<%BANNER%>

Optimization Methods in Intensity Modulated Radiation Therapy Treatment Planning

Permanent Link: http://ufdc.ufl.edu/UFE0021522/00001

Material Information

Title: Optimization Methods in Intensity Modulated Radiation Therapy Treatment Planning
Physical Description: 1 online resource (148 p.)
Language: english
Creator: Aleman, Dionne M
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: beams, fluences, imrt, optimization
Industrial and Systems Engineering -- Dissertations, Academic -- UF
Genre: Industrial and Systems Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The design of a treatment plan for intensity modulated radiation therapy is a mathematical programming problem which is not yet satisfactorily solved. Current techniques include dividing the problem into several subproblems, which are then solved sequentially. My research addresses several of these subproblems, particularly, beam orientation optimization (BOO), fluence map optimization (FMO) and fractionation. The integration of the BOO and FMO subproblems is considered, as well as improved techniques to model the dose deposition of a beamlet.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Dionne M Aleman.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Romeijn, Hilbrand E.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021522:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021522/00001

Material Information

Title: Optimization Methods in Intensity Modulated Radiation Therapy Treatment Planning
Physical Description: 1 online resource (148 p.)
Language: english
Creator: Aleman, Dionne M
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: beams, fluences, imrt, optimization
Industrial and Systems Engineering -- Dissertations, Academic -- UF
Genre: Industrial and Systems Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The design of a treatment plan for intensity modulated radiation therapy is a mathematical programming problem which is not yet satisfactorily solved. Current techniques include dividing the problem into several subproblems, which are then solved sequentially. My research addresses several of these subproblems, particularly, beam orientation optimization (BOO), fluence map optimization (FMO) and fractionation. The integration of the BOO and FMO subproblems is considered, as well as improved techniques to model the dose deposition of a beamlet.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Dionne M Aleman.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Romeijn, Hilbrand E.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021522:00001


This item has the following downloads:


Full Text





OPTIMIZATION METHODS IN INTENSITY MODULATED RADIATION THERAPY
TREATMENT PLANNING,



















By
DIONNE M. ALEMAN


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2007




































S2007 Dionne M. Aleman










To my ever-patient wife Nancy, and to my father Roberto, who, if not for the

shortconlings of current cancer treatments, might still be with us tod a-









ACKNOWLEDGMENTS

Il Ia!y thanks to Nancy Huang, ClIn-Il..ph.: Fox and Bart Lynch for so helpfully and

happily explaining the physics of medical physics to me on a wide range of topics, even

when those topics are not relevant to my own research.

This work was supported in part by the NSF Alliances for Graduate Education and

the Professoriate, the NSF Graduate Research Fellowship and NSF grant DMI-0457394.








































2 FLITENCE MAP OPTIMIZATION.

2.1 Introduction.
2.2 Literature Review.
2.3 Model Formulation.
2.4 Spatial Considerations
2.5 A Primal-Dual Interior Point Algorithm for FMO
2.5.1 Primal-Dual Interior Point Algorithm.
2.5.2 Hessian Approximations
2.5.2.1 Single Hessian Approximation.
2.5.2.2 BFGS Hessian Update.
2.5.3 Insignificant Beamlets.
2.5.4 Warm Start
2.6 Results.
2.6.1 How Small of a Duality Gap is Necessary?
2.6.2 Computational Results
2.6.3 Clinical Results
2.6.4 Spatial Coefficient Results
2.6.5 Warm Start Results.
2.7 Conclusions


TABLE OF CONTENTS


page


ACKNOWLEDGMENTS

LIST OF TABLES.

LIST OF FIGURES

ABSTRACT

CHAPTER

1 INTRODUCTION

1.1 Intensity Modulated Radiation Therapy Treatment Planning .....


___





1.2 Dissertation Summary


1.2.1 Fluence Map Optimization.
1.2.2 Beam Orientation Optimization
1.2.3 Fr-actionation
1.2.4 Modeling the Dose Deposition of a Beam
1.3 Contribution Summary.
1.3.1 Fluence map optimization
1.3.2 Beam Orientation Optimization
1.:3.3 Fr-actionation
1.3.4 Modeling the Dose Deposition of a Beam











:3 BEAM ORIENTATION OPTIMIZATION ..... .. . 46

:3.1 Introduction .. ... . .. 46
:3.2 Literature Review .. ... ... .. 47
:3.3 Model Formulation ......... .. .. 48
:3.4 Alixed-Integer Model Formulation ...... .. . 50
:3.5 Beam Data Generation ......... .... .. 52
:3.6 A Response Surface Approach to BOO ..... ... .. 54
:3.6.1 Overview of Response Surfaces ..... .. . 55
:3.6.2 Determining the Next Observation .... ... .. 58
:3.6.2.1 Maximizing the expected improvement .. .. .. .. 59
:3.6.2.2 Obtaining an upper bound on the uncertainty .. .. .. 59
:3.6.2.3 Branch-and-Bound ...... .. .. 61
:3.6.3 Method of Obtaining the Next Observation .. .. .. 69
:3.7 Neighborhood Search ......... .. .. 69
:3.7.1 Introduction ......... .. .. .. 69
:3.7.2 Neighborhood Search Approaches .... .. .. .. 70
:3.7.3 A Deterministic Neighborhood Search Method for BOO .. .. .. 70
:3.7.3.1 Neighborhood Definition .... ... .. 71
:3.7.3.2 Neighbor Selection . ..... .. 72
:3.7.3.3 Implementation . ..... .. 72
:3.7.4 Simulated Annealing ......... .. 7:3
:3.7.4.1 Neighborhood Definition .... ... .. 75
:3.7.4.2 Neighbor Selection . ..... .. 75
:3.7.4.3 Implementation . ..... .. 75
:3.7.4.4 Convergence ........ .. .. 76
:3.7.5 A New Neighborhood Structure ...... .. 77
:3.8 Results ............ .. ..... .... 79
:3.8.1 Evaluating Plan Quality . ...... .. 79
:3.8.1.1 Target coverage . ..... .. 79
:3.8.1.2 Critical structure sparing .... .. .. 80
:3.8.2 Response Surface Method Results ... .. .. 81
:3.8.2.1 Proof of concept ..... .... . .. 8:3
:3.8.2.2 Adding a non-coplanar beam to a coplanar solution .. 84
:3.8.2.3 Clinical results . ..... .. .. 85
:3.8.3 Neighborhood Search Method Results ... . .. 88
:3.8.3.1 Add/Drop algorithm results ... . .. 89
:3.8.3.2 Simulated Annealingf results ... . .. 89
:3.8.:3.3 Clinical results . ..... .. .. 91
:3.9 Conclusions and Future Directions ..... .. .. 92
:3.9.1 Response Surface Conclusions .... ... . 92
:3.9.2 Neighborhood Search Conclusions ... ... .. 95












4 FRACTIONATION ......... .. .. 96

4.1 Introduction ......... . .. .. 96
4.2 Model Formulation ......... . .. 97
4.3 Results. ............ .. ......... 100
4.3.1 Computational Results . ...... .. 101
4.3.2 Clinical Results ......... .. .. 102
4.3.3 Spatial Coefficient Results . ..... .. 103
4.4 Conclusions and Future Directions ...... .. .. 111

5 A MONTE CARLO METHOD FOR MODELING DOSE DEPOSITION .. 120

5.1 Introduction ......... . .. .. 120
5.2 Monte Carlo Engine ......... . .. 121
5.3 Dose Distribution of a Beamlet . ...... .. .. 121
5.3.1 Depth-Dose Curve ......... .. .. 122
5.3.2 Lateral Penumbra ......... .. .. 123
5.4 Methodology to Model a Beamlet . ..... .. 124
5.4.1 Modeling the Depth-Dose Curve .... ... .. 125
5.4.2 Modeling the Lateral Penumbra ..... ... .. 128
5.5 Results ............. .. ......... 132
5.6 Conclusions and Future Directions ...... .. .. 138

REFERENCES ......... . .... .. 147

BIOGRAPHICAL SK(ETCH ....._... . .. 148










LIST OF TABLES


Average run times for 5-heant treatment plans ......

FATO value obtained using e = 0.001 .....

Comparison of duality gaps .....

Performance measures of interior point method warm starts ......

Performance measures of projected gradient method warm starts ....

Sparing criteria varies for each critical structure ....

Sizes of test cases .......

Mininiun FMO value obtained and time required to obtain it ......

Target coverage achieved by the treatment plans .......

Percentage of plans in which an organ is spared ......

Definitions of intplenientations .......

Case sizes and run times using identical algorithm and weighting paramet

Sparing criteria varies for each critical structure ....

Computation times in minutes of Monte Carlo simulations ..... .

Computation times for dose distribution fits ......

Variation of fits .......


Table

2-1

2-2

2-3

2-4

2-5

:3-1

:3-2

:3-3

:3-4

:3-5

:3-6

4-1

4-2

5-1

5-2

5-3


page

:36i

:36i

:37

4:3

44

80

82

86i

86;

87

91

102

10:3

1:32

1:34

1:38


;ers ..












LIST OF FIGURES


Figure

2-1 Progression of duality gap .....

2-2 Dose received by targets as a function of the duality gap .....

2-3 Dose received by saliva glands as a function of the duality gap ......

2-4 Quality of DVHs for various duality gaps ......

2-5 The spatial coefficients used for two cases .....

2-6 Comparison of spatial and non-spatial treatment plans .....

2-7 Comparison of spatial and non-spatial treatment plans ......

:3-1 A linear accelerator and the available movements ......

:3-2 FATO value as a function of two angles .....

:3-3 Initial regions .......

:3-4 Partitioning a region into subregfions ......

:3-5 Accounting for syninetry ......

:3-6 The flip neighborhood .....

:3-7 Selection probabilities in Nix (0) and NihF (0). .

:3-8 Proof of concept results ......

:3-9 Comparison of response surface, Add/Drop and equi-spaced targets ...

:3-10 Comparison of response surface, Add/Drop and equi-spaced targets ...

:3-11 Add/Drop and simulated annealing comparison of FMO convergence ..

:3-12 Comparison of Add/Drop and 7-heant equi-spaced plans ..... .

:3-13 Comparison of simulated annealing and 7-heant equi-spaced plans ....

4-1 Target DVHs, saliva DVHs and axial slices in Fractions 1 and 2 .....

4-2 Target DVHs, saliva DVHs and axial slices in Fractions 1 and 2 .....

4-:3 Target DVHs, saliva DVHs and axial slices in Fractions 1 and 2 .....

4-4 Target DVHs, saliva DVHs and axial slices in Fractions 1 and 2 .....

4-5 Target DVHs, saliva DVHs and axial slices in Fractions 1 and 2 .....


page

34

35

35

37

38

39

. 40

. 46

. 51

. 64

.. 67

. 68

. 78

. 78

. 84

.. 87

. 88

. 90

. 9:3

. 9:3

.... 104

. .. 105

. .. 106

.... 107

.... 108










4-6

4-7

4-8

4-9

4-10

4-11

4-12

4-13

4-14

5-1

5-2

5-3

5-4

5-5

5-6

5-7

5-8

5-9

5-10

5-11

5-12

5-13

5-14

5-15

5-16


Target DVHs, saliva DVHs and axial slices in Fractions 1 and 2 .. .. .. .. 109

Target DVHs, saliva DVHs and axial slices in Fractions 1 and 2 .. .. .. .. 110

DVHs and axial slices in Fr-actions 1 and 2 using spatial coefficients .. .. .. 112

DVHs and axial slices in Fr-actions 1 and 2 using spatial coefficients .. .. .. 113

DVHs and axial slices in Fr-actions 1 and 2 using spatial coefficients .. .. .. 114

DVHs and axial slices in Fr-actions 1 and 2 using spatial coefficients .. .. .. 115

DVHs and axial slices in Fr-actions 1 and 2 using spatial coefficients .. .. .. 116

DVHs and axial slices in Fr-actions 1 and 2 using spatial coefficients .. .. .. 117

DVHs and axial slices in Fr-actions 1 and 2 using spatial coefficients .. .. .. 118

Dose distribution of a single beamlet in various tissues ... .. . .. 122

Colorwash of the lateral penumbra of a finite sized pencil beam .. .. .. .. 124

Plot of the lateral penumbra of a finite sized pencil beam .. .. .. .. 125

Observed depth-dose curve in water for several histories .. .. .. 126

Polynomial fits of several histories ....... ... .. 127

Variation of polynomial fit as function of degree ... .. .. .. 128

An error function and an error function pair .... .. . 129

Lateral penumbra for several numbers of Monte Carlo histories .. .. .. .. 130

Error function fits of several histories . .... .. 131

Error function pairs summed to approximate a beamlet in water .. .. .. .. 135

Depth-dose curves in muscle tissue. . ...... .. 135

Lateral penumbra curves in muscle tissue. ..... .. 136

Depth-dose curves in lung tissue. ........ ... .. 136

Lateral penumbra curves in lung tissue. ...... .... . 137

Depth-dose curves in heterogeneous muscle and lung tissue. .. .. .. .. 138

Variation of fits as a function of number of histories .. .. .. .. 139









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

OPTIMIZATION METHODS IN INTENSITY MODULATED RADIATION THERAPY
TREATMENT PLANNING,

By

Dionne M. Aleman

December 2007

Cl.! ny~: H. Edwin Romeijn
Major: Industrial and Systems Engineering

The design of a treatment plan for intensity modulated radiation therapy a

mathematical programming problem which is not yet satisfactorily solved. Current

techniques include dividing the problem into several subproblems, which are then solved

sequentially. My research addresses several of these subproblems, particularly, beam

orientation optimization (BOO), fluence map optimization (FMO) and fractionation.

The integration of the BOO and FMO subproblems is considered, as well as improved

techniques to model the dose deposition of a beamlet.









CHAPTER 1
INTRODUCTION

1.1 Intensity Modulated Radiation Therapy Treatment Planning

Every year, approximately 1.4 million people in the United States alone are newly

diagnosed with cancer (American Cancer Society, [1]). More than half of these patients

will receive some form of radiation therapy (jl!urphy et al. [2], Perez and Brady [3]), and

approximately half of these patients may significantly benefit from conformal radiation

therapy (Steel [4]). During this therapy, beams of radiation pass through a patient,

thereby killing both cancerous and normal cells. Although some patients die of their

disease despite sophisticated treatment methods, many patients may suffer unpleasant side

effects as a result of the radiation therapy which may severely detract from the patient's

quality of life.

Thus, the radiation treatment must be carefully planned so that a clinically

prescribed dose is delivered to targets containing cancerous cells so that the cancer

will be eradicated. Simultaneously, a small enough dose must be delivered to the nearby

organs and tissues (called critical structures) so that they may survive the treatment. This

is achieved by irradiating the patient using several beams sent at different orientations

spaced around the patient so that the intersection of these beams includes the targets,

which thus receive the highest radiation dose, whereas the critical structures receive

radiation from some, but not all, beams and may thus be spared. Currently, a technique

called intensity modulated radiation therapy (IMRT) is considered to be the most effective

radiation therapy for many forms of cancer.

The problem of designing an IMRT treatment plan for an individual patient is a

large-scale mathematical programming problem that is not yet solved satisfactorily.

Current treatment planning systems decompose the planning problem into several stages,

and the corresponding subproblems are solved sequentially. These subproblems include

determining the number and orientation of the beams of radiation, the radiation dose










distribution of each heant and the decomposition of a single treatment plan into several

smaller fractions.

This work addresses the integration of the beam orientation optimization (BOO) and

fluence nmap optimization (FMO) subproblents based on a convex formulation of the latter

and associated efficient algorithms for solving it, an approach which has not received much

attention in previous studies. The fractionation problem, the problem of dividing a single

treatment plan into the 35 treatments (fractions) the patient will actually receive, is also

addressed. Also, the problem of modeling the dose deposition of a beam is also considered.

1.2 Dissertation Summary

In IMRT, each heant is modeled as a collection of hundreds of small beantlets, the

fluences of which can he controlled individually. These fluence values are known as a

fluence nmap, and optimization of these fluences given a fixed set of beams is known as

fluence nmap optimization. The optimal solution value of the FATO problem quantifies the

quality of the treatment plan, where quality means the ability of the plan to deliver the

prescribed radiation dose to the specified target structures while sparing critical structures

by ensuring that they receive an acceptably low amount of radiation. Thus, the quality of

a set of beams can he measured by the optimal solution of the FATO problem performed

with those beams. Thus, the problem of selecting the best directions from which to

deliver radiation to the patient (the BOO problem) is based on the treatment plan quality

indicated by the optimal solution value to the corresponding FMO problem.

1.2.1 Fluence Map Optimization

One of the most popular subproblents of the intensity modulated radiation therapy

(IMRT) treatment planning problem is the fluence nmap optimization (FMO) problem.

In IMRT, each heant of radiation can he discretized in hundreds of smaller heantlets, the

radiation intensities (fluences) of which can he modulated independently of the other

heantlets. For a given set of beams, the beamlet fluences can greatly influence the quality

of the treatment plan, that is, the ability of the treatment to deposit the prescibed amount










of dose to cancerous target structures while simultaneously delivering a small enough dose

to critical structures so that they may continue to function after the treatment. These

fluence values are known as a fluence nmap, and optimization of these fluences given a fixed

set of beams is known as fluence nmap optimization.

Because the fluences of the beamlets can drastically affect the quality of the

treatment plan it is critical to obtain good fluence maps for radiation delivery. As the

FMO problem is one of the most popular subproblents in IMRT optimization, it has been

extensively studied in the literature. Several problem structures and algorithms to solve

various models are presented in many studies.

1.2.2 Beam Orientation Optimization

In a typical head-and-neck treatment plan, radiation beams are delivered front 5-9

nonlinally-spaced coplanar orientations around the patient. These coplanar orientations

are obtained front rotating the gantry only. Several components of a linear accelerator can

rotate and translate to achieve more orientations than those obtained front rotating the

gantry. The available orientations consist of the orientations obtained front rotation of the

gantry, colliniator and couch, as well as the three translation directions of the couch.

Beam orientation optimization (BOO) is the problem of selecting front the available

beam orientations the best set to use in delivering a treatment plan. Given a fixed set

of beams, different fluence maps (radiation intensities of heantlets) yield treatment plans

with different qualities. Therefore, the quality of an optimized fluence nmap should be

considered when selecting a set of beam orientations to use in a treatment plan. Optimal

fluence maps may be difficult to obtain depending on the FMO model. Thus, it is coninon

in the literature for scoring approximations and other heuristics to be used to estimate the

quality of a beam solution.

Regardless of the objective function used in the BOO problem, the problem is

fundamentally nonlinear as the physics of dose deposition change with direction. Because

nonlinear progranining problems are difficult to solve, most approaches to the BOO










problem rely on global search algorithms to obtain a solution, which may or may not he

optimal.

1.2.3 Fractionation

An important subproblem related to the FATO problem which has not yet received

much attention is the fractionation problem. Rather than deliver an entire treatment plan

in one session, a treatment plan is divided into several sessions, called fractions. This

is done to take advantage of the fact that normal, healthy cells recover faster from the

radiation than cancerous cells. To obtain the treatment plans for the fractions, in practice,

a single FMO treatment plan is developed and then divided into the desired number of

fractions, usually around :35. This division of a treatment plan is a non-trivial task, as

the target voxels, geometric cubes of tissue, must receive 1.8-2.0 Gy of radiation in each

fraction.

With a single IMRT treatment plan, it is practically impossible to devise a constant

dose-per-fraction delivery technique because only a single FMO problem is solved to

obtain the treatment plan, which is then simply divided into a number of daily fractions.

If a single plan is optimized to deliver doses to multiple target-dose levels, then the dose

per fraction delivered to each target must change in the ratio of a given dose level to the

maximum dose level. For example, my- PTV1 has a prescription dose of 70 Gy, PTV2 has

a prescription dose of 50 Gy, and the number of fractions is :35. If a single treatment plan

is divided among the :35 fractions, then PTV1 will receive 70/:35 = 2.0 Gy in each fraction,

but PTV2 will only receive 50/:35 = 1.4 Gy, and thus any cancerous cells in PTV2 may

not he eradicated by the treatment. Similarly, if only 25 fractions are used in order to

ensure that PTV2 receives 2.0 Gy per fraction, then PTV1 receives 70/25 = 2.8 Gy per

fraction, well above the desired dose.

1.2.4 Modeling the Dose Deposition of a Beam

The FATO problem is arguably the most significant in determining the quality

of the treatment plan. The FATO problem depends heavily on the calculation of dose










received in each voxel of a patient. This dose is typically approximated by assuming a

linear relationship with the radiation intensities of the beamlets delivering the radiation.

Although this approximation is accepted as satisfactory, it is not truly accurate.

The dose in a voxel is determined by the paths the photons in the radiation beams

follow through the patient. Some photons may collide with particles inside the patient

and scatter in any direction, while others may collide with particles and be absorbed.

Still other photons may pass entirely through the patient with no collisions. Due to the

unpredictable nature of the radiation beam inside the patient, the dose received in a

voxel can only be accurately obtained through Monte Carlo simulations. A simple linear

relationship is assumed between total dose and heamlet fluences and is commonly accepted

as a satisfactory dose approximation in IMRT optimization. Errors of as much as t:I' .

have been reported for photon beams near tissue inhomogeneities (11 .. et al. [5]).

For IMRT optimization, particularly with advent of image-guided IMRT (IGIMRT),

or 4D IMRT, the FATO problem must he solved extremely quickly to create real-time

treatment plans. Thus, the speed of the FATO problem is paramount. Lengthy Monte

Carlo simulation can provide an accurate measure of the dose deposited in a voxel,

but this technique is time intensive and impractical for clinical use and particularly for

treatment planning optimization.

1.3 Contribution Summary

1.3.1 Fluence map optimization

Nonlinear functions to approximate biological behavior and desired dose distributions

are common in the previously proposed FMO models in the literature, as are mixed-integer

programming models. These models can he difficult and computationally expensive to

solve. To make the FATO problem more tractable, we employ a model with a convex

objective function and linear constraints. This desirable structure allows our model to be

solved quickly and to optimality with the primal-dual interior point algorithm we have

developed specifically for this problem.










One of the greatest benefits of an interior point algorithm is that a globally

optimal solution can be found for many problem structures, and in particular, convex

problem structures. As our FMO model is convex, the interior point algorithm can

locate the globally optimal solution to within a specified duality gap. While there are

other algorithms that can theoretically return a globally optimal solution to a convex

problem (and many algorithms that cannot), interior point methods have the advantage of

providing a known duality gap and generally fast computation times. Because the duality

gap is known in each iteration, the user can make knowledgeable trade-offs between

computation time and solution optimality without having to guess how far from the

optimum the final solution may be. This allows for a scientific comparison of different

IMRT delivery techniques as we can solve the different problems to a specific duality gap.

Several alterations to the standard primal-dual interior point method were made

to improve its performance. Beamlets that are likely to have little or no contribution to

the treatment plan are removed a priori and different approximations to the objective

function Hessian are tested to save time in calculating the true Hessian in each iteration.

The use of warm starts to initialize the interior point method is also examined. The

solutions obtained provide quality treatment plans in a clinically feasible amount of time.

The incorporation of spatial information into the FMO model is also considered.

The probability of tumor metastasis increases with proximity to gross tumor mass. By

using the distances of voxels from target structures, the voxels can be weighted according

to their importance in the treatment plan. For example, it should be less important to

spare saliva gland voxels near a target structure than it should be to spare saliva gland

voxels far from a target. The use of spatial coefficients will help the model identify quality

treatment plans that will prevent future metastasis.

1.3.2 Beam Orientation Optimization

For head-and-neck cancers, typical IMRT treatment plans use 5-9 equi-spaced

coplanar beams. Coplanar beams are those beams obtained from the rotation of only










the gantry of the linear accelerator, the machine which delivers radiation beams to the

patient. If all other components of the linear accelerator are fixed, the rotation of the

gantry sweeps out a set of coplanar beams. The couch can rotate and translate in three

dimensions, and the head of the gantry can rotate independently, creating an even larger

set of beams. Beams obtained front the movement of more than one component front the

linear accelerator are known as non-coplanar beams.

Intuitively, one may expect that the number of beams required for a high-quality

treatment plan can he reduced, or the quality of the treatment plan for a given number

of beams can he improved, if the beam orientations are chosen optimally and/or front

a larger set. In particular, we investigate the effect of considering more coplanar or

non-coplanar beams. A treatment plan consisting of fewer beams is preferable because

the number of beams used in a plan directly affects the length of the actual treatment.

If fewer beams are used to treat a patient, then each treatment takes less time and more

patients can he treated in a d w, which is beneficial front both a clinical and economic

perspective. Longer treatment times also allow for more errors due to possible patient

motion.

We view the BOO problem in IMRT treatment planning as a global optimization

problem with expensive objective function evaluations, each of which involves solving

a FMO problem. We propose a response surface method that, unlike other approaches,

allows for the generation of problem data only for promising beam orientations on-the-fly

as the algorithm progresses, enabling the consideration of far more candidate orientations

than is currently feasible. Our response surface approach to BOO allows us to develop

high quality plans using just four beams for head-and-neck cases, in contrast to the

current practice of using 5-9 heants. The response surface method also provides for

convergence to the globally optimal solution.

We have developed neighborhood search methods to solve our BOO model. One

method is simulated annealingf, a proper global optimization algorithm, and the other










is a local search heuristic designed specifically for the BOO problem. The local search

heuristic, which we call the Add/Drop method, returns a locally optimal solution in a

small amount of time. The simulated annealing algorithm has the ability to escape local

nmininia, and is theoretically able to return a globally optimal solution given enough time.

For each of these algorithms, we have devised a new neighborhood structure based on

observations of known optimal BOO solutions compared to the simulated annealingf and

Add/Drop BOO solutions. This new neighborhood structure provides faster objective

function value convergence in both algorithms.

1.3.3 Fractionation

In practice, a single FMO treatment plan is developed and then divided into the

number of desired fractions. Dividing a single FMO into multiple treatments is a

non-trivial task, owing to the need of maintaining a constant dose-per-fraction to each

the target structures, which may have different prescription doses. Therefore, any division

of a single FMO plan into multiple fractions can lead to suboptinmal treatments. We

propose a new method of formulating the fractionation problem which yields optimal

fluence maps for each cancerous target structure. These fluence maps can then he easily

divided into optimal fractions.

The proposed fractionation model is solved using the same prinial-dual interior point

method presented for the FATO problem. The solutions provide high quality fluence maps

for each target, and in a clinically acceptable amount of time.

1.3.4 Modeling the Dose Deposition of a Beam

We propose obtaining a limited number of Monte Carlo histories to obtain a noisy

dose distribution which can then he transformed into a very accurate, smooth dose

distribution suitable for optimization techniques in a reasonable amount of time.

Because the particles in a heantlet scatter in three dimensional space, multiple

dose distributions must he considered to satisfactorily model the beamlet's affect on

the patient's tissue. These distributions arise front the amount of radiation the beamlet










deposits as a function of depth (the depth-dose curve), and from the amount of radiation

radiating outward from the center of the beamlet (the lateral penumbra). The depth-dose

curve is modeled using a high-degree polynomial and the lateral penumbra is modeled as

the sum of error functions. The parameters of the error functions are determined using a

Levenbergf-Marquardt quI I---5 li.-on minimization method.

Using these techniques, dose distributions with satisfactory accuracy can be obtained

using at least a factor of 10 fewer Monte Carlo histories than would otherwise be required.

This can greatly decrease the amount of time required to obtain dose data for beamlets in

the FMO problem of IMRT treatment planning without sacrificing accuracy.









CHAPTER 2
FLUENCE MAP OPTIMIZATION

2.1 Introduction

IMRT is differentiated from conformal radiation therapy by the dose distributions

that can he delivered hv each heam. Rather that just delivering a uniform radiation field

of radiation, the dose distribution of a beam can he any desired distribution. This ability

allows for greater flexibility and accuracy in targeting the target structures while avoiding

the critical structures.

The dose distribution of a beam is achieved as follows. In IMRT, each heam can

he thought of as consisting of several hundred smaller beamlets, each of which can have

its own radiation intensity (fluence) independent of its neighbors. By modulating the

intensities of these beamlets, any dose distribution can he achieved. Given a fixed set of

beams, the optimization of these intensities is called fluence map optimization.

2.2 Literature Review

Because the FATO problem is one of the most studied problems of IMRT, many

different approaches have been taken to formulate the FATO problem, based on both

"physical" (Bortfeld [6]) and biologicall" (Alber and Nusslin [7], Jones and Hoban [8],

K~allman et
14]) objective functions and constraints. Linear programming (LP)-hased multi-criteria

optimization (Hamacher and Kiifer [15]) and mixed-integer linear programming (!l~l.P')

(Bednarz et

Constraints to enforce various measures of treatment quality are also taken into

account in different FMO models. Hamacher and K~iifer [15] include the homogeneity

of the dose received by the targets as a constraint in their FMO model. Full-volume

constraints, which require that the dose in every voxel of a structure he within pre-determined

upper and lower bounds, are common for controlling the dose in each structure. Models










employing full-volume constraints are found in Bednarz et
[15], Lee et
volume constraints, constraints requiring that dose in only a subset of voxels he within

pre-determined upper and/or lower bounds, are also common. Formulations with partial

volume constraints are found in Lee et

In addition to varying constraints, there are many competing methods of formulating

the FMO objective function to reflect the quality of the treatment plan. Shepard et
describe several different objective formulations. These formulations include minimizing

the sum of doses received at all voxels; minimizing a weighted combination of doses

received at each voxel, where the weights depend on the structure in which the voxel

resides; and minimizing the deviation of the dose in each voxel from the recommended

prescription.

Romeijn et
proposed in the medical physics literature are equivalent to convex penalty function

criteria when viewed as a multicriteria optimization problem. For each set of treatment

plan evaluation criteria from a very large class, there exists a class of convex penalty

functions that produces an identical Pareto efficient frontier. Therefore, a convex penalty

function-based approach to evaluating treatment plans is used to investigate the BOO

problem. Although this approach could be used in a multicriteria setting, Romeijn

et
evaluation criteria that produces high-quality treatment plans for a population of patients,

eliminating the need to solve the FATO problem as a multicriteria optimization problem for

each individual patient.

2.3 Model Formulation

A convex penalty function-based approach to the FMO model as described in

Romeijn et









making the trade-off between delivering the prescribed radiation dose to the target

structures while sparing the critical structures. Using this approach, the FMO problem

can formulated as a quadratic programming problem with linear constraints as follows.

Denote the set of all potential beam orientations as B. The structures (both targets

and critical structures) are irradiated using a predetermined set of beam angles, denoted

8, where each beam On E a, & = 1,..., k and k is the number of beams in 0. Each beam

is decomposed into a rectangular grid of beamlets with m rows and a columns, yielding

typically 100-400 beamlets per beam. The position and intensity of all beamlets in a beam

can be represented by a vector of values representing the beamlet intensities, called bixels.

The set of all bixels in beam On is denoted by Beh. The core task in IMRT treatment

planning is findings radiation intensities for all beamlets.

Denote the total number of structures by S and the number of targets by T. Each

structure s is discretized into a finite number v, of volume cubes, known as voxels.

Typically, around 350,000 voxels are required to accurately represent the targets and

surrounding structures of a head-and-neck cancer site.

Because a beamlet must pass through a certain amount of tissue to reach a voxel, the

dose received in a voxel from a beamlet may not be the full delivered intensity. Denote

Dij, as the dose received by voxel j in structure a from beamlet i at unit intensity. The

Dij, values are known as dose deposition coefficients. Let xi denote the intensity of bixel i.

This brings us to the following expression for the dose zys received by voxel j in structure




h=1 i6EBBh
Although the goal of IMRT treatment planning is to control the dose received by

each structure, if hard constraints are imposed on the amount of dose received by each

structure because such a solution may not exist. In some cases, it may be necessary to

sacrifice organs in order to treat targets, and if that possibility is not allowed in the model,

then a feasible or a satisfactory solution may not exist. Thus, in our model, a penalty is










assigned to each voxel based on the dose it receives for a given set of beamlet intensities.

Let Fj, denote a convex penalty function for voxel j in structure s of the following form:

1j(s -


where T, is the dose threshold value for structure s, w, and ps are weighting factors for

underdosing, and w, and p, are weighting factors for overdosing. The expression (-)*

denotes max{0, -}. The function is normalized over the number of voxels in the structure

using the coefficient 1/v,. By setting,, ws, > 0 and p,, p~ > 1, convexity is ensured.

A basic formulation of the FMO problem is then:



s=1 j=1

subjct t zy= Dasxyj= 1,..,v, a= 1,...,S
h=1 i6EBBh
xi > 0 i E Be,, h= 1,..., k


The FMO problem is the black-box function F(0) in the BOO model to quantify the

quality of beam vector 0. In contrast with the methods presented by all of the previously

cited FMO studies except for Das and Marks [27], Haas et al. [28] and Schreibmann [29],

this measure of beam vector quality is an exact measure of the FMO problem, rather than

using heuristic methods or scoring approaches which cannot accurately optimize the beam

orientations.

2.4 Spatial Considerations

With IMRT optimization, it is possible to generate treatment plans with similar FMO

objective function values but very different levels of clinical treatment quality. C'!s I, et

al. 2003 [30] illustrate this possibility with two treatment plans that have nearly identical

target coverage when plotted on a dose-volume histogram, but while one plan delivers

an acceptable homogeneous dose, the other plan results in significant underdosing of the

target structure.










C'!s Iu et al. 2003 [30] show that the probability of microscopic tumor extension

decreases linearly with distance from the gross tumor volume, implying that cold spots

located near the gross tumor volume are far more likely to allow for tumor metastasis after

treatment. Likewise, cold spots located far from the gross tumor volume are unlikely to

result in tumor metastasis.

To reduce the likelihood of obtaining an unsatisfactory plan with a good dose-volume

histogframs, spatial coefficients are introduced into the FMO model. For each voxel, we

consider its position relative to the primary target as a measure of how acceptable/unacceptable

overdosing or underdosing may be. Voxels further from the gross tumor volume are

penalized more heavily than voxels closer to the gross tumor because it is less acceptable

for a voxel far away from the actual tumor to receive an overdose, as the cancerous cells

are unlikely to spread very far from the tumor location (C'!s I.. et al. [30]). This additional

penalization is called the spatial coefficient, and is denoted cj, for voxel j in structure s.

For voxels inside the target structures, the probability of cancer spread is 1, as cancer

already exists in those voxels. Let S' denote the set of gross tumor structures. Let dej, be

the minimum distance from voxel j in structure a to structure e. The spatial coefficient cj,

for voxel j in structure s is


1 j v,, a ( S'
m~j r i nl ( 1 m a x 0 .00 1 e p ( A d ) e g + A 1 v s


where Ay, pr and Pe are weighting coefficients. The objective function for the FMO

problem becomes


s=1 j=1

2.5 A Primal-Dual Interior Point Algorithm for FMO

To solve the FMO and fractionated FMO models, a primal-dual interior point method

is emploi-. I For a convex problem such as the FMO model presented in the preceding

section, this method yields an optimal solution in short amount of time.










The primal-dual interior point algorithm moves through the interior of the solution

space along a central path (a path through the interior of the solution space) toward the

optimal solution. The central path is defined by perturbing the KKET conditions described

below. These conditions ensure primal feasibility, dual feasibility and complementary

slackness. If these conditions are satisfied for a convex programming problem with linearly

independent constraints, they yield the optimal solution. Thus, we only need to solve this

system to obtain an optimal solution to our FMO model (which has a convex objective

function and linear, linearly independent constraints). The KKET system can be difficult to

solve, so the conditions are perturbed in order to obtain a solution.

The general idea of the primal-dual interior point algorithm is to start from an initial

feasible solution, use the perturbed KKET conditions to obtain a step direction close to the

central path, and then move the current solution some step length along that direction.

The amount of pertubation in the KKET conditions is gradually decreased so that in each

step, the solution becomes closer to the optimum. The interior point method allows for

the duality gap, the gap between the objective functions of the primal and dual problems,

to be calculated, thus providing a measure of how close the current solution is to the

optimum. For a problem with continuous variables, when the objective functions of the

primal and dual problems are equal (duality gap of zero), the solution is optimal.

A mathematical description of the primal-dual interior point method can be found

in Nocedal and Wright [31]. Further explanation is provided only as needed to define

variables in the algorithm. In the FMO problem, G(x) = -Ix, so the KKET conditions for

the FMO formulation are


-sS1 DayCou F ~eN D~r i si = 0 ie N (2-1)

sixi = 0 if NV. (2-2)

si > 0 ieN1 (2-3

xi > 0 i NV, (2-4









where the Equation (2-4) ensures that the solution is feasible, as the only constraints

in the FMO problem are nonnegativity. The complimentary slackness constraint (2-2)

forces the solution to the above conditions to be on the boundary of the solution space.

Since a point in the interior of the solution space is desired, the complimentary slackness

constraint must be relaxed.

The complimentary slackness constraint (2-2) is relaxed by changing each sixi = 0 to

sixi = p-, where p > 0. This, along with requiring that x > 0 and s > 0 for feasibility,

ensures that a solution to the perturbed KKET conditions is an interior point.

Let a be the size of decision variable vector x. A solution is "close encalll to the

central path if the duality measure p in iteration k is


1k k T (2-5)


and ||XkSk k~l k ~, Where Xk is a matrix with xi values as diagonals and zeros

elsewhere, and Sk is a matrix with si values as diagonals and zeros elsewhere.

As the algorithm progresses, p is reduced to zero until the solution is sufficiently close

to optimality. To reduce p, in each iteration we set p = pe, where a e [0, 1] is called the

centering parameter. If the duality gap is very large, a can be reduced so that p is reduced

faster.

In each iteration, the current solution (x, s) is moved in a direction (ax, as) for some

step length a~ is given by

Xk~+1 X~ aXk

Sk~+1 S~ aSk

Let Xk = diag(xk"), Sk = diag(sk"), H(Xk) = 2 (Xk). The directions axk" and ask"

can be determined by solving the following equations:


[(Xk -1 Sk H(Xk) AXk -rDF (k -1 rs (2-6)

ask (Xk -1 rus +SkXk) (2-7)










In order to solve this system, we must obtain axk" from Equation (2-6) by taking the

inverse of [(Xk -1Sk + H]. Because computing the inverse of such a large dense matrix is

very time consuming, a Cle.!.~ -l:y factorization to solve this system quickly.

The primal-dual interior point method requires a feasible (x, s) solution in each step.

Thus, a maximum step length camax must be imposed on each step direction to ensure that

x > 0 and s > 0:


amax, = mim i mm -xi/ax4 mm -s/Ag


Because the inverse of each xi is required to determine the step directions, it is

undesirable to have any xi = 0, which would result from using step length camax. Instead,

only a percentage rl < 1 of amax, is used:


a = mn{1, amax}(2-8)


The benefit of this primal-dual method is that in each step, we can calculate the

objective of the dual problem (simply s~x), thus providing a bound on how far the current

solution is from optimality.

2.5.1 Primal-Dual Interior Point Algorithm

The primal-dual interior point algorithm is as follows:


Initialization

1. Select initial values for e, a and rl (we use e = 5, a = 0.01, and rl = 0.95).
2. Set xo 0.05 (very close to 0) and calculate V4(xo) and H(xo) = 2 (XO)
3. Set so = p~(Xo -]
4. Set I-o = ( ~V (xo))i/100.
5. Set k = 0.

Algorithm

1. If the duality gap is very large ((Xk"+1 TS"+1 > 107e), set a = 0.01a.

2. Set pk = k*~









3. Solve for the step direction (Axk, ask") aS described in Equations (2-6) and
(2-7). Note that this involves calculating the Hessian H(Xk)

4. Solve for the step length a~ as described in Equation (2-8).

5. Set xk+1 = Xk + ca~x and sk"+1 = Sk cGas.

6. If the duality gap (Xk+1) T'k+1 < 6, Stop. Otherwise, set pk+1 = (Xk+1) T'k+1l
and k <- k + 1 and repeat.


2.5.2 Hessian Approximations

The most time-consuming step in the primal-dual interior point algorithm is

calculating the Hessian of the objective function in each iteration. For clarity, let C

denote CESes1/vs gg and F)'(x)~ denote F)'(Ez Dljxl). The Hessian of the FMO

problem is then given by

C Fj" (x)D~ ... C Fj" (x)DljDnj
H(x) =

Ll\/UEl F)(xDajr\ .. F'()

Note that only the pairwise Dij products differ in each element of the Hessian. By

precomputing these cross products, only Cses 1/U v, e Fjl'(CEN ljzXI) has to be
recomputed in each iteration. The matrix of the Dij products yields the sparsity (or

density) pattern of the Hessian, which cr li- constant throughout the algorithm. Because

the Hessian is symmetric, the matrix values only need to be computed for half of the

matrix, further improving efficiency.

Despite these observations, computing the Hessian is still so expensive that it renders

the algorithm impractical. Methods of approximating the Hessian are implemented to

speed up the algorithm.

2.5.2.1 Single Hessian Approximation

One way of speeding up the algorithm is to compute the Hessian just once during

initialization to obtain H(xo), and then rather than re-compute the Hessian in each

iteration, use H(xo) as an approximation to H(Xk). We call this the Single Hessian










approximation. Although the convergence of such an approximation has not yet been

mathematically proven, tests run on several head-and-neck cases for 5-heam and 7-heam

plans show that the Single Hessian does in fact converge to the known optimal solution.

2.5.2.2 BFGS Hessian Update

Another Hessian approximation is the Broyden-Fletcher-Goldfarb-Shanno (BFGS)

Hessian update. The approximation to the Hessian in iteration k is Bk, with Bo = H(xo).

The update to the approximated Hessian in each iteration is

qkqk Bkrkp k
Bkc+1 = Bke+.T.~

where


pk = Xk"+1- Xk

qk = o ~Xk+1) V ~Xk


Note that this update ensures that Bk is alr- i-n symmetric and positive definite,

so the Cle.!.~ -l:y factorization can still be applied to obtain the step direction. This

approximation also empirically converges to the known optimal solution for 5- and 7-heam

head-and-neck cases.

2.5.3 Insignificant Beamlets

Insignificant heamlets are those that hear little contribution to the quality of the

FMO plan. Letting d denote the diagonal elements of the initial Hessian H(xo), the set of

insignificant heamlets BI is defined as


BI = i :

These beamlets are removed by removing the ith row and the ith column in H(xo) for

every i E BI, and then updating the number of pixels to the number of remaining pixels.

The insignificant heamlets must he re-inserted into the solution xk in order to calculate










the voxel doses, objective function, gradient and Hessian, but the inversion of the Hessian

is done to the Hessian with the had heamlets removed, providing significant time savings.

2.5.4 Warm Start

For the sake of theoretical accuracy, a truly optimal solution cannot have the had

heamlets described in Section 2.5.3 removed. Without removing the had heamlets a priori,

the interior point method must he run for an impractical amount of time to obtain a

near-optimal solution,; Or, e = 0.001. The interior point method is typically started with

a decision variable vector x equal to almost zero. If the algorithm were to be started at a

point closer to the final solution, denoted x""'"r~, time savings could be gained, allowing all

beamlets to be considered in the interior point algorithm in a reasonable amount of time.

Such an approach is a called a c i.~ I1 -1 .11

One difficulty in using a warm start with the interior point method is that a warm

start solution may have some :rftort" = 0, which is not allowed because the inverse of each

:rs must he taken. To correct this problem, any xrfert" = 0 is simply replaced with some

very small value y. Because these zero-valued variables are less important to the problem

than nonzero variables, y should be less than the minimum nonzero value of xu-ort". Let
1 m ni 1,.,, { rf ar~ : :4 ari > 0}. T h en ] = m in {0 .0 0 1 ] *}.


o :rjarm~ i ( BI

if BI

An additional problem with warm starts in the interior point method is that the KKET

variable vector s is unknown at the warm start point. Depending on the algorithm used

to obtain the warm start, some information about so-ort" and ps,,,,, s and p at the warm

start point, respectively, may not he available. If no information is available about s from

the warm start, then so = 0. If an interior point algorithm is used to obtain the warm

start, then s"'"'" is available. If the warm start did not include the insignificant heamlets,

some corrections must he made to account for the insignificant heamlets which will be










optimized in the final solution. Let so be the initial s used in the interior point method

after the warm start has been obtained. Then,


aswarm i ( BI
Si



where the value chosen for so corresponding to insignificant beamlets arises from the

general initialization s = p~(Xo)-1

2.6 Results

The true Hessian, Single Hessian approximation, and BFGS update implementations

of the primal-dual interior point algorithm are tested on six cases head-and-neck cases to

obtain coplanar, equi-spaced 5-beam plans. The tests are run on a 2.33GHz Intel Core 2

Duo processor with 2GB of RAM. The method is tested for both leaving in and removing

the insignificant beamlets, as well as the proposed alternative to computing the Hessian.

The optimality of the interior point method solutions is verified by comparison to the

known optimal solutions obtained by Java with CPLEX (ILOG).

An acceptable duality gap must be determined in order to implement the interior

point method. While we consider a duality gap of a = 0.001 to be acceptably close to

optimal, it may be unnecessary to achieve such a small duality gap to obtain a quality

solution. A duality gap of 0.001 may be sufficiently small to ensure optimal solutions given

objective function values using certain weighting parameters, depending on the parameters

used in the FMO objective function, the value of the objective function may vary widely.

Because of the potential range of values, a stopping criteria based on a relative duality gap

rather than an absolute duality gap is preferable. Clri the objective function value in an

iteration is f. Define the relative duality gap in an iteration to be e' = e/ f.

An examination of the relative duality gap necessary is presented in Section 2.6.1.

Computational results are presented in Section 2.6.2 and clinical comparisons are provided

in Section 2.6.3.










2.6.1 How Small of a Duality Gap is Necessary?

Because the run time of the algorithm is dependent on the required duality gap, it

is desirable to only require the algorithm to achieve as small a duality gap as necessary

to ensure a clincally good solution. The duality gap decreases quickly in the first few

iterations, and then subsequently decreases by only a small amount per iteration, as

shown in Figure 2-1A. If these iterations with only marginal improvements are found to

be unnecessary in terms of clinical quality, significant time can be saved by stopping the

algorithm once the duality gap is reasonably small, as opposed to waiting until the duality

gap is very small.

To check the importance of the duality gap, the FMO value and dose delivered to the

targets and the saliva glands were plotted against the duality gap in each iteration using

the true Hessian and without removing insignificant beamlets. For a representative case,

the FMO values per duality gap are shown in Figure 2-1B. It is clear that the duality gap

decreases rapidly in the first few iterations, but subsequent iterations yield increasingly

smaller drops in the duality gap.

Similarly, the amount of dose received by the targets and critical structures does not

change significantly toward the end of the algorithm. Figure 2-2 plots the dose received by

the two targets, PTV1 and PTV2, starting from a duality gap of 0.15' The prescription

doses are 70 Gy for PTV1 and 50 Gy for PTV2, common dose values used in the cancer

clinic at Shands Hospital at the University of Florida. Neither the dose received by 95' of

the targets nor the size of the hotspots and coldspots changes significantly in this duality

gap range (Figure 2-2A). The hotspots are measured by the percent of the target receiving

1101' and 1211' of the prescription dose, while the coldspots are measured by the percent

of the target receiving at least 9 :' of the prescription dose (Figure 2-2B).

Figure 2-3 shows for two representative cases the amount of dose received by the

saliva glands starting from a relative duality gap of 0.15' Both cases show that the










Objective function and relative duality gap v. iteration
x14
0.3

,'' 10.25 0.


~3
o' \0.15
O
2L 0.1

1 ~ \ $0.05

0 5 10 15 20 2
iterations

Figure 2-1. The duality gap drops sharply in early iterations, but very slowly thereafter.
The relative duality gap monotonically decreases after several iterations.


change in dose received by the saliva glands as the duality gap decreases is not clinically

relevant.

Fr-om these figures, it appears that a duality gap as large as 0.1 could provide

clinically acceptable plans. Since the algorithm may terminate with a duality gap less than

the one specified as the stopping criteria, a duality gap larger than 0.1 will also be tested

for acceptability.

2.6.2 Computational Results

Table 4-1 shows the average run times for each of the implementations of the

algorithm. Relative duality gaps of 0.15' 0.10I' 0.05' and 0.01 are compared.

The value of 8 used to define the central path is 0.5. As expected, using the Single

Approximation Hessian alternative with the insignificant beamlets removed is the fastest

method, while using the true Hessian is the slowest method, regardless of whether the

insignificant beamlets are removed. Interestingly, for large duality gaps, it is slightly faster

to leave the insignificant beamlets in the model when using the true Hessian. Otherwise, it

is faster to remove the insignificant beamlets.

The final FMO values are di;11l-p I4 for each of the tested methods using a duality

gap of 0.001, which is sufficiently small to ensure optimal solutions given typical objective

function values (Table 2-2). For each case, the final FMO value is nearly identical,















Target hotspots and coldspots
100- -PTV1 at 1 .10
-.-.----------~-~''''- --- PTV1 at 1.20
80
PTV1 at 0.93
--PTV2 at 1.10
60 PTV2 at 1.20
PTV2 at 0.93


Target coverage at 95%


- PTV1
---PTV2


(365

o 6


0.1 0.05
relative duality gap (%)


0.1 0.05
relative duality gap (%)


Figure 2-2. Dose received by targets as a function of the duality gap. A) The amount of
dose received by at least 95' of each target is used to assess proper target
coverage. B) The percent of each target receiving 110'; and 1211' of the
prescription dose indicates hotspots, while C, :' of the prescription dose
indicates coldspots.


Saliva gland dose at 50%


Saliva gland dose at 50%


30


25


O

15


-L. parotid gland
- -R. parotid gland
--- L. SMB gland
- R. SMB gland


-R. parotid gland
- -L. parotid gland
------ R. SMB gland
- L. SMB gland


0 28


10 22
0.1 0.05 0 0.1 0.05 0
relative duality gap (%) relative duality gap (%)
Figure 2-3. The amount of dose received by at least 501' of each saliva gland remains
relatively constant even for large duality gaps. Two representative cases are
shown.









Table 2-1. Average run times for 5-beam treatment plans.
Remove insigf. Avemage run time (s)
Hessian type beamlets? e = 0.001 e' = 0.15 e' = 0.1 e' = 0.05 e' = 0.01
True no 113.8 55.48 55.48 58.58 71.75
True yes 105.6 55.25 56.29 59.09 70.56
BFGS no 43.9 13.59 14.17 14.66 16.67
BFGS yes 40.9 13.19 13.66 14.30 15.88
Single Approx. no 18.1 8.83 8.98 9.29 10.13
Single Approx. yes 16.8 8.6;9 8.84 9.14 9.90

Table 2-2. FMO value from using e = 0.001.

Remove insig.
Hessian type beamlets? Case 1 Case 2 Case 3 Case 4 Case 5 Case 6
True Hessian no 2546.22 2200.70 2289.95 2566.38 5024.97 2585.40
True Hessian yes 2546.22 2200.70 2289.95 2566.38 5024.97 2585.40
BFGS update no 2546.23 2200.70 2289.95 2566.39 5024.97 2585.40
BFGS update yes 2546.24 2200.70 2289.95 2566.39 5024.97 2585.40
Single Approx. no 2546.38 2201.11 2290.40 2566.56 5025.06 2585.82
Single Approx. yes 2546.38 2201.15 2290.44 2566.62 5025.14 2585.82


indicating that the Hessian alternatives and the removal of the insignificant beamlets still

provide for convergence to the optimal solution.

The percentage increases in the FMO values using an absolute duality gap of 0.001

and relative duality gaps of 0.15' 0 .101' 0.05' and 0.01 are shown in Table 2-3.

2.6.3 Clinical Results

For each of the duality gaps tested, the DVHs of the solutions obtained using the

Single Approximation Hessian with the insignificant beamlets removed are compared.

Since the each of the interior point implementations obtains nearly identical solutions, it

does not matter which implementation is used to produce the DVHs.

As previously stated, the prescription doses used are 70 Gy for PTV1 and 50 Gy for

PTV2, marked by a vertical line in Figure 2-4A. As saliva glands are the most difficult

organs to spare in head-and-neck cases, the only critical structures shown are the saliva

glands (Figure 2-4B). All other glands are spared in every implementation. The sparing

criteria used for saliva glands is that no more than 501' percent of the saliva gland can










Table 2-3. Percent increase in objective function value from various relative duality gaps
as opposed to an absolute duality gap of e = 0.001.
Remove insigf. Avy. increase in obj. fn. (.~.)
Hessian type beamlets? e' = 0.15 e' = 0.1 e' = 0.05 e' = 0.01
True no 0.58 0.58 0.27 0.05
True yes 0.58 0.48 0.25 0.06
BFGS no 0.99 0.54 0.30 0.05
BFGS yes 0.94 0.57 0.26; 0.07
Single Approx. no 1.26 0.89 0.6;0 0.19
Single Approx. yes 1.21 0.87 0.57 0.16

Interior point method: Target DVHs Interior point method: Saliva DVHs

100 E'=0.15% 100 -E=0.15%
-E'=0.10% ---E=0.10%
co 80--e'=0.05% 2U 80--e=0.05%
o e'=0.01% O e=0.01%
E 60t -\ E 60

E 40- E 40-

20 20


00 1 0 20 30 40 50 60 70 80 90 00 1 0 20 30 40 50 60 70 80 90
Dose [Gy] Dose [Gy]
A B
Figure 2-4. Quality of DVHs for duality gaps e'=0.01 .~ 0.05' .~ 0.1 and 0.15' A) The
target coverage is nearly identical. B) The saliva gland sparing for the different
duality gaps is similar, but the solution for e'= 0.15' sacrifices one saliva
gland. The sparing criteria is marked hv a star.


receive more than 30 Gy in order to be spared. This point is marked by a star in Figure

2-4B.

Each of the duality gaps achieves good target coverage. While they each provide

similar saliva gland dosage, the plan obtained using e' = 0.15' slightly surpasses the

sparing criteria used for saliva glands.

2.6.4 Spatial Coefficient Results

To assess the possible treatment plan improvement afforded by spatial coefficients,

spatial parameters were tuned and then compared to treatment plans obtained without

using spatial information. To demonstrate the spatial coefficients, Figure 2-5 di pl oni~ the























10 20 30 40 50 60 70

Figure 2-5. The spatial coefficients used for two cases.


coefficients used for two cases. In addition to tuning A, p and /3 to values of 1.07, -0.32

and 0.77, respectively, a minimum spatial coefficient of 0.025 was also set for target voxels.

By definition, the maximum value of a spatial coefficient is 1.

These spatial parameters generally produce treatment plans of nearly identical

quality to the best plans obtained without using spatial information, though with the

added benefit of preventing misleading dose-volume histograms. In some cases, the spatial

coefficients were able to outperform the non-spatial plans. Figures 2-6 and 2-7 illustrates

two such cases.

In Figure 2-6, the spatial coefficients yield improved target coverage and spare all

saliva glands, as opposed to the non-spatial plan which only spares three of the four saliva

glands. There is less dose outside the desired target in the plan using spatial coefficients.

In Figure 2-7, the spatial coefficients reduce the amount of overdose in the primary

targets. In this patient, both the spatial and non-spatial plans spare all saliva glands.

2.6.5 Warm Start Results

Warm start solutions were obtained using the interior point method and the projected

gradient algorithm (Nocedal and Wright [31]). The interior point method warm starts

were tested with each Hessian possibility and a large duality gap of 200, both with and

without insignificant heamlets removed. The projected gradient algorithm was tested using















Target DVHs: Non-spatial

------ 1










10 2030 4050 6070 80 90
Dose [Gy]
Saliva gland DVHs: Non-spatial

-left parotid gland 1
S ---1.-f Euibi j dbli3ij it*L l3 10.3

\ r~gr~l E~iTbin lT~li3ijitLi glar..i


Target DVHs: Spatial

100 -----.










00 10 20 3040 5060 7080 90
Dose [Gy]
Saliva gland DVHs: Spatial

100: -.-left parotid gland
---left submandibular gland
-right parotid gland
6 ~---right submandibular gln








0 10 2030 40 50 67080 90
Dose [Gy]
Snatial


100
.


ti
LL
a,
E
o


0b 10 2030 4050 6070 80 90
Dose [Gy]
Non-snatial


-6 -4 -2 U 2 4 b -4 -2 U 2 4

A B

Comparison of spatial and non-spatial treatment plans. A) Non-spatial
parameters result in slightly low target dosage and fail to spare one saliva
gland. B) Spatial parameters allow for improved target coverage and spare all
saliva glands.


Figure 2-6.














Target DVHs: Non-spatial

100 -----










00 10 2030 4050 6070 80 90
Dose [Gy]
Saliva gland DVHs: Non-spatial

100 -left parotid gland
---left submandibular gland
~right parotid gland

tE. P r~gr~l ELITC. 13Fu~ljitL jl 311.3






0010 2030 4050 6070 80 90
Dose [Gy]
Non-spatial


Target DVHs: Spatial

-----


1020 3040 5060 7080 90
Dose [Gy]
Saliva gland DVHs: Spatial

-left parotid gland
---left submandibular gland
-right parotid gland
---right submandibular gland


100
80


00 10 20 3040 5060 7080 90
Dose [Gy]
Spatial


A B

Figure 2-7. A) Non-spatial parameters result in slightly low target dosage and fail to spare
one saliva gland. B) Spatial parameters allow for improved target coverage and
spare all saliva glands.










several stopping criteria and without insignificant b~eamnlets removed. It was observed

that the projected gradient algorithm is fast enough that the time required to remove and

re-insert the insignificant beamlets as necessary caused the algorithm to slow down. To

be theoretically close to optimal, the interior point method used after the warm start has

duality gap of 0.001 and no beamlets removed.

The determine the how close the warm start solution is to the final solution, the

percent improvement in objective function value the final solution obtains over the warm

start is measured. To assess how close to optimality the final solutions using a warm start

are, the percentage by which their objective function values are greater than the objective

function value of a near-optimal solution is measured. Lastly, the decrease in run times

over obtaining a near-optimal solutions are provided. These results for the interior point

and projected gradient warm starts are di; 1lai-x d in Tables 2-4 and 2-5, respectively.

From Table 2-4, it is clear that using an interior point warm start can provide

significant time savings over the near-optimal solution times. There is also a significant

increase in the FMO objective function value. From the amount of increase in the

objective function value, the interior point warm start does not appear to converge to

the optimal solution, and is unlikely to provide acceptable solutions. It is interesting to

note that the improvement from the warm start solution to the final solution is very small.

This indicates that KKET information obtained from the warm start and used in the final

algorithm were unhelpful in improving the solution.

For the projected gradient algorithm, once there is less than 6 percent decreases from

one iteration to the next, the algorithm terminates. Several 5 values are tested. As with

the interior point warm starts, the projected gradient warm starts also provided significant

time savings, as shown in Table 2-5. The final solutions from the projected gradient warm

start methods are nearly identical to the near-optimal solutions. The final interior point

method also significantly improves the objective value of the warm start solution. This

implies that despite not having KKET information about the warm start, the interior point










algforithm is still able to converge to the optimal, or at a least near-optimal, solution usingf

the KKET value approximations and adjustments to the warm start vector described in

Section 2.5.4.
















O~~Ln
013040


cr3

cr,


03cr
000


oO


003


003


V3000


~~0~~~0~~~00000000





0~000~0000~00000000












000000000000000000


me
%%,


so
9%


aa
aa


UIm
0001


a


UIm
0001


~ulLn Ln Ln Ln Ln Ln Ln Ln Ln Ln Ln Ln Ln Ln Ln Ln Ln Ln


bD
~ C
m m
~

ecb







a


cb
m
m

X


080808080808080800


aaaaaa
aaaaaa
666666

13 13 13 13 13 bD bD bD bD bD bD


U1

~~~~~
P~PPP



















bmOQCOCUMM000~
01ncr3bm3 VO010O
oCoiorodcjOO


OOOHOOO0mmow~ub
000000000+000000





0~0~0~000000000

















000000000000000


E
~do


,"'" ,~
t~~E
`C)~Cb
,
"a~
o


O
N
i
a,
-o
a
N

"
cb

m

X


aaaaa
aaaaa
66666
~~~~~
~~~~~

UIUIUIUIUI


UIUIUI


PPPPP~s~s~s










2.7 Conclusions

The primal-dual interior point method is an effective algorithm for obtaining fluence

maps that deliver quality treatment plans. The proposed Hessian alternatives appear

to converge to the optimal solution, even when insignificant beamlets are removed. The

removal of the insignificant beamlets provides significant time savings in all instances. The

interior point method may also be run with a duality gap as large as 20 and still achieve

quality treatment plans, thus decreasing the amount of time required to run the algorithm.

Of the implementations tested, the fastest method that still provides quality solutions

without using a warm start is to use the Single Approximation Hessian alternative, remove

insignificant beamlets and employ a relative duality gap of 0.1

When the interior point method is started with one of the warm starts discussed, time

savings were again significant. Although the interior point warm starts generally provided

more improvement in computation time than the project gradient warm starts, the final

solutions using the projected gradient warm starts were much closer to optimality. The

fastest and most effective warm start method is to use the projected gradient algorithm

with 6 = 500, followed by the interior point method with e = 0.1 and the Single

Approximation Hessian. This combination results in a near-optimal solution with an

average total computation time of 8.32 seconds.










CHAPTER 3
BEAM ORIENTATION OPTIMIZATION

3.1 Introduction

In a typical head-and-neck treatment plan, radiation beams are delivered from 5-9

nominally-spaced coplanar orientations around the patient. These coplanar orientations

are obtained from rotating the gantry only. As shown in Figure :3-1, several components

of a linear accelerator can rotate and translate to achieve more orientations than those

obtained from rotating the gantry. The available orientations consist of the orientations

obtained from rotation of the gantry, collimator and couch, as well as the three translation

directions of the couch.












p~*:








Figure :3-1. A linear accelerator and the available movements; the gantry rotation is
highlighted.


BOO is the problem of selecting from the available beam orientations the best set

to use in delivering a treatment plan. Given a fixed set of beams, different fluence maps

(radiation intensities of beamlets) yield treatment plans with different qualities. Thus, the

quality of an optimized fluence map should be considered when selecting a set of beam

orientations to use in a treatment plan.









3.2 Literature Review

M lityJ approaches have been taken to solve the BOO problem. Evolutionary

algorithms (Schreibmann [29]) and variants of evolutionary algorithms, particularly

genetic algorithms (Ezzell [32], Haas et al. [28], Li et al. [33]) have been emploi-e I Li

et al. [34] use a particle swarm optimization method, which is conceptually based on

evolutionary algorithms. Bortfeld and Schlegel [35], Djajaputra et al. [36], Lu et al. [37],

Pugachev and Xing [38], Rowbottom et al. [39] and Stein et al. [40] have all emploi-e 4

variations of simulated annealing to determine a beam solution. Soderstrom and Brahme

[41] selected coplanar beam orientations using two measures, entropy and the integral

of the low frequency part of the Fourier transform of the optimal beam profiles, both of

which are based on the size and shape of the target structure. Soderstrom and Brahme

[42] also use an iterative technique to determine the optimal number of coplanar beams

required using BOO. Das and Marks [27] use a quI I-;--N. vi.on method. Rowbottom et al.

[43] use artificial neural network algorithms to select beam orietations. Gokhale et al.

[44] use a measure of each beam's 1. 1111 of least re -!-1 Ills from the patient surface to

the target location to determine the best beam directions. Meedt et al. [45] use a fast

exhaustive search to obtain a non-coplanar solution. The concept of beam's-eye view

(BEV) has also been commonly used to approach the BOO problem (Chen et al. [46], Cho

et al. [47], Goitein et al. [48], Lu et al. [37], Pugachev and Xing [38, 49, 50]).

Despite the varying techniques to quantify the quality of a beam solution, it is widely

accepted that the optimal solution to the FMO problem presents the most relevant

measure (Bortfeld and Schlegel [35], Dj ii Ilputra et al. [36], Holder and Salter [51], Lee

et al. [20, 21], Li et al. [33, 34], Meedt et al. [45], Morrill et al. [52], Oldham et al. [53],

Rowbottom et al. [39, 43, 54], Schreibmann et al. [29], Soderstrom and Brahme [41],

Stein et al. [40], Wang et al. [55, 56], Woudstra and HE ~iiin Il [57]). Given this accepted

measure of treatment quality, the shortcoming of the previous works is twofold. First,

they predominantly only consider coplanar angles, and not necessarily even the entire










coplanar solution space, while those that do consider non-coplanar beams only consider

a hand-selected subset of the available orientations. Second, the 1 in 4 Gry of the previous

studies do not select heam solutions using the FATO problem as a model for determining

quality; instead, the beam solutions are chosen based on scoring methods (e.g., BEV, path

of least resistance) or approximations to the FAIO. By not optimizing the beam solution

with respect to the exact FMO problem, the BOO methods cannot guarantee convergence

to an optimal solution.

Of the previously cited works, only Das and Marks [27], Gokhale et

orientations. This is likely due to the computational difficulties associated with the

inclusion of non-coplanar orientations as well as the widespread belief that non-coplanar

orientations do not improve the quality of a treatment plan.

Also, of those works that addressed non-coplanar beams, Das and Marks [27] require

that the beam distances he maximized, essentially requiring that heam solutions must

he equi-distant and thus restricting the size of the solution space; Meedt et
consider :3,500 heams (a minute subset of orientations available by rotation of the couch

and the gantry); and Wang et
With the exception of Das and Marks [27], Haas et
the previous studies have based their BOO approaches not on a beam solution's optimal

solution to the FATO problem, but on locally optimal FMO solutions or on various scoring

techniques. Without basing BOO on the optimal FMO solutions, the resulting beam

solutions have no guarantee of optimality, or even of local optimality.

3.3 Model Formulation

The goal of radiation therapy treatment planning is to design a treatment plan that

delivers a prescribed level of radiation dose to the targets while simultaneously sparing

critical structures by ensuring that the level of radiation dose received by these structures

is less than a structure-specific radiation dose. These two goals are contradictory if the










targets are located near critical structures. This is especially problematic for certain

cancers, such as tumors in the head-and-neck area, which are often located very close

to, for instance, the spinal cord, brain stem and salivary glands. In order to model the

BOO problem, a quantitative measure that appropriately makes trade-offs between

these contradictory goals must be developed. Let F(0) be a black-box function that

quantifies the quality of the treatment plan if radiation is delivered from beam vector

8 = (01,...,0k), Where k is the user-specified number of orientations that may be used. F

is formulated in such a way that the optimal plan yields the minimum function value.

For k beams orientations to be optimized in the treatment plan, the vector of decision

variables representing the beam orientations is defined as 8 = (01,...,0k)T. The decision

vector 8 is used as input into the black-box function F(0) to determine the ability of the

beam vector to deliver the prescribed treatment without unduly damaging normal tissue

and critical structures. The BOO problem is then formulated as


min F (0)

subject to On E B h = 1,..., k,


where B is the set of candidate beams. The candidate set of beams can be selected

according to any user-specified criteria, for example, the beams can be coplanar or

non-coplanar, continuous or discrete, or only represent a subset of the available beams.

It is also possible to fix some beams and only optimize a subset of the total number of

beams to be used. Theoretically, the linear accelerator is able to capture a continuous set

of orientations, but due to machine tolerances, the actual beams delivered may not be

exactly the desired beams. Therefore, it is common to only consider a discretized set of

beam orientations.

In our BOO model, the black-box function F(0) is the convex FMO problem

described in Section 2.3, thus ensuring an exact measure of the quality of each beam

vector. Even though F(0) is convex, this formulation of the BOO problem is fundamentally










nonlinear because the physics of dose deposition change with each beam orientation, that

is, the effect of a beam on each patient can be drastically different than the effect of a

neighboring beam. To illustrate the nonlinearity of the problem, Figure 3-2 shows the

FMO problem as a function of just two coplanar beam angles. From this illustration, it is

evident that the FMO function, particularly in higher, more realistic dimensions, is likely

to also be multi-modal.

Although the FMO problem itself can be solved quickly using the convex model

presented in Section 2.3, in order to perform the FMO, lengthy calculations must be made

in order to determine each candidate beam's effect on the patient. These calculations,

described in Section 3.5, require a 13 minutes per beam to calculate, and thus make each

evaluation of the FMO problem expensive. Despite the time required for each function

evaluation, the limiting factor in beam orientation optimization is the hard drive space

required to store the beam data for each candidate beam. If the candidate set of beams is

small, this data can be pre-computed and stored, allowing the FMO problem to be solved

quickly in the BOO problem. But, if the candidate set of beams is large-for example,

consisting of non-coplanar orientations--then the data cannot be pre-computed due to

storage requirements.

Because of these difficulties with the BOO problem, previous studies have been largely

unable to consider the entire solution space of available beams. By using the response

method, which is specifically designed to model expensive nonlinear black-box functions,

we can iteratively identify promising beam vector solutions and generate beam data for

these solutions on-the-fly, thus circumventing the issue of storage space and allowing for

the consideration of all deliverable beam orientations.

3.4 Mixed-Integer Model Formulation

As an alternative to the BOO model given in Section 3.3, if the set of beam

orientations B is finite, the BOO and FMO problems can be formulated together and

solved simultaneously as a mixed-integer linear or nonlinear program (D'Souza et al. [58],















4000


3000


>2000 \




1- 30S 50 lt

300
240 30035
180 -1.--.~240
10660120 180

Angle 2 0 0Anl1



Figure 3-2. FMO value as a function of two angles.


Ehrgott and Johnston [59], Ferris et al. [17], Lee et al. [20, 21], Lim et al. [60], Shepard

et al. [22], Wang et al. [61]). The FMO formulation can be combined with BOO in the

following model. Let ye be a binary variable indicating whether or not beam 8 E B is used.

If beam 8 is used in the treatment plan, then all the beamlets in 8, Be, are 1 o"

that is, they can have positive fluences up to some pre-determined maximum intensity M~.

The simultaneous BOO+FMO MIP model is then


minimize F(z)

subject to zys = DijsXi j = 1,., vs, a = 1, ..., S
h= 1 = P.
xi < M~yo i E Bo, 8 E B


8BE
xi >0 i EBo, 0 6B

ye E {0, 1} 0 E










In order to solve such a problem, all beam data must he pre-computed for every heam

orientation. As described in Section :3.5, heam data requires a tremendous amount of

time and space to compute and store. Because of this requirement, only a small subset of

all possible beam orientations can he considered due to time and space constraints for a

BOO+FMO MIP formulation.

3.5 Beam Data Generation

For each heam orientation that is considered, lengthy calculations must he made to

determine the beam's effect on the patient's tissue and organs. This includes determining

in which structure each voxel lies, which voxels are hit by which heamlets and the amount

of intensity of each heamlet is deposited in each voxel through which it passes.

Beamlet dose computation models used in IMRT rely heavily on r n, -tiracing

algorithms for voxel classification and determination of the radiological path (Fox et


the path of a radiation beam and classifies voxel centers as inside or outside of segmented

targets and critical structures. The radiological path is the effective distance traveled by

a beamlet when the effect of traveling through tissues of different densities is considered.

The exact radiologfical path of a beamlet through the patient is required to correct for

tissue heterogeneities in determining the dose deposition coefficients (Siddon [64]).

Siddon's r n, -tiracing algorithms (Siddon [6:3, 64]) have been the standard methods

used for r n, -tiracing in radiotherapy since the 1980s. In Siddon's polygon and voxel

r n, -tiracing algorithms for voxel classification (point-in-polygon testing), structures

are represented as :3D polygonal objects, known as Siddon Prisms, and the signs of

cross-products of rays passing through the polygons are used to determine whether a voxel

lies inside or outside a structure. Despite its overwhelming use, Siddon's algorithm for

pub~l-_on r n -tiracing becomes very costly due to the number of voxels in a patient. Fox et


cross-products by translating the polygon structure onto a coordinate system, replacing










the need for a cross-product by the sign of the second coordinate of each voxel in the

coordinate system.

In Siddon's algorithm for determining radiological paths (Siddon [64]), the radiological

path must he determined for each voxel for every heamlet. This involves computations

for millions of beamlet-voxel combinations. As reported by Jacobs et
amount of computational time is required for these repeated calculations. Fox et
combine the incremental voxel r n, -tiracing algorithm presented by Jacobs et
method of virtual stereographic projection to significantly reduce the computational cost

of obtaining radiological path lengths.

Using their polygon translation and incremental r n, -tiracing algorithms, Fox et
achieve a 100-300 fold improvement in computation time over Siddon's point-in-polygon

algorithm. Because of the significant reduction in computation time, these methods are

used to generate beam data.

Because these beam data calculations must he performed for each of millions of

beamlet-voxel combinations, heam data generation is a lengthy process, requiring a 1:3

minutes per beam using the algorithms described by Fox et
formulation, the beam vector is pre-determined and the beam data for the beam vector

is calculated once and stored a priori. For a 5-heam case, this requires ml50 MB of space

to store. As with a typical FMO problem, in a simultaneous FMO+BOO mixed-integer

programming (\!lP) formulation, heam data for each of the candidate beams in a must

he generated a priori. If candidate beams are considered only for coplanar angles on a 100

grid, that is, only every 10th angle, heam data would have to be computed for :36 heams,

which requires m5 hours to compute and x800 MB of space to store. If we also wanted to

consider the possibility of rotating the couch on a 100 grid in addition to the gantry, heam

data would then have to be computed for :362 heams, which would require m170 hours and

S60 GB of space for just one plan.










Clearly, the storage space requirements for each heam restricts the number of beams

that can he considered in a simultaneous FMO+BOO MIP formulation. This issue is

typically addressed by simply restricting the number of candidate beams in B. Lee et
[20] restrict the set B to only contain 18 pre-selected heam orientations, which can he

coplanar or non-coplanar. If only gantry and couch rotations are allowed on a 100 grid, a

beam set of 18 heams comprises only a small percent of the available beam orientations.

As more ranges of motion are allowed, this percentage falls even further. The inclusion

of all possible beam orientations significantly increases the size of the solution space and

could possibly allow for improved treatment plans, but the beam data for all orientations

cannot he pre-computed. In order to consider these orientations, we use a method that

allows us to generate the beam data on-the-fly only as necessary.

3.6 A Response Surface Approach to BOO

The shortcoming of the previous works on BOO is twofold. First, they predominantly

only consider coplanar angles, and not necessarily even the entire coplanar solution space,

while those that do consider non-coplanar beams only consider a hand-selected subset

of the available orientations. Second, the 1 in 4 Gry of the previous studies do not select

heam solutions using the FATO problem as a model for determining quality; instead, the

beam solutions are chosen based on scoring methods (e.g., BEV, path of least resistance)

or approximations to the FAIO. By not optimizing the beam solution with respect to

the exact FMO problem, the BOO methods cannot guarantee convergence to an optimal

solution.

Of the previously cited works, only Das and Marks [27], Gokhale et

orientations. Of these works, Das and Marks [27] require that the beam distances he

maximized, essentially requiring that heam solutions must he equi-distant and thus

restricting the size of the solution space; Meedt et









minute subset of orientations available by rotation of the couch and the gantry); and

Wang et
With the exception of Das and Marks [27], Haas et
the previous studies have based their BOO approaches not on a beam solution's optimal

solution to the FATO problem, but on locally optimal FMO solutions or on various scoring

techniques. Without basing BOO on the optimal FMO solutions, the resulting beam

solutions have no guarantee of optinmality, or even of local optinmality.

Because beam data generation is costly, a method that iteratively identifies only

promising beam orientations is required. The response surface (R S) method is such an

algorithm. In contrast to the previous studies, our approach to the BOO problem allows

for the inclusion of all possible beam orientations which are measured according to the

exact FMO problem, thus ensuring convergence to optinmality due to the properties of the

response surface method.

The RS method is designed to efficiently model expensive black-box functions. In this

application, the FMO solver is our black hox and the set of beams to be used is the input.

As in Alenian et
[68] and Jones et
3.6.1 Overview of Response Surfaces

The response surface method identifies promising solutions based on the performance

of previous solutions. The function value and expected intprovenient over the current

best solution of a certain point is estimated based on the function behavior learned front

previously sampled points and their calculated objective function values. The function

values of points are related by correlation functions that depend on each point's distance

front the previously sampled points. Front the correlation functions, the algorithm predicts

the probability that the best solution will improve at unexplored points in the solution

space. Using this probability, a promising solution is identified. For the BOO problem,










beam data only needs to be generated for these promising solutions, thus saving both

computation time and storage space.

The response surface method models the objective function as a stochastic process of

the form

F(0) = p- + e(0), (3-1)

where p is a constant representing an average of the function F and e(0) is a random error

term associated with the point 0. In the general case, the error terms between two points,

ii- 8() and 0(2), arT COrrelated by





where d(8( ), 0(2)) is a Weighted distance measure between 8(1 and 0(2). Ill1ultiVely, if tWO

points are very close together, the correlation between them will be close to one, similarly,

if two points are very far apart, the correlation between them will approach zero. Jones et

al. [69] propose the following weighted distance measure in general:



h= 1

where the parameters ch and ph are weighting factors corresponding to the importance

of each variable h and the smoothness of the function F in the direction of variable h,

respectively. If small changes in variable h cause large changes in the function F, then ch

should be large to reflect that two points with relatively small differences in the value of

variable h should be Il apart due to the large difference in their function values, and

thus have a low correlation. The parameter ch can take on any value, whereas 1 < ph < 2,

with ph = 2 corresponding to objective function smoothness and ph = 1 corresponding to

less objective function smoothness.

In the application to BOO, O = (01,...,0k) is the vector of k angles from which

radiation will be delivered. Because no beam is more important than another beam, each

beam orientation h contributes equally to the FMO function, so ch = c and ph = p for










all & = 1,..., k. To maintain tractability of the subproblems described in the following

sections, the angles are treated as though they are points on a line rather than points on

a circle and so a Euclidean distance metric is used to determine the distance between two

points. The weighted distance measure for BOO is then


d 0 0() = (1)_ p2)

where || ||, denotes the e,-norm. To ensure tractability of the subproblems described in

Section 3.6.2, the value p = 2 is used.

The idea of the RS method is to iteratively evaluate the true function F at certain

beam vectors 8, and then construct the conditional stochastic process given these function

values. This conditional stochastic process is then used to decide where to evaluate the

function F next. Due to the time and space required to generate the beam data necessary

to evaluate the function F, it is desirable to only evaluate points that will either improve

the best solution with a significant probability or significantly increase our knowledge of

the function. The optimization models to determine the next observation are described in

Section 3.6.2.

Let 8( ), ") be a previously sampled points. R, is the matrix of correlations

between the previously sampled points, y, is the vector of function values F(0C')) of the

previously sampled points and 4, and 6, be estimators of the average and variance of the

function F, respectively. The response surface algorithm is given by:


Initialization:

1. Choose values for the parameters c and p.

2. Choose an initial sample size, n, and a set of angles 8("), i = 1,..., n. Evaluate
the function F at each of these points, yielding the values yi, i = 1,..., n.

Iteration:

1. Compute or update the values of &L, R-], ps, So, and F, the minimum
observed objective function value.









2. Determine the next point to observe using one of the methods described in
Section 3.6.2 and call this point 8 "+)

3. Find the value y,+l = F(8 3)), set n <- n + 1, and repeat.

3.6.2 Determining the Next Observation

Because the function F is expensive to evaluate, we want to sample as few points

as possible. Thus, in each iteration, an optimization problem is solved that determines

the "b. -I next point at which to observe the true function F. Some of the optimization

problems that have been proposed in the literature depend on the uncertainty of the

predictor as a function of 8, as well as the expected improvement over the current best

solution (Jones [68], Jones et al. [69]).

Let r,(0) be the vector correlations between 8 and the a previously sampled points.

The uncertainty is then given by

2/~~~ 2 ~[1 -1TR;'~1r~LT,-(0)]" 2
s (0) = &(1-r(0 ,0 + 1R1

where

o-, = (y,-1,) R, (y, 14,)

is the estimator of the variance O.2 based on the n observations. The expected improvement,

denoted I,(0), is given by

I,(0) = s,(0) [z@ (z) + 4(z)] (3-4)

where

sne (0)

and F, = min~yl,..., y,} is the current best solution and F,(0) is the estimated function

value of 8 given the a previously sampled points. # and 4 are the c.d.f. and p.d.f. of a

standard normal random variable, respectively.










The selection of the next point will be based on selecting the point that maximizes

either the uncertainty or the expected improvement, or a combination of both. Denote the

beam vector to be chosen as the vector 0.

3.6.2.1 Maximizing the expected improvement

.Jones [68] and .Jones et
point 8 for which the expected improvement over the current best solution value, I,2(0), is

largest. This corresponds to solving the following optimization problem:


max In (8)

subject to Oi, E B b = 1,..., k


Although this is a difficult optimization problem, it can he solved using a branch-and-hound

technique, but in order to do so, an upper bound on I, (0) must he obtained. This can

he done by solving for the expected improvement in equation (:34) while substituting

an upper bound on the uncertainty and a lower bound on F,(0), used in equation (:35)

to determine the value x. The method of hounding F,(0) is taken directly from .Jones

[68] and .Jones et
Q (0) is improve from the original formu~lationn inn .one pt
instabilities, and is presented in Section :3.6.2.2. The branch-and-hound algorithm used to

maximize 1,(0) is described in Section :3.6.2.3.

3.6.2.2 Obtaining an upper bound on the uncertainty
Due, to the complexityT of the c2(0) function, maximizing the uncertainty is a difficult

problem to solve. It can he relaxed into a linearly constrained quadratic programming

problem as follows (.Jones et
maximization problem is an upper bound on the uncertainty that can he used in

determining an upper bound on I,2(0) as described in Section :3.6.2.1.

Let r = {rl,...,r,z}, where r is a vector of decision variables independent of 0. By

treating both r and 8 as decision variables, a quadratic objective function is obtained.









Because r is now a decision variable independent of 8, an equality constraint must be

added to the problem to ensure that r assumes the correct correlation values according

to the correlation definition in equation (3-2). This constraint is nonlinear, but it can be

relaxed by expressing the single equality as two inequalities (< and >) and then replacing

the nonlinear terms generated by In(r,) ansd c|| e(" || with linear underestimators

ai + biri and pi,h + qi,heh, respectively. The different types of linear estimators require

different values for ai, bi, pi~h and qi,h, and are differentiated by a superscript c for the

chord underestimators and a superscript t for the tangent line underestimators in the

model formulation, denoted Problem s2-UB.

Unfortunately, this relaxation provided by Jones et al. [69] can become numerically

unstable if two sampled points are very close together. If such a situation arises, the

bounds of the corresponding correlation value can become so close that due to round-off

error, the lower bound rC can become slightly larger than the upper bound rf resulting

in infeasibility. To avoid such an instability, instead of bounding ri using constraints, the

amount by which ri is outside of its feasible range is penalized by adding penalization

terms w = min{0, ri r } and I,"~ = min{0, ri rg }. This final formulation is given in

Problem s2-UB. This formulation has only two more variables and two more constraints

for each sampled point, so the increased problem size does not significantly increase the

amount of time required to solve the problem.










PROBLEM s2-UB: Choose r and 8 to

min -~2i [[1 -1TR-r]2



subject to rr (af +bys)c (p,t -q B,4) <0 i= 1,..n
h= 1

(ai + ci~ (p,C + qf,0) < 0 i =1, n
h= 1
I,.'<0 i= 1,...,n






la'T < On






Using the upper bound on the uncertainty provided by Problem s2-UB, the point

yielding the maximum uncertainty is obtained by using the same branch-and-bound
method described in 3.6.2.3, except that s (0) is maximizedr rather than I,(0).

Alternatively, another approach would be to choose the next point based on

maximizing uncertainty rather than the expected improvement. The branch-and-bound

approach described in Section 3.6.2.3 can be adapted to solve that problem rather than

maximizing the expected improvement.

3.6.2.3 Branch-and-Bound

A branch-and-bound method is used to determine the maximum expected improvement

in each iteration. At some point in the algorithm, a points, O ',...,0 "), have already

been observed. The solution space is divided into regions based on these previously

sampled points and consider each region as a separate subproblem.

Each of these subproblems is solved using branch and bound. First, the upper bound

on the uncertainty is determined as described in Section 3.6.2.2 using the subregion's










lower and upper bounds on 0. Next, the lower bound FL on Ez(0) is determined using the

method in .Jones [68] and .Jones et al. [69].
The upppr boundl on c2(0) and lower bound on F are now used to determine an

upper bound on I,2(0) over the current subregion by solving for I,2(0) substituting

F,(8) = FL and s,2(8) = s" as described in .Jones [68] and .Jones et al. [69]. In addition,

the 8 that yielded the maximum uncertainty can he used to evaluate the function I,2(8)

yielding a lower bound on I, (0) over the interval liz < 8h < uh, h = 1,. ., k. This value is

used to update the current best lower bound found (i.e., if the current best lower bound is

less than the new lower bound found, the current best lower bound is replaced by the new

one; otherwise, the current best lower bound is unchanged).

If the upper bound is less than the current best lower bound, the subregion is

discarded as not interesting. If the lower and upper bound are very close, we ;?i that

we have found the optimum over the current subregion. Otherwise, the upper bound is

significantly larger than the current lower bound, so the subregion is further divided into

subregfions as described below and the procedure is repeated for each of the new regions.

This is the branching step.

At some point, there are no more subregfions to consider, as we have either decided

they are not interesting or have found the optimal solution for that subregion. Then, the

algorithm terminates and the current best lower bound is the optimal solution for I1z(8)

over the current region.

This branch-and-hound procedure is applied to each of the regions, and the overall

largest I,t(0) value is then the maximum I1z(0), and the corresponding 8 is the next point

at which to evaluate the Fl\O function.

Selecting the subregions. An important component of the branch-and-hound

algorithm is the method of selecting the subregions. The definition of these subregions,

as well as1 the order in which they are explored, can have significant impact on both the

amount of time and memory required to perform the algorithm. As our implementation









of the branch-and-bound method requires that the entire solution space be divided into

subregions before the branch-and-bound algorithm begins, the selection of these initial

regions may also affect the speed of the algorithm.

Initial regions. Before beginning the branch-and-bound process, the solution space

of the decision variables, On E [0, 360] for all & = 1,. ., k, is divided into a set of initial

regions. If 8 represents non-coplanar orientations, we consider two owsi~ of selecting the

regions defined by the non-coplanar orientations. First, we consider the entire solution

space as the only region, that is, instead of dividing the solution space into several

subregions, we only consider one subregion that encompasses the entire solution space (see

Figure 3-4A).

Second, denote a subset of variable indices HC { 1,..., k}. For each index h e H,

order the a previously sampled points increasingly by h. For each previously sampled

point i = 1, .. ., a 1, consider the regions defined by lh = 0 and uh = 360 for h ( H,

and la, = 0~ and usI = 0 .I! Also .onIsider. the region dlefinedl byv lh = 0 and ush = 36i0

for h ( H, and la, = 0 and u, = 0 ). Similarly, consider the region defined by lh = 0 and

us, = 360 for it H, and la = 0 *) and us = 360. Figures 3-4A-3-4D) illustrate the initial

regions for different H values where k = 2. Denote the initial region set where H = 0

as BO (Figure 3-4A), H = 1} as B1 (Figure 3-4B), H = 2} as B2 (Figure 3-4C) and

H = {1, 2} as B2 (Figure 3-4D).

Note that in the coplanar case, it is only necessary to test the initial region scheme for

one angle because the angles are interchangeable.

Bounds for discrete and continuous variables. If 8 is discrete, the points on the

boundary between between the two subregions will be contained in both subregions, thus

creating an inefficiency. This can be seen in Figure ??, whr stepita hc

we branch and the blue line represents the division of the region into two subregions. The

boundary line is contained in both the top interval and the bottom interval. This overlap

can be avoided when 8 is integral by adjusting the bounds between subregions in such a




















































Initial region scheme B3



300


240 4


180


120 (


60



0 60 120 180 240 300 36
Gantry angle


Initial region scheme BO


Initial region scheme B1


240


S180


120


120


0 60 120 180 240 300 360
Gantry angle


0 60 120 180 240 300
Gantry angle


Initial region scheme B2



300


240 g


180


120 0


60



0 60 120 180 240 300 360
Gantry angle


Initial regions in the branch-and-bound algorithm. A) Initial regions with

H = 0 (BO). B) Initial regions with H = {1} (Bl). C) Initial regions with

H = {2} (B2). D) Initial regions with H = {1, 2} (B3).


Figure 3-3.









way as to prevent overlapping between any subregions. If the lower bound liz on 8h in a

subregion is fractional, then we discard the non-integral solutions by setting li, = [lal.

Similarly, if the upper bound to, on 8h in a subregion is fractional, then uhi = Luiz]. If the

liz and Uhi hounds are integral and liz = uhi, overlapping is avoided by setting liz = liz 1

(see Figure ??). If 8 is continuous, the bounds cannot he adjusted.

Branching scheme. The basic principle of the branch-and-hound method is to

decompose regions into smaller subregions in such a way that as many subregions as

possible can he discarded as unini.~ i. -1 it, leaving a reduced number of subregions that

must actually be searched. The branch-and-hound method is a well studied problem,

and as such, there are numerous methods of selecting the subregions. Regions may be

divided into two equal subregions bisectionn), or more generally, into multiple subregions

which may or may not he equal in size (niultisection) (Csallner et
Soubry [71]). Some other coninon methods include selecting only a subset of variables on

which to branch (Epperly et
(Barrientos and Correa [7:3], Thoai [74], Tuy [75]) and applying decomposition algorithms

(Phong et
In our branching step, we form the subregfions based on some point in the region. The

region is divided at this point along one of the indlices. In Figure :3-4A1 Of is the point

at which we branch. We branch by dividing the region horizontally into two subregions

at Of ", talking into account the adjustments to the bounds described above so as to avoid

overlapping regions. For k = 2, in each branching step, we alternately divide the region

horizontally (along index 2) and vertically (along index 1) as shown in Figures :3-4B-3-4D.

After br~anlching hlorizonrtally onlce at Of asi shown inl Figure :3-4B,: we examine thle top

region and select Of$) as Ithe point at which wve branc~h. We/ then branch by dividing: this

subregion vertically at Of$). We proceed in the same manner for 8( where we branch

horizontally, and so on until the convergence criteria is met.










In the general case, we divide the region into two subregfions along the branching

index while cycling through each of the indices & = 1,..., k sequentially. For the

branching index h, the bounds for one new subregion are lh = lh and uh = Ob,h -i

and the bounds for the other new subregion are lh = Ob,h and uh = Uh. The lower

and upper bounds on the region for the remaining indices are unchanged for both new

subregions, i.e. lh = lh and Uh = Uh for & / h.

In the non-coplanar case, a beam in 8 may be represented by more than one index.

For example, if a single non-coplanar beam consisting of couch and gantry rotation is

optimized, the vector 8 consists of Or representing the gantry angle and 02 representing

the couch angle. The branching index he { 1, 2} represents branching on either the

gantry angle or on the couch angle. If two such non-coplanar beams are optimized,

then 8 consists of Ox and 02 representing the gantry and couch angles of the first beam,

respectively, and 03 and 04 representing the gantry and couch angles of the second beam,

respectively. The branching index he { 1, 2, 3, 4} then represents branching on a single

component of a single beam.

Accounting for symmetry. In the case where 8 represents a set of coplanar beam

angles, the ordering of the variables in 8 is irrelevant to the FMO value obtained at 0. For

example, if 8( ) = (10, 20, 30, 40) and 0(2) = (20, 30, 40, 10), then F(0C 3) = F(0(2)). Thus,

it is redundant to consider both 8( ) and 0(2), and elimination of these redundant regions

can greatly decrease the size of the solution space.

For example, if we consider the two-dimensional case (k = 2), the solution space is a

square region with 0 < Or < 360 and 0 < 02 < 360. The points above the line 01 < 02 arT

equivalent to the points below the line, so we only need to consider one of these regions.

Clri we branch by splitting the region into four equal quadrants, as shown in Figure 3-5A.

If we arbitrarily choose to only examine the points above the line 01 < 02, then quadrant 4

can be eliminated.









































II Gantry angle u


A

Branching scheme
u2




()0(2
ob

(1
s eSb


Branching scheme


Branching scheme


Gantry angle


Branching scheme


Gantry angle


Gantry angle


Figure 3-4. Partitioning a region into subregions. A) Partitioning a region into subregions
without accounting for overlap. B) Preventing overlapping regions. C) Regions
after two branches. D) Regions after three branches.











360


4
360

0 360
360
A B

Figure 3-5. Accounting for symmetry. A) Accounting for symmetry in 2D. B) Accounting
for symmetry in 3D.


In three dimensions, the solution space is a cube. If we branch by splitting the cube

into eight equal cubes, the region to be examined is shown in Figure 3-5B, where the

origin is the back bottom left corner of the cube. From this figure, we can see that a

sizable portion of the solution space can be discarded.

In regions where there are both viable and redundant solutions (for example,

quadrants 2 and 3 in Figure 3-5A), the addition of constraints requiring that Or < ... < Ok,

in the problem of maximizing the expected improvement ensure that only the unique

portion of the region is considered.

If more than one non-coplanar orientation is optimized, a similar symmetry to the

multiple coplanar orientation symmetry exists. Consider an implementation where two

non-coplanar beam orientations are optimized, and these orientations are obtained from

rotating both the gantry and the couch. Each beam is represented by two variables

in the solution vector: one variable indicating the degree of gantry rotation, and one

variable indicating the degree of couch rotation. Let 01 and 02 be the gantry rotation and

couch rotation of the first beam, respectively, and 03 and 04 be the gantry rotation and

couch rotation of the second beam, respectively. Then, the solution vector {r, 02, 3, 84










is identical to the solution vector {03, 84, 81, 82}. Because the couch angle selected is

dependent on the gantry angle (and vice versa), this symmetry can be exploited by only

removing redundant solutions from one of the beam variables, that is, by requiring that

01 < 03 (removing redundancy from the gantry angles) or 02 84 (removing redundancy

from the couch angles). In general, if d degrees of motion are used to obtain m beam

orientations, and the linear accelerator motion variables are in the same order for each

beam, then Ok, I k+d I k1+2d ** 8k +(m-1)d foT Some k~ E {1,..., d}.

3.6.3 Method of Obtaining the Next Observation

The RS algorithm allows for two methods of selecting the next point to observe:

by maximizing the expected improvement or by maximizing the uncertainty. In these

tests, the point to observe is obtained by first selecting the point that maximizes the

expected improvement until the maximum expected improvement falls below a certain

threshold, and then switching to the point that maximizes the uncertainty. Once the

maximum uncertainty also falls below a certain threshold, the algorithm terminates. By

first selecting according to the expected improvement, the method quickly obtains a good

solution. By then selecting according to uncertainty, theoretical convergence to the global

minimum is ensured.

3.7 Neighborhood Search

3.7.1 Introduction

Fr-om Aleman et al. [79], we test the simulated annealing algorithm on the BOO

problem, as well as existing and new variants of a greedy neighborhood search heuristic

called the Add/Drop algorithm (see K~umar [80]) to obtain a good solution to the BOO

problem. In each step of the Add/Drop algorithm, a beam in the current beam set is

replaced by a neighboring beam that yields an improving solution. As with the simulated

annealing implementation, we also apply our new neighborhood to the Add/Drop

algorithm and compare its performance to a commonly used neighborhood structure.










3.7.2 Neighborhood Search Approaches

Neighborhood search approaches are common methods of obtaining solutions to

global optimization problems. For a vector of decision variables, a neighbor is obtained by

perturbing one or more of the decision variables. A neighborhood for a particular vector

of decision variables is the set of all its neighbors for a given method of perturbatingf the

decision variable vector. A solution is considered to be locally optimal if there are no

improving solutions in its neighborhood.

Both deterministic and stochastic neighborhood search algorithms have been applied

to a wide v .vi.~ iv of optimization problems. A deterministic neighborhood search algorithm

is one in which the entire neighborhood, or a pre-defined subset of the neighborhood,

is enumerated in each iteration to find an improving solution. Stochastic versions of

neighborhood search approaches, for example, simulated so~~! n., .11 randomly select

neighboring solutions in an attempt to find an improving solution in each iteration.

For the BOO problem, we consider two neighborhood search methods. The first is

a deterministic neighborhood search algorithm that finds a locally optimal solution, and

the second is the simulated annealingf algorithm, which, although based on neighborhood

searches, provably converges to the globally optimal solution for certain neighborhood

structures.

3.7.3 A Deterministic Neighborhood Search Method for BOO

Deterministic neighborhood search methods are optimization algorithms that

start from a given solution and then iteratively select the best point in the current

neighborhood as the next iterate. The best point in the neighborhood can be found by

complete enumeration if the neighborhood is small, or by optimization is the neighborhood

is large or if objective function evaluations are expensive. Due to the complexity of

the BOO problem, even when only a subset of available orientations is considered, we

will focus on smaller neighborhoods and use enumeration. The neighborhood could

alternatively be searched heuristically, for example by searching the neighborhood until









the first improving solution is found, rather than the best improving solution. If no

improved solution can be found the current solution is a local optimum.

In our implementation of the Add/Drop algorithm, a small neighborhood is desired

for enumeration purposes. In each iteration, a neighborhood for just a single beam is

considered. Cli-a beam set consisting of k beams is desired. Letting the neighborhood of a

single beam On in 0 be denoted as &h(0), the Add/Drop algorithm is as follows:


Initialization:

1. Choose an initial starting solution 80()

2. Set 0* = 80() and i = 0.

Iteration:

1. Select he { 1,..., k}), then generate 8 E &h(6 ).~

2. If F(0) < F(0*), set 0* = 0(i+l) = 0 and set i <- i + 1.

3. If all points in U zA(0(i))") have been sampled w-ithoult improvement, stop w-ith
0* as a local minimum. Otherwise, repeat Step 1.


3.7.3.1 Neighborhood Definition

In each step of the Add/Drop algorithm, a beam in the current solution is replaced

with an improving beam in its neighborhood. Rather than define a neighbor as related

to an entire beam vector, the neighborhoods of individual beams are considered. The

neighborhood of a single beam On in 8 is defined as


M(0) =(081,..,0a_l,0mod360,0s 1,...,0k)



In other words, the neighborhood of a beam is all beams within + 6 degrees taking into

account the cyclic nature of the angles. The cyclicality of the angles refers to the fact

that all angles can be represented by degrees in [0,360]. For example, 4000 = 400 and

-1000 = 2600. The expression 8 mod 360 captures this cyclicality.










3.7.3.2 Neighbor Selection

The process of selecting a neighboring point in each iteration consists of two steps:

selecting the index b to change and then selecting an improving angle in Niz(0) to replace

0;z. If h is selected as i mod k + 1, the algorithm will cycle through each index sequentially,

similar to a Gihhs Sampler (see, for example, Geman and Geman [81] and Gelfand and

Smith [82]). The Gihhs Sampler also uses a similar two-step approach to generating a

new point by sequentially generating a new value for each variable in turn. If h is selected

randomly in each iteration, the resulting algorithm is similar to a Hit-and-Run method

(see, for example, Smith [8:3] and Bi41isle [84]), in which a variable to be changed is selected

randomly, and then a new value for that variable is also selected randomly within a

neighborhood.

Once b is selected, the new value for 8hi can he generated by enumeration or by a

heuristic method. The Add/Drop algorithm compares the quality of the new solution to

the current solution, and then only accepts improving solutions. This greedy approach

results in a locally optimal solution.

3.7.3.3 Implementation

The index of beam angle to be changed in each iteration, h in Step 1 of the algorithm

in Section :3.7.3, is chosen as h = i mod k + 1 to cycle through each index in a sequential

manner. In the Add/Drop implementation, once b is determined, 8 in iteration i is

chosen as 8 = argmrineagatyei, {F(j)}. By replacing ~ac~h he~am hv the mnost imnproving:

neighbor, the Add/Drop algorithm is a greedy heuristic which terminates when there is no

improving neighbor for any heam.

A multi-start aspect is added by repeating the algorithm with multiple initial starting

points. For example, one strategy to select starting points would be to select a random

starting point according to a particular distribution. Another strategy would be to select

an equi-spaced solution and rotate it a fixed number of times to obtain new starting points

until the initial equi-spaced solution is repeated. Equi-spaced heam solutions are common










in clinical practice for an odd number of beams. The reason that such a method is not

generally used in practice for even-numbered heams is that the resulting beam set would

contain parallel-opposed heams (heams that lie on the same line), which are not used by

convention as it is believed that the effect of a parallel-opposed heam is very similar to

simply doubling the radiation delivered from a beam. If an equi-spaced solution is not

possible given a beam set of k heams and the discretization level of the candidate beam

set B, then th~e solutions canl he rounded so that 0 oi) E B, hI = 1,..., k.

3.7.4 Simulated Annealing

The simulated annealing algorithm used is similar to the classical simulated annealing

approach proposed in K~irkpatrick et al. [85]. The simulated annealing algorithm is based

on the Metropolis algorithm, wherein a neighboring solution to the current iterate is

generated, and if it is an improving point, it becomes the current iterate. Otherwise, it

becomes the current iterate with probability exp{AF/T}, where AF is the difference

in FMO value between the current iterate and the newly generated point and T is the

temperature, a measure of the randomness of the algorithm. If T = 0, then only improving

points are selected. If T is very large, then any move is accepted, which is essentially a

random search.

The simulated annealing algorithm starts with an initial temperature To and performs

a number of iterations of the Metropolis algorithm using T = To. Then, the temperature is

decreased according to some cooling schedule such that {Ti)} 0.

Obvious parallels can he drawn between the simulated annealingf algorithm and the

Add/Drop neighborhood search method described in Section 3.7.3. While the Add/Drop

algorithm deterministically searches the neighborhood for improving solutions, the

simulated annealing algorithm randomly selects neighboring solutions. Rather than

being limited by the ability to only move to improving solutions, the simulated annealing

algorithm may still move to a non-improving solution with a certain probability, thus









allowing for the escape from local minima. The Add/Drop algorithm, on the other hand,

is a greedy algorithm that is specifically designed to find local minima.

The simulated annealing algorithm is essentially a randomization of the Add/Drop

algorithm. In addition to the added randomness, the possibility of changing more than

one beam in each iteration is allowed by selecting a set of indices HC { 1,..., k} to

change, rather than just selecting a single index h. The simulated annealing algorithm is

as follows:


Initialization:

1. Choose an initial beam set 8(o) and calculate its FMO objective function value
Fo

2. Set 8 = 0(o), F = Fo, i = 0.

Iteration:

1. Select H C {1,..., k}, generate 8 E UheH & (i)),)) and calculate its FMO
objective function value F.

2. If F < F, set F F, Fiaz F, 0(i+l) = and 7 = 0. Otherwise, set Fi 1 = F
and 8( ) = 8 with probability exp{(Fi F)/Ti}.

3. Set i <- i + 1 and repeat Step 1.


The simulated annealing algorithm has been previously applied to the BOO problem.

Bortfeld and Schlegel [35] use the .-I" simulated annealing algorithm described by Szu

and Hartley [86] which employs a C 111 Ov: distribution in generating neighboring points.

Stein et al. [40], Rowbottom [39] and Dj ii Ilputra et al. [36] also use a Cauchy distribution

in generating neighboring solutions. Lu et al. [37] randomly select new points satisfying

BEV and conventional wisdom criteria and Pugachev and Xing [38] randomly generate

new points and then vary them according to an exponential distribution. All accept

improving solutions, and with the exception of Rowbottom et al. [39] who only accept

improving solutions (essentially Ti = 0 for all i), all accept non-improving solutions with a










Boltzmann probability. None of the previous BOO studies employing simulated annealing

use the exact FMO as a measure of the quality of a beam set.

3.7.4.1 Neighborhood Definition

Two neighborhood structures are explored. The first neighborhood is similar to that

described in Section :3.7.3.1 in that a neighborhood Az(0) is considered for only a single

beam index h { 1, .. ,k }, just as in the Add/Drop method.

As an extension to changfiningf a single angle in each iteration, we also consider

a neighborhood that involves changing all beams in each iteration, corresponding to

H = {1,..., k} in Step 1 of the simulated annealing algorithnx in Section 3.7.4. This

neighborhood is defined as #1(0) = Uita~h(0). Again, the neighborhoods for the

individual beams are defined as in the first method, with hounds of + 6 degrees.

3.7.4.2 Neighbor Selection

The method of selecting a neighbor depends on the neighborhood structure as

described in Section :3.7.4.1. In the first method where only one beam is changed at a

time, a neighbor is selected using the randonlized approach described in Section :3.7.3.2.

Once b is selected, the probability of selecting a particular solution in Az (0) where the

new 8 is d degrees front Bi is P{D = d}, where D is the realization of a random variable of

some probability distribution defined on the interval [-6, -6 +1,...,6].

For the neighborhood #1(0) where all beams are changed in an iteration, the new

value for each heant he {1,..., k} is generated from Mz(0) in the same manner described

above.

3.7.4.3 Implementation

In addition to easing our algorithms on the exact FMO solution rather than on

heuristics or scoring measures, our simulated annealing approach differs front the previous

studies in the distribution used to generate neighbors, the definition of the neighborhood,

the cooling schedule and the number of iterations/restarts used. Not only do we use a

new neighborhood structure, but also a geometric probability distribution rather than a










uniform or C 1111 ily distribution on the neighborhood. The geometric distribution is similar

in shape to the C 1 t. 1sy distribution in that they both can have fat tails depending on

the choice of probability parameters. The fat tails of these distributions allow for points

far away from the current solution to be selected as successive iterates, which potentially

increases the likelihood of finding a globally optimal solution. The geometric distribution

has the added attractiveness of producing discrete solutions, which is desirable for the

BOO problem in which discrete solutions are preferred.

By using the cooling schedule Ti 1 = ~Ti with a~ < 1, the sequence of temperatures

{Ti} converges to zero as the number of iterations increases. In our approach, the

neighborhood of a beam for both the Na(0) and #1(0) neighborhoods is defined using

6 = 180, that is, Na(0) = B. By defining the neighborhood of each beam to be the entire

single-beam solution space, the simulated annealingf algorithm converges to the global

optimum when using the neighborhood #1(0) defined in Section 3.7.4.1. Though Na~(0)

is large, each beam in Na(0) is assigned a probability so that only the beams closest to

On have a significant probability of being selected. Figure 3-7A shows the probability of

replacing On with beams at varying distances using probability p = 0.25 for the geometric

distribution. Note that the current beam cannot be selected as a replacement.

As with the Add/Drop method, a multi-start aspect is added to the simulated

annealing algorithm by repeating the algorithm using several different starting points.

3.7.4.4 Convergence

Unlike many previously proposed simulated annealing algorithms, our algorithm

converges to the globally optimal solution to the BOO problem under mild conditions.

The following theorem summarizes these conditions.

Theorem 3.7.1. Suppose that

H={1,...,k}

limit, Ti = 0

6= 180









*There is a positive r,. e..1st.l.:7:;i of generating r,.;; .solution in the neighborhood.

Then our .simulated n...... al.:,:l l'l-" ~:thm converges to the 11g l..lrl optimum .solution in the



lim Fi = F* i;ttl77

where F* is the 11g l..lrl optimum value of the BOO problem.

Proof. This follows from Theorem 1 in Biilisle et
3.7.5 A New Neighborhood Structure

For the BOO problem, the neighborhood structure that is typically used for a vector

of beam orientations is simply the collection of beam vectors obtained from changing one

or more of the beams to a neighboring beam, where each heam has its own neighborhood



In addition to #;,(0), we consider a new neighborhood which we call a "flip"

neighborhood. The flip neighborhood of a beam index b consists of M,(0) plus a

neighborhood around the parallel opposed heam of b. The parallel opposed heam is

the beam 1800 away, that is,

h' = (0;z + 180) mod 360

The flip neighborhood can he defined as


Ni (0) (H1,.. ,Hh;z-1,0od:36018,0 iz.. ; Hk)E 6

: H t [Oi, 6. Bi, + 6] U [Oi, + 180 bB, O + 180 + 6P]

Note that the values 6 and 6" may be different. Figure :3-6 depicts a flip neighborhood

for a beam located at Oo degrees, the center of the top shaded wedge representing n#,(0)

where Bi, = 0.

The motivation for the flip neighborhoods arises from the observation that many

of the :3-heam simulated annealing plans generated using the regular neighborhood

contained two heams very close to two heams in the optimal solution (obtained by explicit











enumeration), while the third beam was very close to the parallel opposed beam of

the third beam in the optimal solution. Given this observation, it is intuitive that the

inclusion of the neighborhood around the parallel beam should provide improved solutions.

The neighborhoods &h(0) and ~ih(() with varying 6F values are applied to both

the Add/Drop and the simulated annealing frameworks. For the geometric probability

distribution used in the simulated annealing method, Figure 3-7B shows the probability of

selecting beams at different distances using a flip neighborhood with probability p = 0.25.

Note that the current beam cannot be selected as its own neighbor.


Figure 3-6. Mh(0) (top shaded area) and ~ih(() (top and bottom shaded areas) for 8h=0.


Geometric probability distnbution for standard neighborhood


Geometnc probably/ distribution for Olp neighborhood


-6 -4 -2 0 2
distance from current beam


4 6


distance from current beam

B


Figure 3-7. Selection probabilities. A) Mi~(0). B) iF((0)










3.8 Results

In addition to judging the BOO algorithms according to their computational time, the

plans must also be evaluated for clinical viability. All criteria used are those emploi-v I at

the Davis Cancer Center at Shands Hospital at the University of Florida.

3.8.1 Evaluating Plan Quality

In order to formulate an optimization problem, a quantitative measure of the

treatment plan quality is needed. This measure, the FMO function value, needs to

appropriately make the trade-off between the contradictory goals of covering targets and

sparing critical structures.

Typically, a good plan ensures that at least a certain percent of each target receives

the prescription dose. A coldspot occurs where less than a certain percent of the target

receives the prescription dose. Similarly, a hotspot occurs if a significant percentage of the

target receives more than the prescription dose.

3.8.1.1 Target coverage

Each of the plans contains two target structures, or planning tumor volumes (PTV):

one is the tumor mass observed from imaging scans, which we will call PTV2, and the

other is the PTV2 plus some margin specified by the physician, which we will call PTV1.

The PTV1 structure is used by physicians in case there are elements of the tumor mass

that cannot be seen from the imaging scans. The dose prescribed for PTV1 is less than

the dose prescribed for PTV2.

For target structures, we require that at least 95' of the target receives the full

prescription dose, so the dose that is received by at least 95' of each of the targets is

measured. We want to restrict the amount of the target that receives more than the

prescription dose. Because PTV2 is contained inside PTV1, PTV2 will necessarily have a

sizable, but less important, area receiving an overdose. Thus, we are only concerned with

PTV2 overdose. To evaluate the size of the hotspot, we check the percent volume of PTV2

that receives more than 110I' of the prescription dose. To evaluate the coldspots, we check









Table 3-1. Sparing criteria varies for each critical structure
Structure Percent ( .) < Dose (Gy)
brain stem 100 55
eyes 50 30
mandible 100 70
optic chiasm 100 55
optic nerves 100 50
parotid glands 50 30
skin 100 60
spinal cord 100 45
submandibular glands 50 30


the percent volume of both PTV1 and PTV2 that receives at least C, :' of the prescription

dose. The prescription doses are set to 54 Gy for PTV1 and 73.8 Gy for PTV2, which are

the dose values used at Shands Hospital at the University of Florida.

3.8.1.2 Critical structure sparing

The critical structures involved in each case vary, depending on their proximity to

the tumor. The critical structures can be classified into two general groups according

to their ability to survive radiation dose. Parallel structures, e.g., saliva glands, will

continue to function as long a certain percentage of the organ receives less than a certain

amount of dose. Serial structures, on the other hand, will cease to function if any of the

organ receives over a certain amount of dose. The spinal cord is one example of a serial

structure--if it receives too much dose, the effect is equivalent to cutting it in half, leaving

the patient paralyzed. The sparing criteria for each of the common critical structures in

head-and-neck cases are listed in Table 4-2. The critical structures involved in each case

vary, depending on their proximity to the tumor.

There are four saliva glands: one submandibular and one parotid gland on each of

the right and left sides. The saliva glands are of particular importance because their loss

can greatly decrease the patient's quality of life, but because of their location relative

to the usual tumor positions, they can be difficult to spare. Studies show that a patient

can lead a relatively normal life with three of the four glands spared. The loss of other










organs, especially the spinal cord, will also greatly affect the patient's quality of life, but

head-and-neck tumors are usually situated in such a way that other organs can be easily

spared in the FMO optimzation. Thus, the results presented place particular emphasis on

the sparing of saliva glands.

Rather than relying strictly on FMO value, a tool commonly used by physicians to

judge the quality of a treatment plan is the dose-volume histogram (DVH). This histogram

is a measure of the cumulative dose received by a given structure. It specifies the fraction

of each structure's volume that receives at least a certain amount of dose. Although there

are several critical structures to be considered in head-and-neck cases, the saliva glands are

notoriously the most difficult to spare due to their proximity to common tumor locations.

Thus, for clarity, the DVH results provided include only target structures and saliva

glands. Each of the treatment plans spares all organs not shown in the DVHs.

In the DVH results provided, vertical lines indicate target prescription doses, and

asterisks mark the sparing criteria for the saliva glands.

3.8.2 Response Surface Method Results

The response surface method was tested on six head-and-neck cases using a Windows

XP computer with a 3.2 GHz Pentium IV processor and 2 GB of RAM. The sizes of the

test cases for plans with three beams are shown in Table 3-2. Each algorithm was allowed

to run for 12 hours, which is not an unreasonable run length because BOO will not be

performed on a d n,-to-d .; basis. It is anticipated that BOO will be performed once

overnight between the time the patient is imaged and the time the patient begins radiation

therapy. A good beam vector chosen before treatment begins should continue to provide

quality treatment plans throughout the patient's treatment, which is typically 35 d .ve.

The beam orientations from which linear accelerators are capable of delivering

radiation are not restricted to integer value degrees. In this study, integral beam

orientations are desired to account for setup tolerances. For the same reasons, beam

orientations are considered on a 100 grid. To obtain integral solutions, in the subproblem









Table 3-2. Sizes of test cases.
Case # bixels # voxels
1 514 345,629
2 5463 352,284
3 613 347,233
4 549 268,823
5 423 271,156
6 585 389,565
Avg. 538 329,115
Low 423 268,823
High 613 389,565


of maximizing I(0), the integer constraint is relaxed in the problem of determining an

upper bound on s2(0), and the resulting solution is rounded to integer values.

The branching scheme used treats the rounded solution as integral and branches so

as to avoid overlapping subregions as described in Section 3.6.2.3. Results are provided

for each possible initial region scheme. The point at which branching is performed in each

region, Ob in Section 3.6.2.3, is chosen as the midpoint of the region. Also, ri and On in the

underestimating terms in Problem s2-UB in Section 3.6.2.2 are taken to be the midpoints

of their respective intervals.

It is anticipated that the weighted distance measure in equation 3-3 will have

an significant impact on the algorithm's performance. Intuitively, a small weighted

distance corresponds to a small correlation between points, which will cause the algorithm

to behave locally. In order to induce the algorithm to behave globally, the algorithm

must assume less correlation between two points. If the points are less correlated, the

algorithm will be less likely to stay in the neighborhood of previously sampled points.

The correlation between two points can be decreased by increasing the weighted distance

between the points, which can be done by increasing c or p. If c becomes sufficiently large,

the correlation between points will be effectively zero, thus yielding an effectively random

search algorithm. To test these expectations, c was tested with values of 10.0, 100.0 and










500.0. In each test, five randomly selected starting points were used to initialize the RS

algorithm.

To evaluate the algorithm's performance across all of the tested cases, the relative

improvements in FMO value over a 5-heam equispaced plan (denoted 5 equi), a 7-heam

equispaced plan (denoted 7 equi) and a locally optimal :3-heam coplanar plan obtained

using a local search algorithm called the Add/Drop local search heuristic introduced by

K~umar [80] and denoted 3 A/D are compared.

3.8.2.1 Proof of concept

To test the accuracy of the RS method, a single case was tested wherein the problem

of adding a single coplanar beam to an equi-spaced, coplanar :3-heam solution over a 10

grid was considered. The algorithm was initialized with two randomly selected starting

points. By considering such a small scale problem, the solution space in each iteration can

he explicitly enumerated in order to exactly obtain the next best point to sample. The

ability to enumerate the solution space will also allow us to determine how accurately the

RS method models the FMO objective function.

At each point that has been sampled, both the uncertainty and the expected

improvement will be zero. This result is not only theoretically true, but also intuitive

because once the FMO value at a certain point is known, there will be no improvement

over the current best FMO value by sampling that point again. It is also expected that as

the algorithm progresses, the approximation of the FMO function will become increasingly

accurate, with the approximation obtaining the exact FMO values at sampled points.

Figures :3-8A-:3-8D demonstrate how the RS method behaved as predicted at different

points in the RS algorithm. The expected value is zero at sampled points and the

approximation of the FMO function almost perfectly fits the true FMO function by

the time the algorithm terminates.

The importance of the starting points, the points sampled before the algorithm begins

to give the method some baseline information about the FMO function, was also tested.











2 points sampled
-True FMO value
-FMO appromixation











60 120 180 240 300 3(

A
80 points sampled


20 points sampled
-True FMO value
-FMO appromixation











S 60 120 180 240 300 3

B
148 points sampled algorithm terminates


F9





50





Figure 3-8.


60 120 180 240 300 360 0 60 120 180 240 300 360


C D


Proof of concept results at various stages of the RS algorithm. A) After two
points. B) After 20 points. C) After 80 points. D) After 148 points, when the
algorithm terminates.


The RS method was run with 100 randomly generated sets of starting points, and the RS

method obtained the global optimum in 90.1n' of trials, indicating that the performance of

the algorithm is not significantly dependent on the starting points.

3.8.2.2 Adding a non-coplanar beam to a coplanar solution


Next, the problem of adding a non-coplanar beam to a 3-beam locally optimal

coplanar solution was considered. The locally optimal solution is obtained using the

Add/Drop algorithm. The beam data for the non-coplanar beam being optimized is

generated on-the-fly, and consists of gantry and couch rotations, where the both gantry

and couch are allowed to rotate a full 3600 on a 100 grid. As the final solution of the

non-coplanar RS plan will be a 4-beam plan, the results from the response surface solution










are compared to the locally optimal coplanar 4-beam Add/Drop plan, denoted 4 A/D.

The plans will also be compared to an equi-spaced, coplanar 7-beam plan, denoted 7 equi,

which is commonly used in practice to treat head-and-neck cancers.

There is relatively little deviation in the final solutions between the different

parameter choices and initial regions schemes, as shown by Table 3-3. The results also

indicate that the starting points chosen do not significantly affect the outcome of the

algorithm. This implies that the response surface algorithm is robust with respect to

varying implementations.

Although the 4 RS solutions obtained an average of 5.4 !' decrease in FMO value

from the 7 equi plans, the 4 RS solutions did in fact obtain an average of 16.1"'

improvement in FMO value over the 4 AD solutions. Despite the differences in FMO

value, all treatment plans examined were similar in clinical quality, as discussed in Section

3.8.2.3.

Although the algorithm was allowed to run for 12 hours in each scenario, the

minimum FMO value obtained by the RS method was found early on. On average,

the best FMO value found was obtained in 6.15 hours after sampling 27-40 points.

For each of the RS method variations tested, both the number of points sampled

and the relative improvements in FMO value are nearly identical. This indicates that the

algorithm is robust with respect to parameter and implementation changes. The time

spent generating beam data comprises approximately b !' of the algorithm's run time,

while the response surface portion on average accounts for only 1;:' Thus, it is expected

that changes to the RS method, including improvements to the branch-and-bound routine,

will not have a very strong impact on the number of points the algorithm will sample in

its allotted run time.

3.8.2.3 Clinical results

The target coverage achieved by the different treatment plans are di;1l-p Ixd in Table

3-4. On average, the 7 equi plan was able to deliver the most amount of dose to PTV2,









Table 3-3. Minimum FMO value obtained and time required to obtain it.
M~in. FM~O value Time (hrs)
Case Avg. St. Dev. Avg. St. Dev.
1 565.24 8.82 5.35 5.07
2 570.51 12.83 7.49 3.78
3 927.34 20.6;0 7.05 2.21
4 710.92 7.72 6.54 3.39
5 512.22 20.04 6.96 3.33
6; 799.95 34.07 3.48 3.40

Table 3-4. Target coverage achieved by the treatment plans
4 RS 4 A/D 7 equi
PTV2 dose at 95' volume 73.16 Gy 72.56 Gy 73.81 Gy
PTV2 .receiving > 110I of Rx 23.18 15.07 31.63
PTV2 .receiving > C, : of Rx 98.87 98.67 99.57
PTV1 dose at 95' volume 54.71 Gy 54.41 Gy 55.09 Gy
PTV1 receiving > C, : of Rx 97.95 98.01 97.46


but the 4 RS plan is very close. Both of the 4-beam plans obtain smaller hotspots and

better PTV1 target coverage than the 7 equi plan. The 4 A/D plan on average underdoses

PTV2, which could lead to recurrence of the cancer. This underdosage could also account

for the smaller hotspot in the 4 A/D plans.

Figures 3-10 and ?? illustrate two representative cases where the 4 RS, 4 A/D and

7 equi plans each have clinically acceptable target coverage. The vertical line at 73.8 Gy

indicates the prescription dose for PTV2.

The ability of each of the treatment plans to spare the organs in the cases tested is

shown in Table 3-5. Surprisingly, both the 4 RS and the 4 A/D plan are equivalent to

or outperform the 7 equi plan in terms of organ sparing. In the 4-beam plans, the left

submandibular gland is spared in 8 :' of the treatment plans developed, whereas it is only

spared in Iul' of the 7 equi plans. One case illustrating equivalent organ sparing is shown

in Figure 3-10, and one case demonstrating improved organ sparing over the 7 equi plan is

shown in Figure ??. Just as PTV2 underdosage in the 4 A/D plans likely contributed to

the smaller hotspots, it is possible that the improved organ sparing in the 4 A/D plans is

also a result of the underdosage.















Table 3-5. Percentage of plans in which an organ is spared.


Structure
brain stem
mandible
left optic nerve
right optic nerve
left eye
right eye
optic chiasm
left parotid gland
right parotid gland
left SMB gland
right SMB gland
spinal cord
skin


4 RS
101 1' .
101 1' .
101 1' .
101 1' .
101 1' .
101 1' .
101 1' .
101 1' .





101 1' .
101 1' .


4 A/D
101II' .
101II' .
1011 .
1011 .
101II' .
101II' .
101II' .
101II' .


ill' .
8CI :' .
50I .


7 equi
101 1'
101 1' .
101 1' .
101 1' .
101 1' .
101 1' .
101 1' .
101 1' .





101 1' .
101 1' .


case001 Target DVHs


case001 Saliva gland DVHs

100 7-beam equi
4-beam A/D
S80~ -4-beam RS

S60 r. parotid
b -r. SMB
E 402




Dose [Gy]


7-beam equi
-4-beam A/D
-4-beam RS


....

~60

E 40

> 20


PTV
--GTV


00 10 20 30 40 50 60 70 80 90
Dose [Gy]


7-beam equi-spaced (dotted), 4-beam Add/Drop (dashed) and 4-beam RS
non-coplanar (solid) target and select saliva gland DVHs. A) Target coverage
is nearly identical. B) The tumor surrounds the right submandibular gland, so
the FMO solver recognizes that it cannot be spared and allows it to receive as
much dose as necessary to ensure good target coverage in all plans. All other
saliva glands are spared in all plans.


Figure 3-9.












case05 TrgetDV~scase005 Saliva gland DVHs
10C 7beamequi100 7-beam equi
----4-beam A/D -4-beam A/D
80 -o4-beam RS 80~ -4-beam RS
60PTV 60~ 1 SMB
a. ~-GTV
E 40 E 40
> 20~ > 20

0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
Dose [Gy] Dose [Gy]

A B

Figure :3-10. 7-heant equi-spaced (dotted), 4-heant Add/Drop (dashed) and 4-heant RS
non-coplanar (solid) target and select saliva gland DVHs. A) Target coverage
is nearly identical. B) The left submandibular gland is spared by the two
4-heant plans, but not by the 7-heant plan. All other saliva glands are spared
in all plans.


3.8.3 Neighborhood Search Method Results

The simulated annealing method was tested on six head-and-neck cases using a

Windows XP computer with a 2.1:3 GHz Pentium Al processor and 2 GB of R AM. On

average, a :340 FAIOs were calculated in :30-nxinute run time allowed for the simulated

annealing and Add/Drop algorithms. Beams were selected on a 5-degree grid, yielding 72

candidate coplanar beams.

The simulated annealing and Add/Drop algorithms were used to obtain 4-heant

coplanar plans using regular and flip neighborhoods. In order to compare the quality of

the treatment across different plans, the plans are compared in terms of the percentage

intprovenients of each plan's FMO value intprovenient over the FMO value of the locally

optimal :3-heant plan obtained front the Add/Drop local search heuristic described

by K~umar [80]. The Add/Drop plans are denoted 3 A/D and 4 A/D for the :3- and

4-heant plans, respectively. The 4-heant plans generated by the simulated annealing and

Add/Drop algorithms are compared to the typical 5- and 7-heant equi-spaced plans,









denoted 5 equi and 7 equi, respectively. The simulated annealing plans are denoted by the

implementation numbers, which refer to the parameters used, given in Table 3-6.

Figure 3-11 demonstrates the improved convergence times possible using the flip

neighborhood.

3.8.3.1 Add/Drop algorithm results

The Add/Drop algorithm was allowed to run for 30 minutes to generate a 4-beam

plan. The &h(0) neighborhood with 6 = 20 and the Na'F(0) with 6F = 0 and 6F = 20

neighborhoods are tested for the Add/Drop algorithm. The value 6 = 20 is chosen

to approximate the neighborhood size that is expected from the simulated annealing

implementation using a large flip neighborhood, where 6F = 180. More details on the

simulated annealingf implementations are provided in Section 3.8.3.2.

Using &h(0), the 4-beam Add/Drop solution is nearly identical identical to the

7-beam equi-spaced plan, while the flip neighborhoods allow the Add/Drop algorithm

to find 4-beam solutions that exceed the quality of the 7-beam plans. Figure 3-12

demonstrates the quality of the solutions, while Figure 3-11(a) illustrates that the flip

neighborhoods provide faster FMO convergence than that of Mi~(0).

3.8.3.2 Simulated Annealing results

Several parameter sets were tested for the simulated annealingf algorithm. For

simplicity, each of the parameter sets and methods of generating a neighboring solution

are numbered according to Table 3-6. Each implementation contains a total of 500

iterations, i.e., 500 sampled points, thus yielding a fair comparison between the parameters.

To ensure clinical practicality, the algorithm was allowed to run for a maximum of 30

minutes or 500 iterations, whichever came first.

For the cooling schedule, we update the temperature according to an exponential

cooling schedule, Ti 1 = a~Ti, where a~ < 1. Due to the random nature of the algorithm,

the algorithm is restarted five times, each time with a different initial starting point. The

first initial starting point is an equi-spaced solution, and each subsequent starting point is











4-beam Add/Drop 4-beam simulated annealing
750- 700-
--Regular neighborhood I -Regular neigborho
---flip, 6F=0 ---flip, 6F=0
flip, 6F=20 flip, 6F=180
650
700





600 -0
LL~ ~ ~ 500t L i



A B

Fiue 3-1 Coprsno M ovegne )AdDo.B)Smltdanaig

th reiu inta soltio roae yddgeshrecniaengsarcoidedn










no-ipovn sltion fo teinitial iteaton of te agorthm




enigre solutiCon paceisono consider a aneighorhod As shrown in Figuraed 3-A, teai


t probbiityof seleting aouo betame 20 wy usingres ther M(0)neighbohoo withe gre omietric

Sddist e ribuio wtha ps = 025isony 0.3' an' s on aso grid. Whe cnsider thi siufficed ntl sallto


nth constider neighborhoodslarer thane r =o 20 for() and To = 20. fo =Or ((0) inth





Add/ro alorih.ut si the Add/Dro imlmettin the( 8 neighborhoods b ((0) susds ta h







with 6" = 0 is also considered.










Table :3-6. Definitions of implementations.
Number n m NV a To
1 100 1 1 0.9 0
2 10 10 1 0.9 0
:3 100 1 1 0.99 0
4 10 10 1 0.99 0
5 100 1 1 0.9 75
6 10 10 1 0.9 75
7 100 1 1 0.99 75
8 10 10 1 0.99 75
9 100 1 all 0.9 0
10 10 10 all 0.9 0
11 100 1 all 0.99 0
12 10 10 all 0.99 0
1:3 100 1 all 0.9 75
14 10 10 all 0.9 75
15 100 1 all 0.99 75
16 10 10 all 0.99 75


Figure :3-11(b) shows that the flip neighborhoods converge in FMO value significantly

faster than does the Nix(0) neighborhood, while Figure :3-13 shows that the flip neighborhoods

provide comparable solution quality to both the non-flip simulated annealing and 7-heam

equi-spaced solutions.

3.8.3.3 Clinical results

Because there is no fundamental way of quantifying a treatment plan, a tool

commonly used by physicians to judge the quality of a treatment plan is the dose-volume

histogram (DVH). A DVH is a graphical measure of the cumulative dose received by a

given structure. It specifies the percentage of each structure's volume that receives at least

a certain amount of dose, thus providing an intuitive means of assessing the quality of a

treatment plan.

The plans tested plans each contain two target structures. The gross tumor volume

(GTV) is the tumor mass observed from imaging scans. The clinical tumor volume (CTV)

is the GTV plus some margin specified by the physician. The CTV is used by physicians










in case there are elements of the tumor mass that cannot be seen from the imaging scans,

and the dose prescribed for the CTV is less than the dose prescribed by the GTV.

DVHs for a representative case comparing the 7-beam equi-spaced plan with the

simulated annealing plans obtained using a regular neighborhood and flip neighborhoods

with 6" = 0 and 6F = 180 are shown in Figure 3-13. Comparison of the 7-beam

equi-spaced plan and the Add/Drop plans using a regular neighborhood and flip

neighborhoods with 6F = 0 and 6F = 20 are shown in Figure 3-12. The sparing criteria

used for the saliva glands, no more than 501' of the gland receiving 30 Gy, is marked by

the star in Figures 3-13 and 3-12. The prescription dose for the GTV is 73.8 Gy, which

is marked by the vertical line in Figures 3-13 and 3-12. As previously stated, for target

structures, we require that at least 95' of the target receives the full prescription dose.

Figure 3-13 reveals that the 7-beam equi-spaced plan actually overdoses the target

and has a larger hotspot than the 4-beam simulated annealing plans. The 7-beam

equi-spaced plan only spares three of the four saliva glands, whereas the 4-beam simulated

annealingf plans spare three or more saliva glands. The simulated annealingf plans obtained

using the flip neighborhoods spare all four saliva glands, while the plan obtained how the

Na(0) neighborhood only spares three saliva glands, indicating that the flip neighborhoods

do in fact find superior solutions in terms of clinical quality.

Figure 3-12 shows that the 4-beam Add/Drop plans obtain nearly identical solutions

when compared to the 7-beam equi-spaced DVHs. The flip neighborhoods perform

clinically comparably to the regular neighborhood plans, and all of the Add/Drop plans

are comparable to the 7-beam equi-spaced plan in terms of saliva gland sparing and target

coverage.

3.9 Conclusions and Future Directions

3.9.1 Response Surface Conclusions

We have shown that for head-and-neck cases, quality plans with fewer beams than

a standard treatment plan can be obtained if BOO is applied. The response surface















Add/Drop: Target DVHs
-?-beam equi-spaced
-flip, FF=20
-flip, FF=0
i i Regular neighborhood


Add/Drop: Saliva gland DVHs
-?-beam equi-spaced
--flip, FF=20
--flip, FF=0
Regular neighborhood


o u-


E


80

E 60

E
2 40


0 10 20 30 40 50 60 70 80 90
Dose [Gy]


A B


Figure :3-12. Comparison of Add/Drop and 7-heam equi-spaced plans. A) The Add/Drop
plans achieve nearly identical target coverage when compared to the 7-heam
equi-spaced plan. B) The saliva gland sparing in the Add/Drop plans and the
7-heam equi-spaced plan is clinically equivalent.


Simulated Annealing: Target DVHs


Simulated Annealing: Saliva gland DVHs
-?-beam equi-spaced
--flip, FF=180
--flip, FF=0
Regular neighborhood


7-beam equi-spaced
flip, FF=180
flip, FF=0
Regular neighbohod


100 T


E
3 0


E 60
IL.
E
2 40

20


0O 10 20 30 40 50 60 70 80 90 0O 10 20 30 40 50 60 70 80 90
Dose [Gy] Dose [Gy]

A B


Figure :3-1:3. Comparison of Add/Drop and 7-heam equi-spaced plans. A) Unlike the
7-heam equi-spaced plan, the 4-heam simulated annealing plans do not
overdose the target. B) The simulated annealing plans are also capable of
sparing more saliva glands than the 7-heam equi-spaced plan.










algorithm operates in a clinically reasonable time frame, and is generally successful in

selecting non-coplanar beam orientations to improve the FMO value over that of locally

optimal coplanar solutions. The FMO value of the 4-heam response surface plans was

also only slightly larger than that of the 7-heam equi-spaced coplanar treatment plans,

indicating comparable treatment plans despite the decrease in the number of beams used.

In terms of clinical results, the most significant benefit of the non-coplanar solutions

over the locally optimal coplanar solutions was the ability to deliver a higher amount

of dose to the target structures. Both the non-coplanar and locally optimal coplanar

solutions were able to obtain treatment plans with organ sparing that is comparable to or

improved upon the 7-heam equi-spaced coplanar treatment plans.

While the inclusion of non-coplanar orientations in BOO is useful in terms of FMO

value and target coverage, the resulting improvements in the treatment plan may not

ahr-l-~ .-4e clinically significant. With better parameter tuning or neighborhood structure,

it is possible that the Add/Drop algorithm can obtain coplanar treatment plans with more

desirable target coverage, thus making the response surface plans and the Add/Drop plans

clinically equivalent. This -II__- -; that the inclusion of non-coplanar beam orientations

does not significantly improve the quality of a treatment plan. Although most BOO

research is restricted to coplanar orientations, there has not yet been a study assessing the

solution quality of coplanar versus non-coplanar solutions. With this study as evidence,

both researchers and practioners now have a basis for restricting the solution space to the

smaller, more tractable set of coplanar beams for head-and-neck heam optimization.

The patient cases in this work were all head-and-neck cases. Different tumor sites,

e.g., breast, lung and prostate, could also benefit from BOO, and perhaps may experience

greater improvements in treatment plan quality. In future work, these sites will be tested

to assess the general clinical usefulness of non-coplanar orientations and the response

surface method.










3.9.2 Neighborhood Search Conclusions

We have shown that for head-and-neck cases, quality plans with fewer beams than

a standard treatment plan can he obtained if BOO is applied. The simulated annealing

and Add/Drop algorithms both regularly obtained quality treatment plans with as few

as four beams in only 30 minutes. The use of the flip neighborhood improves the rate of

Fl\1 convergence in both algorithms, and even has the ability to improve organ sparing

as shown in the simulated annealing results. The simulated annealing and Add/Drop

algorithms performed comparably to each other, with neither algorithm indicating a

significant benefit over the other.

It is possible to incorporate flip neighborhoods into other BOO algorithms that rely

on neighborhood searches to yield improved treatment plans in clinically acceptable time

frames.










CHAPTER 4
FR AC TION ATION

4.1 Introduction

Typically, head-and-neck treatment plans each contain two target structures, or

planning tumor volumes(PTV): PTV1 and PTV2. Let PTV1 he the tumor mass observed

from imaging scans, and let PTV2 he PTV1 plus some margin specified by the physician.

Rather than deliver an entire treatment plan in one session, a treatment plan is

divided into several sessions, called fractions. This is done to take advantage of the fact

that normal, healthy cells recover faster from the radiation than cancerous cells. To

obtain the treatment plans for the fractions, in practice, a single FMO treatment plan is

developed and then divided into the desired number of fractions, usually around :35. This

division of a treatment plan is a non-trivial task, as the target voxels must receive 1.8-2.0

Gy of radiation in each fraction.

With a single IMRT treatment plan, it is practically impossible to devise a constant

dose-per-fraction delivery technique because only a single FMO problem is solved to

obtain the treatment plan, which is then simply divided into a number of daily fractions.

If a single plan is optimized to deliver doses to multiple target-dose levels, then the dose

per fraction delivered to each target must change in the ratio of a given dose level to the

maximum dose level. For example, my- PTV1 has a prescription dose of 70 Gy, PTV2 has

a prescription dose of 50 Gy, and the number of fractions is :35. If a single treatment plan

is divided among the :35 fractions, then PTV1 will receive 70/:35 = 2.0 Gy in each fraction,

but PTV2 will only receive 50/:35 = 1.4 Gy, and thus any cancerous cells in PTV2 may

not he eradicated by the treatment. Similarly, if only 25 fractions are used in order to

ensure that PTV2 receives 2.0 Gy per fraction, then PTV1 receives 70/25 = 2.8 Gy per

fraction, well above the desired dose.

We propose a new method of approaching the fractionation subproblem wherein an

FMO treatment plan is developed for each target structure, rather than developing a










single treatment plan for all target structures. The individual treatment plans can then be

easily divided into optimal fractions.

The primal-dual interior point algorithm presented by Aleman et al. [88] is used to

solve the FMO and fractionation models to optimality.

4.2 Model Formulation

The fractionation model builds on the FMO model described in Chapter 2. To solve

the fractionation problem, we consider developing an individual fluence map solution for

each target. For a case with two targets, two plans must be developed: (1) a plan that

delivers the prescription dose to PTV1 and PTV2, and (2) a plan that "box.-l- the dose

received by PTV1 to reach the prescribed dose level. These two fluence maps can then be

divided into the appropriate number of fractions easily. For the example of 50 Gy and 70

Gy prescription doses for PTV2 and PTV1, respectively, this would yield 25 fractions of

treating both PTV1 and PTV2 to 50/25 = 2.0 Gy, and another 10 treatments of treating

just PTV1 to (70 50)/10 = 2.0 Gy.

For simplicity, we call these individual fluence maps Ilin mI I s~~", rather than using the

term to describe the daily treatments. The development of these fluence maps separately

would result in suboptimal solutions. To optimize these fluence map sets simultaneously,

we consider each bixel in each fraction as an individual decision variable. As there number

of fractions is equal to the number of targets (T), this results in a fluence map developed

for each target.

In the single FMO formulation, dose to voxel j in structure s is defined as zys

C~yx, aizi = 1,..., T, and the penalty associated with it, as Fs,(zys). Because the

fractionation model will be concerned with dose-per-fraction as well as cumulative dose,

new variables must be defined to express these values.










Define x{, f =,~ 1...,T, as the fluence ofbheamlet, i in fraction f. The amount of dose

received by a voxel j in structure s in fraction f is defined as


zf= Dexf, j=1..vs=1..Tf=1,.T (4-1)
i= 1

Critical structures are thought to be affected by only the cumulative dose received

from all treatments, rather than the just the dose in any one particular fraction. This

cumulative dose received by a voxel is


zjs= Dexf,i j= 1,..., v, s= T+1,..., S (4-2)
f =1 i= 1

Critical structures are penalized in the same manner as in the original FMO model, that

is, F,(zys), = T +1, ..., S.

Targets require a more complex treatment in the fractionation model. In each

fraction, we are primarily concerned with dose received by the targets in that particular

fraction. Thus, new variables are needed to express the amount of dose per fraction

received by a voxel (zf, in Equa~~tion (4-1)).

Since we must also ensure that the cumulative dose received by each target reaches

the prescribed dose, variables to express the cumulative dose received by a voxel are

required. Intuitively, this cumulative dose should be the sum of all the doses received in all

fractions. If the cumulative dose for targets is defined this way, then over/underdosing

in one fraction can result in under/overdosing in another to compensate, which is

undesirable. To prevent such a scenario, another new variable called the artificial dose

is required (zy, in Equation (4-3)). Rather than simply summing up the dose received

in each fraction, we will assume that in the previous fraction, the target voxel received

exactly the correct prescription dose for the previous fraction. Thus, no compensating will

be necessary. The artificial dose is just the prescription dose from the previous fraction










(Pf_l) plus the dose received in the current fraction:


Ojs = Py-1 +z~f, j = 1,..., v, s = 1,..., Tf = 1,..., T (4-3)


Since each of the target voxels being irradiated in fraction f is treated as target f, the

penalty functions for these voxels is


T~jv


Once a target has received its prescription dose, ideally, it should not receive any

further dose. As target f is treated in fraction f, for all fractions after f, target f should

be treated as normal tissue. Specifically, targets that no longer require dose will be treated

as skin, denoted structure S. Therefore, these target voxels, along with actual skin voxels,

will be penalized with penalty function Fs. The dose received by these target voxels is

the prescription dose of the voxel (P,) plus the dose received in all subsequent fractions

(E, z ) This leadsn to the followningr penaltyi functions for voxerls penalized as normal

tissue in fraction f:

Fsr Ps +: zf j + Fs (zys)
s=1 yEV, =s+1 ev
As with the traditional FMO model, penalty functions are normalized according to

the number of voxels in the structure. For critical structures, this normalization factor

is still 1/v, since there are ahr-l-w v, voxels being treated as critical structure s. In each

fraction, the number of target voxels depends on which targets still need to be treated.

Each fluence map set will only "see" the target voxels that are included in its prescription

dose level. Thus, define the number of target voxels treated in fluence map f as


vyf = vs f =1,...,T
S=f










The number of voxels treated as skin in each iteration can be expressed by or vf + vs,

where or 07 is the number of target voxels being treated as skin and vs is the number of

actual skin/unspecified tissue voxels.

Identical to the traditional FMO, the critical structures are normalized and penalized

by
S-1

s=T+1 "jEV

Letz b a ectr o al zy, zf, and if, variables. The objective function is obtaiined

by summing the normalized penalty functions:


-firac (z) =: 17- U +s FUs Ps +C z + F(zys)]
f=1 s=1 j6%, =s+1 j6vs
Tj~ CS-1 C~~

s= f j6V s=T+1 jEV,

The fractionation model is then formulated as


minimize FTfrac(z)

subject to zf= D
i= 1



f =1 i= 1

(js = P,_, +z~f,

x>0


j= 1 ,.Vs,s


1,... T, f


1,...,T


j=l 1 ,.,v,s


1,... T, f= 1,..., T


As the objective function is the sum of quadratic functions and the constraints are all

linear, the fractionation formulation, just like the basic FMO formulation.

4.3 Results

The fractionation model is tested using the primal-dual interior point algorithm

in Aleman et al. [88]. One significant benefit of employing a primal-dual interior point

algorithm is that the solution generated is guaranteed to be optimal to within a certain










tolerance that can be specified by the user. Thirteen head-and-neck cases using five

equi-spaced beams are tested. Each test case consists of two targets, PTV1 and PTV2,

with prescription dose levels of 70 Gy and 50 Gy, respectively.

According the -II_0----- -0.1. made on algorithm parameters in Aleman et al. [88],

the primal-dual interior point algorithm was implemented with a Single Approximation

Hessian and a stopping criteria of a relative duality gap of 0.1 Although it was also

recommend to remove il-aisls 11beamlets, these removal of these beamlets actually

increases run time in the fractionation model. Thus, insignificant beamlets are left in the

fractionation model.

4.3.1 Computational Results

The tests are run in Matlab (il II11Works, Inc.) on a 2.33GHz Intel Core 2 Duo

processor with 2GB of RAM. Table 4-1 shows the sizes of each case in terms of the

number of decision variables (the number of bixels) and the size of the patient area being

treated (the number of voxels). The computation times obtained are display in Table 4-1.

On average, the fractionation model was solved in 22.03 seconds. With the same algorithm

parameters and weighting parameters, a single FMO treatment plan can be determined

in an average of 16.28 seconds, thus there is only a 35' increase in computation time

required to develop two FMO plans for the fractionation model. This relatively small

increase in time could be accounted for by the fact that the weighting parameters used

in the objective function were specifically tuned for the fractionation model. Using

parameters specifically tuned to the single-FMO model, the single-FMO model can be

solved on average in 9.36 seconds. Compared to this average run time, the FMO model

requires 2.4 times as much computation time to develop two models as opposed to one,

which is a more intuitive expectation of the interior point method's performance.









Table 4-1. Case sizes and run times using identical algorithm and weighting parameters.
Single FM~O Fractionation
Case Bixels Voxels Iterations Time (s) Iterations Time (s)
1 813 85,017 16 8.39 16 19.60
2 1320 189,234 103 82.69 14 55.34
3 935 86,255 24 11.75 11 18.79
4 692 58,636 15 6.87 11 11.47
5 1044 102,262 14 13.16 12 29.70
6 1005 84,369 13 10.31 12 25.58
7 822 71,873 17 9.14 14 18.88
8 802 92,307 59 22.92 14 20.19
9 911 65,541 18 10.84 17 26.12
10 642 66,634 25 7.94 16 12.44
11 279 56,847 29 2.75 14 2.99
12 994 96,105 17 12.30 12 27.13
13 823 72,729 33 12.55 14 18.15
Average 852 86,755 29 16.28 14 22.03


4.3.2 Clinical Results

Because there is no fundamental way of quantifying a treatment plan, DVHs are

examined in addition to objective function values to assess the quality of a treatment

plan..

The prescription doses used are 70 Gy for PTV1 and 50 Gy for PTV2. These are

common prescriptions used in the cancer center at Shands Hospital at the University of

Florida. Figures 4-1-4-7 show both dose volume histograms (DVHs) and axial slices for

several cases. The DVHs show that in the first fraction, both PTV1 and PTV2 are treated

to 50 Gy, and in the second fraction, only PTV1 is treated to an additional 20 Gy. The

prescription dose for the fraction is marked by a vertical line. The amount of dose received

by each target in each fraction is clinically acceptable.

As this study focuses on head-and-neck cases where the most conflict lies in treating

the targets while sparing the saliva glands, only DVHs of the saliva glands are shown. All

other organs, including skin/unspecified tissue, receive a low enough amount of dose to be

spared in the treatment. The sparing criteria for each of the common critical structures in
head-and-neck cases are listed in Table 4-2. The critical structures involved in each case









Table 4-2. Sparing criteria varies for each critical structure
Structure Percent ( .) < Dose (Gy)
brain stem 100 55
eyes 50 30
mandible 100 70
optic chiasm 100 55
optic nerves 100 50
parotid glands 50 30
skin 100 60
spinal cord 100 45
submandibular glands 50 30


vary, depending on their proximity to the tumor, and thus DVHs for some cases do not

include all saliva glands.

DVHs of the saliva gland doses in Fr-action 1 show that the saliva glands receive the

1 in & Rly of dose in the first fraction. Because the cumulative amount of dose received

determines whether or not critical structures can be spared, the DVHs for Fraction 2

depict the cumulative dose of these organs. The sparing criteria used for saliva glands is

that no more than 501' of the gland can receive more than 30 Gy. This point is marked as

a star. For most cases, all of the saliva glands are spared.

Figures 4-1-4-7 also show the dose received in each fraction as a colorwash of a slice

of the patient. Fraction 1 delivers a homogeneous dose of 50 Gy to both PTV1 and PTV2

while generally avoiding overdosing any of the marked critical structures. In Fraction 2,

the dose to PTV1 is boosted by 20 Gy without delivering any unnecessary dose.

4.3.3 Spatial Coefficient Results

The concept of employing spatial information as described in Section 2.4 is also

applied to the fractionation model. One set of spatial coefficients is used to obtain both

fractions. For the fractionation treatment plans, the spatial coefficients are A = 1.02,

p- = -0.92, /7 = 0.97 and the minimum coefficient for target voxels is 0.6.

Generally, the DVHs for both targets and critical structures using spatial coefficients

are similar to those obtained without using spatial coefficients. In fact, in the cases tested,












Target DVHs: Fraction 1 of 2

-PTV1 U PTV2


Target DVHs: Fraction 2 of 2
-PTV1


0

10

0O


0O


10

8
6

E4







LL.

E 4

2


0 10 20 30 40 50 60 70 80 90
Dose [Gy]

Target DVHs: Cumulative dose

0 ------ PTV2
---PTV1











00 10 20 30 40 50 60 70 80 90
Dose [Gy]
Fraction 1 of 2


1020 3040 5060 7080 90
Dose [Gy]
Saliva gland DVHs: Cumulative dose

-left parotid gland
1---leftR submandibular gland
right parotid gland
ru ngh~t submandibular gland




1,



10 2030 4050 0 7080 90
Dose [Gy]
Fraction 2 of 2


100




S60
LL.

E 40

2 0


00


Figure 4-1. Target DVHs, saliva DVHs and axial slices in Fr-actions 1 (left) and 2 (right).











Target DVHs: Fraction 1 of 2


Target DVHs: Fraction 2 of 2
PTV1


--PTV1 U PTV2


100 --


v0102030405060708090
Dose [Gy]

Target DVHs: Cumulative dose

---PTV1


102030405060708090
Dose [Gy]
Saliva gland DVHs: Cumulative dose

-left parotid gland
--- left submandibular gland
:: --- right parotid gland
--- right submandibular gland








102030405060708090
Dose [Gy]
Fraction 2 of 2


100






E 4


102030405060708090
Dose [Gy]
Fraction 1 of 2


-b -4 -Z U Z 4 b


-b -4 U Z 4 b


Figure 4-2. Target DVHs, saliva DVHs and axial slices in Fr-actions 1 (left) and 2 (right).
















































































axial slices in Fractions 1 (left) and 2 (right).


100

S80

S60
LL.

E 40

20


I


Target DVHs: Fraction 1 of 2

.4 -PTV1 U PTV2


Target DVHs: Fraction 2 of 2

-PTV1


u0)102030405060708090
Dose [Gy]

Target DVHs: Cumulative dose

0 -PTV1
--PTV2











00 10 20 30 40 50 60 70 80 90
Dose [Gy]
Fraction 1 of 2


102030405060708090
Dose [Gy]

Saliva gland DVHs: Cumulative dose

left carotid gland


dlbularr gland
g~land
ndlitilar gland


100

c- 8


O 4i




00


---left submanj
rIght palroud










1020 3040 5060 7080 90
Dose [Gy]
Fraction 2 of 2


Figure 4-3. Target DVHs, saliva DVHs and











Target DVHs: Fraction 1 of 2


Target DVHs: Fraction 2 of 2
PTV1


--PTV1 U PTV2


100


v0102030405060708090
Dose [Gy]
Target DVHs: Cumulative dose
)0 ~~~ --- PTV2
I---PTV1


102030405060708090
Dose [Gy]
Saliva gland DVHs: Cumulative dose
left parotid gland
~- ---left submandibular gland
--- right parotid gland
\I --- right submandibular gland


80


E


40 Ei


23

0 10 20 3040 5060 7080 90
Dose [Gy]
Fraction 2 of 2


102030405060708090
Dose [Gy]
Fraction 1 of 2


Figure 4-4. Target DVHs, saliva DVHs and axial slices in Fr-actions 1 (left) and 2 (right).












Target DVHs: Fraction 1 of 2
-PTV1 U PTV2


Target DVHs: Fraction 2 of 2
PTV1


13 10 2030 4050 6070 80 90
Dose [Gy]

Target DVHs: Cumulative dose

)0- --- PTV2 -PV


102030405060708090
Dose [Gy]
Saliva gland DVHs: Cumulative dose

left parotid gland
S---left s11oniand10ular gland
rlght pajroll gland
Srlgh~t 5u0onindltilar gland








1020 3040 5060 7080 90
Dose [Gy]
Fraction 2 of 2


Fraction 1 of 2


-5 0 5


Figure 4-5. Target DVHs, saliva DVHs and


-5 0 5

axial slices in Fr-actions 1 (left) and 2 (right).

















100

r 80

S60
LL.

E 40

20


0 102030405060708090
Dose [Gy]

Target DVHs: Cumulative dose

1- T 2---PTV1|


Target DVHs: Fraction 1 of 2

PTV1 U PTV2


Target DVHs: Fraction 2 of 2

-PTV1


102030405060708090
Dose [Gy]

Saliva gland DVHs: Cumulative dose

--left protid gland
---left s11oniand10ular gland
rlght palroud gland
,4 rlght 5u0onindltilar gland









1020 3040 5060 7080 90
Dose [Gy]

Fraction 2 of 2


100

m
r Eil
o

~ Eil
LL
a,
E 4il
o
2il


10 2030 4050 6070 80 90
Dose [Gy]

Fraction 1 of 2


4

2

0

-2

-4

-6

-8

-10

-12

-5 0 5


Figure 4-6. Target DVHs, saliva DVHs and


-5 0 5


axial slices in Fr-actions 1 (left) and 2 (right).









































































-6 -4 -2 0 2 4


axial slices in Fr-actions 1 (left) and 2 (right).


Target DVHs: Fraction 1 of 2

a .-PTV1 U PTV2


Target DVHs: Fraction 2 of 2
-PTV1


0

10

0O


0O


0 10 20 30 40 50 60 70 80 90
Dose [Gy]

Target DVHs: Cumulative dose

0 ---- -PTV2
---PTV1


10

8
6

E4







LL.

E 4


1020 3040 5060 7080 90
Dose [Gy]
Saliva gland DVHs: Cumulative dose

--left parotid gland
\ ---right parotid gland


102030405060708090
Dose [Gy]
Fraction 1 of 2


102030405060708090
Dose [Gy]
Fraction 2 of 2


-6 -4 -2 0 2 4


Figure 4-7. Target DVHs, saliva DVHs and










there were no instances of either the spatial treatment plans or the non-spatial treatment

plans yielding clinically significant changes in the DVHs. The slices show that there is

improved homogeneity in the target doses when spatial coefficients are used.

The slices also indicate that since the use of spatial coefficients results in the target

voxels weighing more heavily than other voxels, the model is more willing to deliver dose

to critical structures rather than overdose or underdose the target. This helps provide a

uniform dose in the target, and should still be acceptable as the cumulative dose for all

critical structures remains within acceptable levels and there are no instances of sacrificing

organs that were not already sacrificed in the non-spatial plan.

Because more critical structure voxels receive dose in the spatial plans, the dose

deposited in the target structures is more spread out, and thus the nmaxiniun dose

received hv the critical structure voxels is less than in the non-spatial plans. This of course

means that more voxels are exposed to radiation, but the levels are lower and the amount

of radiation still falls within clinically acceptable limits. The resulting intprovenient in

homogeneity is evident for each of the cases, but the effect of the more spread out dose is

best illustrated in the second fraction of each case.

Figures 4-8-4-14 show the DVHs and slices for some of the tested cases. In particular,

Figures 4-9, 4-10, 4-11 and 4-14 demonstrate that the spatial coefficients reduce the

amount of dose delivered outside of the targets when compared to their respective

non-spatial plans in Figures 4-2, 4-3, 4-4 and 4-7.

4.4 Conclusions and Future Directions

The fractionation model presented allows for the creation of guaranteed optimal

fluence maps for each fraction of a patient's treatment. These fluence maps can he easily

divided into the appropriate number of fractions without sacrificing optinmality. Using the

primal dual interior point method, the fractionation model obtains fluence maps for each

target in a clinically feasible amount of time. As expected, the computation time required

to generate two fluence maps for a two-target case is more than the time necessary to










Target DVHs: Fraction 1 of 2


Target DVHs: Fraction 2 of 2
-PTV1


--PTV1 U PTV2


100


102030405060708090
Dose [Gy]
Target DVHs: Cumulative dose


102030405060708090
Dose [Gy]
Saliva gland DVHs: Cumulative dose


- left parotid gland
--- left submandibular gland
Right parotid gland
- right submandibular gland


100

80 I

60 1~
I*


40 \
20

0j 1020304050607;08090
Dose [Gy]
Fraction 2 of 2


Dose [Gy]


Figure 4-8. Target DVHs, saliva DVHs and axial slices in Fr-actions 1 (left) and 2 (right)
using spatial coefficients.


112











Target DVHs: Fraction 1 of 2
-PTV1 U PTV2


Target DVHs: Fraction 2 of 2
-PTV1


0 10 2030 4050 6070 80 90
Dose [Gy]

Target DVHs: Cumulative dose

)0 -~- ----- PT2
---PTV1





0




00 10 2030 4050 6070 80 90
Dose [Gy]
Fraction 1 of 2


0 10 20 3040 5060 7080 90
Dose [Gy]
Saliva gland DVHs: Cumulative dose

100 left parotid gland
---left s11oniand10ular gland
rlgh~t palroud gland
Srlgh~t 5u0onindloular gland


60 1


23


j0 102030405060708090
Dose [Gy]
Fraction 2 of 2


Figure 4-9. Target DVHs, saliva DVHs and axial slices in Fr-actions 1 (left) and 2 (right)
using spatial coefficients.


113










Target DVHs: Fraction 1 of 2
-PTV1 U PTV2


Target DVHs: Fraction 2 of 2
-PTV1


13 10 2030 4050 6070 80 90
Dose [Gy]
Target DVHs: Cumulative dose
)0 -PTV1
10 j---PTV2





0



00 10 20 30 40 50 60 70 80 90
Dose [Gy]
Fraction 1 of 2


102030405060708090
Dose [Gy]
Saliva gland DVHs: Cumulative dose
left parotid gland
\---leftR submandibular gland
: 1igh7t parotid gland
n rght submandibular gland







1020 3040 5060 7080 90
Dose [Gy]
Fraction 2 of 2


Figure 4-10. Target DVHs, saliva DVHs and axial slices in Fractions 1 (left) and 2 (right)
using spatial coefficients.


114

























u0 102030405060708090
Dose [Gy]
Target DVHs: Cumulative dose

0 -----PTV2
O ---PTV1









00 10 2030 4050 60 7080 90
Dose [Gy]
Fraction 1 of 2


Target DVHs: Fraction 1 of 2
--PTV1 U PTV2


Target DVHs: Fraction 2 of 2
PTV1


u0 102030405060708090
Dose [Gy]
Saliva gland DVHs: Cumulative dose

100 ---1 -left parotid gland
--- left submandibular gland
80~ right parotid gland
---nrght submandibular gland



43




0 10 20 3040 5060 7080 90
Dose [Gy]
Fraction 2 of 2


-5 0 5 -5 0 5


Figure 4-11. Target DVHs, saliva DVHs and axial slices in Fractions 1 (left) and 2 (right)
using spatial coefficients.






































Dose [Gy]
Fraction 1 of 2


Target DVHs: Fraction 1 of 2


Target DVHs: Fraction 2 of 2
PTV1


--PTV1 U PTV2


100


102030405060708090
Dose [Gy]
Target DVHs: Cumulative dose


u0 102030405060708090
Dose [Gy]
Saliva gland DVHs: Cumulative dose

00 \. left parotid gland
'j--- left submandibular gland
80 Ilght parotid gland
-rig ht submandibu lar g land







00 10 20 30 4050 60 70 80 90
Dose [Gy]
Fraction 2 of 2


- PTV2
---PTV1|


..


Figure 4-12.


Target DVHs, saliva DVHs and axial
usingf spatial coefficients.


slices in Fractions 1 (left) and 2 (right)












Target DVHs: Fraction 1 of 2
-PTV1 U PTV2


Target DVHs: Fraction 2 of 2
-PTV1


0

10

0O


0O


00 10 20 30 40 50 60 70 80 90
Dose [Gy]

Target DVHs: Cumulative dose

)0 ------PTV2
---PTV1


10

8
6
LL.

E 4

2







10

8




E 4


0O 10 20 3040 5060 7080 90
Dose [Gy]
Saliva gland DVHs: Cumulative dose

100 --left protid gland
*\--- left submandibular gland
80 \ right parotid gland
0 n- rght submandibular gland



E 640

20


00 10 20 30 40 50 60 70 80 90
Dose [Gy]
Fraction 2 of 2


10 2030 4050 6070 80 90
Dose [Gy]
Fraction 1 of 2


-5 0 5 -5 0 5


Figure 4-13. Target DVHs, saliva DVHs and axial slices in Fractions 1 (left) and 2 (right)
using spatial coefficients.











Target DVHs: Fraction 1 of 2

a .-PTV1 U PTV2


Target DVHs: Fraction 2 of 2
-PTV1


100

S80

S60

E 40

20


102030405060708090
Dose [Gy]
Target DVHs: Cumulative dose
..~ -PTV2
---PTV1


102030405060708090
Dose [Gy]
Saliva gland DVHs: Cumulative dose

--left parotid gland
Ti ---right parotid gland


13 102030405060708090
Dose [Gy]
Fraction 1 of 2


102030405060708090
Dose [Gy]
Fraction 2 of 2


Figure 4-14. Target DVHs, saliva DVHs and axial slices in Fractions 1 (left) and 2 (right)
using spatial coefficients.










generate a single FMO plan, but the computation times are still acceptable. Further

parameter tuning could possibly yield better results.

The addition of spatial coefficients in the model allows for improved homogeneity,

but does not seem likely to provide additional organ sparing. The improved homogeneity

alone is enough to warrant the inclusion of spatial information in the model. The model is

sensitive to the changes in the spatial coefficients, so further parameter tuning will have to

be performed in small incremental changes.

Currently, the model assumes that prior to each fraction, each target voxel has

received exactly the prescribed amount of dose up to that point in time. While we

have assumed that over/underdose in one fraction should not he compensated by

under/overdose in another fraction, it may in fact he advantageous to allow for some

degree of compensation. The fractionation formulation proposed affords enough flexibility

to model such a scenario. For example,;i wa physician would like to allow underdose in

target s in previous fractions to be compensated by up to ( Gy of overdose in the current

fraction. Then, for target structure s, the P, term in the objective function would be

replaced by the expression max{~7,, P, (}. As this type of discontinuity already exists in

the model, the structure of the model would not he altered by making this modification.

Other future research possibilities include further parameter testing to employ the

model on other cancer site treatments, for example, lung and prostate cancers.









CHAPTER 5
A MONTE CARLO METHOD FOR MODELING DOSE DEPOSITION

5.1 Introduction

The FATO problem relies on the calculation of the amount of total radiation dose

received in each voxel. The dose in a voxel is determined by the paths the photons in

the radiation beams follow through the patient. Some photons may collide with particles

inside the patient and scatter in any direction, while others may collide with particles

and be absorbed. Still other photons may pass entirely through the patient with no

collisions. Due to the unpredictable nature of the radiation beam inside the patient, the

dose received in a voxel can only be accurately obtained through Monte Carlo simulations.

A simple linear relationship is assumed between total dose and heamlet fluences and is

commonly accepted as a satisfactory dose approximation in IMRT optimization. Errors of

as much as t:I' have been reported for photon beams near tissue inhomogeneities (11 I et


For IMRT optimization, particularly with advent of image-guided IMRT (IGIMRT),

or 4D IMRT, the FATO problem must he solved extremely quickly to create real-time

treatment plans. Thus, the speed of the FATO problem is paramount. While Monte Carlo

simulation may provide the most accurate measure of dose, the lengthy computation

time renders the method impractical for clinical use. We propose a Monte Carlo method

that performs a limited number of histories to obtain a noisy approximation of the dose

distribution of each heamlet to which a smoothing function can he applied in order to

determine an accurate dose distribution. The anticipation is that few histories will be

required, and that this approach can he clinically feasible.

Recently, a similar approach has been taken by .Jeleni and Alber [89] and .Jeleri et
[90] with good results. .Jeleri et
at the beam's edge due to a lack of lateral density correction and the effects arising from

MLC systems, for example, tongue-and-groove and inter-leaf scatter. .Jeleri and Alber










[89] pursue the issue of density scaling, but the MLC effects have not yet been addressed.

Section 5.6 proposes some possible methods of accounting for such MLC effects.

5.2 Monte Carlo Engine

The "Dose Planning Method" (DPM) (Sempau et
to perform the Monte Carlo simulations. DPhi is designed to simulate the transport of

photons in radiotherapy class problems. DPhi is primarily based on the public domain

code PENELOPE (Baro et
This study focuses on modeling a finite sized pencil beam emanating from a 6MV

linear accelerator. A finite sized pencil beam is a beam of finite sized that is parallel to

the point source of radiation. To determine a reasonably accurate measure of the dose of

a single beamlet in a given tissue, approximately one billion histories are run in DPM.

As fewer histories are run, the inaccuracies of the dose resulting from the Monte Carlo

experiment grow. Figure 5-4 shows how the noise in the depth-dose curve of the beamlet

becomes increasingly pronounced in relation to the number of histories. As shown by

Table 5-1, the amount of time required to run each experiment is approximately linear in

the number of histories recorded. Thus, it is impractical to run the number of histories

necessary for acceptable accuracy.

5.3 Dose Distribution of a Beamlet

The accuracy of a treatment plan is contingent upon the accuracy of the calculated

dose deposited by each heamlet in the plan. Because the particles in a beamlet scatter in

three dimensional space, multiple dose distributions must he considered to satisfactorily

model the beamlet's affect on the patient's tissue. These distributions arise from the

amount of radiation the beamlet deposits as a function of depth (the depth-dose curve),

and from the amount of radiation radiating outward from the center of the beamlet (the

lateral penumbra).











5.3.1 Depth-Dose Curve

The depth-dose curve represents the radiation intensity deposited by the beamlet

in the tissue through which it passes as function of depth. Figure 5-1 shows the dose

distribution of a single 6MV heamlet in various tissues obtained from the DPhi simulations.

The dose distribution of a beamlet in water is empirically known, and the results from the

DPhi simulation in water can he easily verified to be correct. Muscle, which has nearly

identical density as water (the densities of muscle and water are 1.04g/cm" and 1.00g/cm ,

respectively), has nearly the same depth-dose distribution as water. As expected, a

beamlet passing through lung tissue, which is significantly less dense than water, does not

lose its intensity as quickly as it travels through the less dense tissue. Lastly, a simulation

with inhomogeneous tissue is considered. A simulation of muscle with a 10-cm thick

1 we c- of lung located at a depth of 10cm shows a dose distribution that when the beamlet

reaches the less-dense segment of lung, its depth-dose curve becomes less steep, indicating

that less radiation intensity is lost through the lung than through the muscle. Once the

deeper lIn-;r of muscle is reached, the steepness of the depth-dose curve increases again.


Depth dose curve of beamlet in various tissues after 1B histories
1.1
-water
1 -~ C mu cle

0.9t ------------- muscle-lung-musl

08-

0;







0.3

0.2

0.1
0 5 10 15 20 25 30
depth (cm)


Figure 5-1. Dose distribution of a single beamlet in various tissues.










Although it may seem unintuitive that the depth-dose curve increases at shallow

depths, this behavior is called the build-up curve, and is explained by the likelihood of

electrons scattering out of the tissue and into air at shallow depths. Because the density

of air is extremely small, an electron that reaches air is likely to travel very far away from

the tissue, and therefore unlikely to return to the tissue and deposit radiation dose. Once

the depth increases passes a certain point, the electrons cannot leave the tissue and the

amount of dose received in the tissue increases. Once that point is reached, the amount

of radiation delivered by the beamlet decreases monotonically in depth, as would be

expected.

5.3.2 Lateral Penumbra

In addition to the dose distribution occuringf as the beamlet penetrates the tissue,

there is a dose distribution spreading away from the beamlet. Just as light emanating

from a flashlight in a dark room does not have a discrete boundary between light and

dark, the radiation delivered by a beamlet also does not have a discrete boundary between

what is and is not irradiated. With a circular flashlight heam shown onto a flat surface,

it is apparent from the distribution of the illuminated portion of the surface that some of

the light is diffused into the surrounding darkness as a result of scatter. If the distribution

of light in the circular projection of the flashlight heam is plotted, a hell-shaped curve

describes the brightest point in the center of the illuminated disc decreasing in brightness

as the edge of the illuminated disc is approached, eventually reaching complete darkness.

This behavior is parallel to the behavior of a beamlet passing through any medium.

Fr-om The Physics of Radiation Therapy [94], the penumbra of a beam is the region

at the edge of a radiation beam, over which the dose rate changes rapidly as a function of

distance from the beam axis. Hence, the distribution of radiation dose originating from the

beamlet described above is called the lateral penumbra. Figure 5-2 shows the colorwash

of dose distribution consistitutingf the lateral penumbra, while Figure 5-3 shows the dose










Lateral penumbra of a finite sized pencil beam x 10-4










distace frm b10

Figue 52. olowas ofthe ateal enubraof finte ize pecilbe9

distribution~~ ~ ~ ~ ~ ~~~~1 18teltrlpnmr t ie et noedmeso bandfo n

billon onteCaro hitores f a -cmfinie szedpencl bam i wae-7

5.4 ethoolog to odela Bem -6
Moeig h os itibto o emlti rltvlysrigtowrdfra0eme
in asinle edim. he dffiult arseswhenmulipl meium aretraersd b -5









boundaryof diffeent tisse ty distance e fetromie beamly n hu eure o

Figreo 5-2Clatorwas of thesn lateral p renumbra of a finit sie patencil bhee am em





ihn a singe mdum h diffirn rctulty arisses when) multi pleeiumsaretavred byudr gmthies










Lateral penumbra of 5-cm finite size pencil beam


0.6-


0.4-


0.2-



0 2 4 6 8 10 12
distance (cm)


Figure 5-3. Plot of the lateral penumbra of a finite sized pencil beam


Knowledge of a beamlet's behaviour given certain tissue inhomogeneities can he very

useful in accurately determining dose in a voxel.

5.4.1 Modeling the Depth-Dose Curve

In the section, we analyze the behavior of the depth-dose curve under both single

tissue and multiple tissue scenarios. The goal of the analyzation is to determine the

minimum number of Monte Carlo histories required to obtain a reasonably accurate

approximating function of the dose deposited at each depth in the tissue. For both

the instances of only a single medium and multiple mediums, this is done by fitting

the depth-dose curve from Monte Carlo experiments with varying numbers of histories

to high-degree polynomial functions. The polynomial fits are then compared to the

polynomial fit of a very accurate measure of the depth-dose curve obtained from an

number of Monte Carlo histories accepted to be satisfactorily accurate.

The number of histories recorded in the Monte Carlo simulation can have a drastic

effect on the accuracy of the data collected. For example, Figure 5-4 demonstrates the vast










Depth dose curve of beamlet in water
1.4
S1 B histories
100M histories
1.2~ -10M histories
1 M histories






0. -iiJl





0.4


0 5 10 15 20 25 30
depth (cm)


Figure 5-4. Observed depth-dose curve in water for several histories.


variation observed in the depth-dose curve of a beamlet in water for histories ranging from

one million to one billion. It is hoped that after a certain number of histories, the function

approximation of the data will closely follow the function approximation of very accurate

data obtained from a large number of histories.

For a beamlet in both homogeneous and heterogeneous tissue, the depth-dose curve

can he modeled using a polynomial function of order k. Although the depth-dose curve

may exhibit changes in concavity in the presence of tissue inhomogeneity, a high degree

polynomial will capture the curve's behavior.

The variation of a k-degree polynomial fitted to n-history Monte Carlo data is

measured by


IIn-- 11'pj~ j2

where due/) is the actual observed depth-dose curve from n' Monte Carlo histories and

p(k~tz) is the vector of approximated depth-dose values obtained from a polynomial fit of











Polynomial fits compared to 1 B-history data in water

1B histories: k=27, var=0.050278
100M histories: k=23, var=0.08015
10M histories: k=28, var=0.21877
\ 1M histories: k=24, var=0.54071


1.1-





08-

0;

oOl

i 0.5

0.4

0.3-

0.2-

0 1


0 5 10 15 20 25 30
depth (cm)


Figure 5-5. Polynomial fits of several histories compared to the observed 1B-history
depth-dose curve in water.


degree k to data obtained from n Monte Carlo histories. It is desirable to have that n' =>

to assess the quality of the polynomial fit compared to more accurate data.

In this study, the accuracy of the polynomial obtained is judged by its variation from

the observed data from a very large number of Monte Carlo histories, that is, n' >> a

in the calculation of v~,n. Figure 5-5 shows that for the illustrated number of histories,

the polynomial fit from 100 million histories closely resembles not only the polynomial fit

from one billion histories, but also the actual data collected from one billion histories. The

polynomial fit to one million histories is clearly an unsatisfactory approximation to the

data collected from one billion histories.

For several numbers of Monte Carlo histories, the best approximating polynomial

function with degree in the range [k, k] is found, that is, k* = arg mink6_,k~] Uk~n}. Several

degrees are tested because the degree of the polynomial can significantly affect the quality

of the fit, even for polynomials that are only one degree apart. Figure 5-6 illustrates the

amount of variation observed in the polynomial approximation as a function of the degree











Variation of polynomial fit (v k,n,1e9 ) as function of degree (k)
0.0

0.7-

0.6-

S0.5-

0.4-

tM 0.3-

0.2-





5 10 15 20 25 30 35 40 45 50
degree of polynomial (k)


Figure 5-6. Variation of polynomial fit as function of degree.


of the polynomial for polynomials fitted to the depth-dose curve of a beamlet in water

obtained from 1 billion histories.

5.4.2 Modeling the Lateral Penumbra

In the section, we analyze the behavior of the lateral penumbra under both single

tissue and multiple tissue scenarios. The lateral penumbra of a beam is a hell-shaped

curve that can he approxiniated as the sunt of error function pairs. The error function,

erf(.r), is twice the integral of the Gaussian distribution with mean 0 and variance of 1/2:






Figure 5-7A demonstrates a sample error function. While a single side of the lateral

penumbra of a heantlet resembles an error function, a closer approximation to a single side

of the lateral penumbra is represented as the average of two error functions given by

a .r' + .ro .
erf erf X g)











Sample error function erf(x)


Sample error function pair for lateral penumbra


06-


-08-
-04-

-02- 04
-0 4


-3-2-10123 0123456

A B


Figure 5-7. An error function and an error function pair. A) Error function. B) Error
function pair.


where a is the amplitude, xo is the offset and a is the variation of the two error functions.

The expression is divided by 2 to take the average of the error function pair. An example

of an error function pair is given in Figure ??B.

Because the lateral penumbra of a beamlet resembles an error function on both the

left- and right-hand sides of the beam center, the lateral penumbra L(x) is represented as

the sum of the average of NV error function pairs, given by


as x zoX Xo,
L~~x) = erf-er,


where ai is the amplitude, xoi is the offset and ai is the variation of error function pair i,

i= 1,...,NV.

To determine the parameters ai, xoi and ai for each of the NV error function pairs, a

Levenberg-Marquardt quI I---N --ton minimization method is emploi-. I This method takes

as input NV and an initial guess of the parameters and returns a locally optimal solution

to the problem of minimizing the variation between the real data and the sum of the error

function pairs.

At a given depth in the tissue, the amplitude of the error function is determined by

the value of the depth-dose curve at that depth. Thus, for each tissue type, it is only










Lateral penumbra of 5-cm finite sized pencil beam in water


0.8-


0.6 -~


S0.4 -


0.2-



0 2 4 6 8 10 12
distance (cm)


Figure 5-8. Lateral penumbra for several numbers of Monte Carlo histories.


necessary to model a single lateral penumbra, and then that model can he extended to

all depths simply by manipulating the amplitude according to the depth-dose curve.

Figure 5-3 shows the lateral penumbra of a 5-cm finite sized heamlet at a fixed depth

in water for a number of Monte Carlo histories deemed to yield a satisfactorily accurate

representation of the dose deposited in the tissue. Using the method described above, the

lateral penumbra was modeled to yield the approximation to the observed data collected

for the various Monte Carlo histories shown in Figure 5-8.

In a similar fashion to the method for modeling the depth-dose curve, the method

for modeling the lateral penumbra consists of fitting the sum of error function pairs to

the lateral penumbra data. The quality of these fits is judged by their variation from the

observed data for a sufficiently large number of Monte Carlo histories to obtain accurate

dose information.











Error function pair fits com pared to 1 B-history data in water

1B histories: var=0.070667
100M histories: var=0.075414
10M histories: var=0.14511
1M histories: var=1.1829


1.2 -


1 -


0.8-


06


0

0


0 2 4 6 8 10 12
distance (cm)


Figure 5-9. Error function fits of several histories compared to the observed 1B-history
lateral penumbra of a beamlet in water.


Just in as the method for determining the quality of the depth-dose curve approximation,

the variation of the error function fit from the actual lateral penumbra is calculated as



va,,, = L "' ""


where L("/) is the observed lateral penumbra data from a simulation of n' histories, and

L("'") is the approximated lateral penumbra obtained from the parameters fitted to the

expression L (X). It is desirable to have that n' > n.

Figure 5-9 di pk.l--s the error function pair fits obtained from the Levenburg-Marquardt

method, as well as the variation of the fits from the observed data from one billion

histories. The variation is measured in the same manner as described in Section 5.4.1.

It is anticipated that although the lateral penumbra exhibits different dose distributions

in materials of different densities, the distribution will only show a fundamental change

in shape if the beam simultaneously hits multiple tissues of varying densities. In such a

situation, the penumbra, which is taken to be symmetric about the center of the beam in


1.4-


1.2-


0










Table 5-1. Computation times in minutes of Monte Carlo simulations
n Water Muscle Lung Muscle-Lung-Muuscle
le9 222.184 211.887 111.318 1863.894
100e6 20.543 21.2563 11.239 18.701
10e6 2.210 2.234 1.269 1.986
le6 0.244 0.339 0.233 0.309


homogeneous tissue, will no longer be symmetric. To model the lateral penumbra under

inhomogeneous material, a sum of error function pairs can still be emploi- II though it

may be necessary to increase the number of error function pairs required. The difficulty

will lie in correctly determining when the addition of additional error function pair will

be needed. A possible measure could be the variation between the lateral penumbra

approximation and the observed data.

5.5 Results

The homogeneous tissues tested are water, muscle and lung, and the heterogeneous

material tested consists of muscle and lung. Each scenario is considered to have a depth

of 30cm. The voxel sizes are 5mm x 5mm x 5mm, and a 5-cm finite sized pencil beam

is considered. For each simulation, tests were run with 1 billion, 100 million, 10 million

and 1 million Monte Carlo histories in DPM on a Mac OS X 10.4.6 machine with dual

2.3GHz PowerPC G5 processors and 8GB of RAM. Due to time constraints, the muscle

tests are run to a maximum of 100 million iterations, and all comparisons to the fit quality

are made to this 100-million-history data instead of the 1-billion-history data used for the

other simulations.

As can be seen from the computation times in Table 5-1, the run time of DPM is

approximately linear in the number of histories. Although a larger number of Monte Carlo

histories yields improved accuracy, the maximum number of histories considered is one

billion because of time limitations and the satisfactory accuracy of the 1-billion-history

runs .









For each of the tested tissue types, the depth-dose curves and lateral penumbras

were modeled using the methods described in Section 5.4. For the polynomial fits of the

depth-dose curve, the values k and k are chosen as 10 and 45, respectively. By choosing

the polynomial approximation over such a large range of degree values, an acceptably

accurate fit is likely to be found.

For the lateral penumbra, NV was chosen as 4 because in addition to the obvious need

for two error functions to model the sides of the lateral penumbra, an additional error

function is needed to model each tail with reasonable accuracy. For example, the four

error functions used to model the lateral penumbra of a beamlet in water (Figure 5-9) are

shown separately in Figure 5-10. The computation times required to obtain each of the

function approximations are di;l li-x d in Table 5-2.

The initial parameters ai, xoi and ai for each error function pair i, i = 1,. .. N, used

to approximate the lateral penumbra are obtained by the following method. Of the four

error function pairs considered, two of the error functions-I = {1, 2}-are used to model

the steep sides of the lateral penumbra, and the other two error functions-I = {3, 4}-are

used to model the tails of the dose distribution. At a given depth z, the amplitude ai is



ai d(z~~)/50 iE I,


where d(z) represents the value of the depth-dose curve approximation at a depth z. The

expression for the amplitude when i el was obtained by experimenting with several

different fractions of d(z).

The a value of the error functions determines the shape of the error function curve.

As a increases, the curve becomes increasingly spread out. Thus, it is desirable to have

a small ai value for i El since the error function in I only need to model the sides of

the lateral penumbra, and a larger ai value for i El since the error function in I need

to model the elongated tails of the lateral penumbra. For the tissues tested, the ai values










Table 5-2. Computation times in seconds of approximating function fits to the dose
distribution. The polynomial fits to the depth dose curve are represented by
D.D., and the error function fits to the lateral penumbra are represented by
Lat. Pen.
Water M~uscle Lung M~uscle-Lung-M~uscle
n D.D. Lat.Pen. D.D. Lat.Pen. D.D. Lat.Pen.. D.D. Lat.Pen.
le9 0.078 2.640 0.078 2.422 0.094 1.062 0.078 n/a
100e6 0.078 1.172 0.078 2.625 0.828 0.906 0.109 n/a
10e6 0.110 3.454 0.109 1.390 2.609 2.594 0.093 n/a
le6 0.094 1.407 0.094 1.172 1.063 0.953 0.078 n/a


used are

ai0.4 iel
0.8 iE I,

These values were obtained through experimentation.

For the 5-cm finite sized pencil beams used in this experiment, the offsets xoi were

empirically set at values of 8.5, -3.5, 11 and -1 for i = 1, ... N, respectively. A method of

identifying the locations of these offsets based on the Monte Carlo data can be developed

by basing the offsets on the slope of the observed data, and is planned for future research.

The results for the fits of both the depth-dose curve and the lateral penumbra of a

beamlet in water are shown in the examples in Section 5.4. Figures 5-11-5-12 show the

results of the fits for the muscle and lung tissues. From the computational results, it is

clear that the time to obtain fits to the Monte Carlo data is insignificant compared with

the amount of time required to run the Monte Carlo histories, even for as few as 1 million

histories.

To test the model in the presence of tissue inhomogeneity, a 10cm-thick 1... -r of lung

between two 10cm-thick lIn-;-rs of muscle is considered. As expected, for the first 10cm,

the depth-dose curve of the muscle-lung-muscle case is identical to that of the muscle

depth-dose curve. Once the beamlet reaches the significantly less dense 1 ... r of lung (lung

has a density of 0.30g/cm3), a predominant change in the depth-dose curve is evident

(Figure 5-1). Once the lIn-;-r of lung is reached, the rate of decrease in the amount of dose

deposited in the tissue decreases, that is, less radiation intensity is lost as the beamlet



















Error functions for 5-cm finite sized pencil beam in water


0 2 4 6
distance (cm)


8 10 12


Figure 5-10. Error function pairs summed to approximate a beamlet in water.


Depth dose curve of beamlet In muscle
1B hionstns
-100M histories
-10M histones
1M histories


Polynomial Ots compared to 1B-history data In muscle
1
1B histones k=27, var=0 046123
-100M histories k=22, var=0072
1 -10M histones k=22, var-0 20284
1M histories k=24. var=0 69751


--~-
-a


10 15
depth (cm)


20 25 30


10 15
depth (cm)


20 25 30


Figure 5-11. Depth-dose curves in muscle tissue.
fits.


Ai) MVont~e Clarlo historiles. B) Polynomial




















Lateral penumbra of 5-cm fnite sized pencil beam In muscle

--1B histones
--- 100M histories
--- 10M histones
1M histories


Error functions for 5-cm finite sized pencil beam In muscle

1B histones var-0 10448
---- 100M histones var=0 11761
---- 10M histories var-012386
1M histones var=0 20889i


05
04
03
02




0 I


0 2


6
distance (cm)


10 12


4 6
distance (cm)


10 12


Figure 5-12. Lateral penumbra curves in muscle tissue.
function fits.


A) Monte Carlo histories. B) Error


Depth dose curve of beamlet In lung


Polynomial Ots compared to 1B-history data In lung

1B histones k=22, var-0056789
100M histones k=22. var=0 110
1 2 ---- 10M histories k=22. var-0 32993
1M histones k=24. var=1 0547



08--

06O


1B histones
100M histories
10M histones
1M histories


10 15
depth (cm)


20 25 30


10 15
depth (cm)


20 25 30


A



Figure 5-13. Depth-dose curves in lung tissue. A)
fits.


B



Monte Carlo histories. B) Polynomial











Lateral penumbra of 5-cm finite sized pencil beam In lung Error function pair fts compared to 1B-history data In lung
1 2 1B histones 1 2 ---1B histones var-0 097127
--- 100M histories --- 100M histones var=0 12
-10M histones --- 10M histories var-012493
1M histories 1~ 1M histones var=0 12937-




~06- 06


04- 04


0 2 4 6 8 10 12 0 2 4 6 8 10 12
distance (cm) distance (cm)

A B


Figure 5-14. Lateral penumbra curves in lung tissue. A) Monte Carlo histories. B) Error
function fits.


passes through the lung tissue. When the beamlet reaches the second 1.v-;r of muscle, this

rate increases again. The same approach used to model the depth-dose curve in a single

tissue continues to work well in multiple tissue. Figures 5-15A and 5-15B illustrate the

ability of a polynomial to approximate the depth-dose curve in inhomogeneous tissue.

Because testing the beamlet in a scenario where it could hit multiple tissues

simultaneously is reserved for future research, results for modeling the lateral penumbra in

the multiple-tissue scenario tested are identical to those for the single-tissue scenario. The

lateral penumbra at a given depth in a certain tissue can be modeled by using the dose

from the depth-dose curve at the given depth as the amplitude of the lateral penumbra.

The dose distribution in the lateral penumbra can then be modeled according to the same

error function pairs used in modeling the lateral penumbra in a single-tissue scenario of

the same medium.

Figure 5-16 illustrates the variations of the fits used to approximate the depth-dose

and lateral penumbra distributions of a beamlet in water as a function of the number

of histories. From this data, it is very clear that the accuracy of the beamlet model is

directly correlated with the number of Monte Carlo histories. It is interesting that there

is not a significant improvement in the beamlet model accuracy from 100 million to 1











Depth dose curve of beamlet In muscle-lung-muscle Polynomlal Oits compared to 1B-history data In muscle-lung-muscle
--- 1B0 histones 1 -- 1B0 histones k=27, var-0 029
--100 histories 1 1 100 histones k=27, var=0 105053
1 4 10M histones 10M histories k=23, var-0 21323
1M histories 1M histones k=22, var=0 88061
12 09







041
0203
0 2 5 10 15 20 25 30 0 2 5 10 15 20 25 30
depth (cm) depth (cm)

A B


Figure 5-15. Depth-dose curves in heterogeneous muscle and lung tissue. A) Monte Carlo
histories. B) Polynomial fits.

Table 5-3. Variation of fits to several numbers of histories with d' = 1 billion.
Water M Iuscle Lung M Iuscle-Lung-M~uscle
73 'Uk*,n,n' Vn,n' 'Uk*,n,n' Vn,n' 'Uk*,n,n' Vn,n' 'Uk*,n,n' Un,n'
le9 0.050 0.071 0.046 0.105 0.057 0.097 0.052 n/a
100e6 0.080 0.075 0.075 0.118 0.110 0.103 0.101 n/a
10e6 0.219 0.145 0.203 0.124 0.330 0.125 0.213 n/a
le6 0.541 1.183 0.698 0.209 1.055 0.129 0.881 n/a


billion histories, and computing 100 million histories requires approximately one tenth of

the amount of time as computing 1 billion histories. Depending on the composition of the

tissue, it may be reasonably accurate to only require 10 million histories, particularly in

the depth-dose curve approximation.

5.6 Conclusions and Future Directions

In conclusion, the Monte Carlo approach presented is emploi-v I to model the dose

distribution of a beamlet using a limited number of histories. Using the polynomial

and error function pair fitting techniques described, dose distributions with satisfactory

accuracy can be obtained using at least a factor of 10 fewer Monte Carlo histories than

would otherwise be required. This can greatly decrease the amount of time required to

obtain dose data for beamlets in the FMO problem of IMRT treatment planning without

sacrificing accuracy.












Figure 5-16. Variations of the fits used to approximate the depth-dose and lateral
penumbra distributions as function of the number of histories.
Variation of approximations as a function of number of histories
1.4
SDepth-dose: water
8 Lateral penumbra: water
1.2~ Depth-dose: muscle
\ 8 Lateral penumbra: muscle
\ Deepth-dose: lung
1 8 Lateral penumbra: lung
\ ------ii-------Depth-dose: mus l-u g m s l
c- 0.8-


> 0.6t


0.4-


0.2E 1 -



106 107 108 109
number of histories



For future testing, more tests on the number of Monte Carlo histories needed will

be run as well, particularly with histories in the range of 10-100 million. More tests of

varying tissues, both homogeneous and heterogeneous, will be run to determine a smaller

range of degrees to be evaluated for the polynomial fit to the depth-dose curve. An

automated method of determining a quality set of initial parameters to model the lateral

penumbra will also be developed. Lastly, the scenario where a beamlet hits multiple

tissues simultaneously will be tested using our model for approximating the lateral

penumbra.

Jeledi and Alber [89] and Jeleri et al. [90] have demonstrated that a beamlet can

be modeled very effectively using an approach based on the one described here. This

approach was improved upon by scaling the modeling parameters according to tissue

density in Jeleri and Alber [89]. Despite the sophistication of the density scaling method

emploi- II the model loses accuracy in the penumbra regions and at the edge of tissue

heterogeneities. This study also used a Levenberg-Marquardt algorithm to determine the










modeling parameters, and although the details of the intplenientation are not provided, it

is possible that with an improved initial guess or damping parameter, the algorithm could

converge to better modeling parameters, thus providing improved prediction of heantlet

behavior at the penumbra.

To further improve upon their work, the effects of the MLC must he considered. One

method of accounting for these effects could be to model the dose deposition of an entire

aperture rather than just the dose deposition of a single beamlet. As the number and

shape of apertures required to deliver an FMO-hased IMRT optimization are unknown,

this method would be most practical if an aperture modulation approach--where aperture

fluences front a pre-defined set of apertures are chosen, instead of fluences from individual

beantlets--is eniploi- 1 instead of an FMO approach, as the number and shape of the

apertures in consideration are predetermined.









REFERENCES


[1] American Cancer Society. Cancer Facts and Figures Report. 2006.

[2] Murphy GP, Lawrence WL, Lenlard RE, eds. American Cancer S8... .:. Iri Tex~tbook on
CI.,..... ., Or:. .. I ~it;, The American Cancer Society, 1995.

[3] Perez CA, Brady LW. Principles and Practice of Radiotherap~y. Lippincott-Raven, 3
edn., 1998.

[4] Steel GG. Basic CL-.,:....; ~Rl EJ..l.: I..J. ,I~ ;i for Radiation Oncologists. Edward Arnold
Publishers, 1994.

[5] Ma C'j!, Mok E, K~apur A, Findley D, Brain S, Bci-;r AL. Clinical implementation of
a monte carlo treatment planning system. Medical Ph ;,:. 1999;26:2133-43.

[6] Bortfeld T. Optimized planning using physical objectives and constraints. Semin
Radial Oncol 1999;9:20-34.

[7] Alber M, Nusslin F. An objective function for radiation treatment optimization based
on local biological measures. Phys M~ed Biol 1999;44:479-493.

[8] Jones LC, Hoban PW. Treatment plan comparison using equivalent uniform
biologically effective dose (eubed). Phys M~ed Biol 2000;45:159-170.

[9] K~allman P, Lind BK(, Brahme A. An algorithm for maximizing the probability
of complication-free tumor-control in radiation-therapy. Phys M~ed Biol 1992;
37:871-890.

[10] Mavroidis P, Lind BK(, Brahme A. Biologically effective uniform dose for specification,
report and comparison of dose response relations and treatment plans. Phys M~ed Biol
2001;46:2607-2630.

[11] Niemierko A. Reporting and analyzing dose distributions: a concept of equivalent
uniform dose. Medical Ph ; -:. 1997; 24:103-110.

[12] Niemierko A, Urie M, Goitein M. Optimization of 3d radiation-therapy with both
physical and biological end-points and constraints. Int J Radiat Oncol Biol Phys
1992;23:99-108.

[13] Wu QW, Dj ii Ilputra D, Wu Y, Zhou JN, Liu HH, Mohan R. Intensity-modulated
radiotherapy optimization with geud-guided dose-volume objectives. Phys M~ed Biol
2003;48:279-291.

[14] Wu QW, Mohan R, Niemierko A, Schmidt-Ullrich R. Optimization of
intensity-modulated radiotherapy plans based on the equivalent uniform dose.
Int J Radial Oncol Biol Phys 2002;52:224-235.










[15] Hamacher HW, Kiifer K(H. Inverse radiation therapy planning a multiple objective
optimization approach. Discrete Applied M~athematics 2002;118:145-161.

[16] Bednarz G, Michalski D, Houser C, Huq hIS, Xiao Y, Anne PR, Galvin JM. The use
of mixed-integer programming for inverse treatment planning with pre-defined field
segments. Phy~s Afed Biol 2002;47:22:35-2245.

[17] Ferris MC, ?1. i-n r RR, D'Souza W. Radiation treatment planning: Mixed integer
programming formulations and approaches. In G Appa, L Pitsoulis, HP Williams,
eds., Handbook on M~odelling for Discrete Op~timization. Springer-V. 11 I_ New York,
NY, 2006;:317-:340.

[18] Langer AI, Brown R, Urie AI, Leong J, Stracher AI, Shapiro J. Large-scale
optimization of beam weights under dose-volume restrictions. Int J Radial Oncol
Biol Phy~s 1990;18:887-89:3.

[19] Langer AI, Morrill S, Brown R, Lee O, Lane R. A comparison of mixed integer
programming and fast simulated annealingf for optimizing beam weights in radiation
therapy. Afedical Ph a 7. 1996; 23:957-964.

[20] Lee EK(, Fox T, Crocker I. Simultaneous heam geometry and intensity map
optimization in intensity-modulated radiation therapy treatment planning. An-
nests of Op~erations Research 200:3;119:165-181.

[21] Lee EK(, Fox T, Crocker I. Integer programming applied to intensity-modulated
radiation therapy treatment planning. Int J Radial Oncol Biol Phy~s 2006;64::301-:320.

[22] Shepard DM, Ferris MC, Olivera GH, Mackie TR. Optimizing the delivery of
radiation therapy to cancer patients. SIAAF Review 1999;41:721-744.

[2:3] Romeijn HE, Alandi I RK(, Dempsey JF, K~umar A, Li JG. A novel linear programming
approach to fluence map optimization for intensity modulated radiation therapy
treatment planning. Phy~s Afed Biol 200:3;38::3521-:3542.

[24] Romeijn HE, Alandi I RK(, Dempsey JF, K~umar A, Li JG. A column generation
approach to radiation therapy treatment planning using aperature modulation. SIAAF
Journal of Op~timization 2005;15:8 :; ;1.2

[25] Romeijn HE, Dempsey JF, Li JG. A unifying framework for multi-criteria fluence
map optimization models. Phy~s Afed Biol 2004;49:1991-201:3.

[26] Romeijn HE, Alani I RK(, Dempsey JF, K~umar A. A new linear programming
approach to radiation therapy treatment planning problems. Operations Research
2006;54:201-216.

[27] Das SK(, Marks LB. Selection of coplanar or noncoplanar beams using
three-dimensional optimization based on maximum heam separation and minimized









nontargfet irradiation. Int J Radiat Oncol Biol Phys 1997;38:643-655.


[28] Haas OC, Burnham K(J, Mills J. Optimization of beam orientation in radiotherapy
using planar geometry. Phys M~ed Biol 1998;43:2179-2193.

[29] Schreibmann E, Lahanas M, Xing L, Baltas D. Multiobjective evolutionary
optimization of the number of beams, their orientations and weights for
intensity-modulated radiation therapy. Phys M~ed Biol 2004;49:747-770.

[30] Chao K(SC, Blanco AI, Dempsey JF. A conceptual model integrating spatial
information to assess target volume coverage for IMRT treatment planning. Int J
Radial Oncol Biol Phys 2003;56:1438-1449.

[31] Nocedal J, Wright SJ. Numerical Op~timization. Springer-V. ~11 q 1999.

[32] Ezzell GA. Genetic and geometric optimization of three-dimensional radiation therapy
treatment planning. Medical Ph;,. 1996;23:293-305.

[33] Li Y, Yao J, Yao D. Automatic beam angle selection in IMRT planning using genetic
algorithm. Phys M~ed Biol 2004;49:1915-1932.

[34] Li Y, Yao J, Yao D, C'I. in W. A particle swarm optimization algorithm for beam
angle selection in intensity-modulated radiotherapy planning. Phys M~ed Biol 2005;
50:3491-3514.

[35] Bortfeld T, Schlegel W. Optimization of beam orientations in radiation therapy: some
theoretical considerations. Phys M~ed Biol 1993;38:291-304.

[36] Djajaputra D, Wu Q, Wu Y, Mohan R. Algorithm and performance of a clinical
IMRT beam-angle optimization system. Phys M~ed Biol 2003;48:3191-3212.

[37] Lu HM, K~ooy HM, Leber ZH, Ledoux RJ. Optimized beam planning for linear
accelerator-based stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1997;
39:1183-1189.

[38] Pugachev A, Xing L. Incorporating prior knowledge into beam orientation
optimization in IMRT. Int J Radiat Oncol Biol Phys 2002;54:1565-1574.

[39] Rowbottom CG, Oldham M, Webb S. Constrained customization of non-coplanar
beam orientations in radiotherapy of brain tumours. Phys M~ed Biol 1999a;44:383-399.

[40] Stein J, Mohan R, Wang XH, Bortfeld T, Wu Q, Preiser K(, Ling CC, Schlegel W.
Number and orientations of beams in intensity-modulated radiation treatments.
Medical Ph ;, -0. ; 1997; 24:149-160.

[41] Soderstrom S, Brahme A. Selection of suitable beam orientations in radiation therapy
using entropy and fourier transform measures. Phys M~ed Biol 1992;37:911-924.










[42] Soderstrom S, Brahme A. Which is the most suitable number of photon beam portals
in coplanar radiation therapy? Int J Radiat Oncol Biol Phys 1995;33:151-59.

[43] Rowbottom CG, Webb S, Oldham M. Beam-orientation customization using an
artificial neural network. Phys M~ed Biol 1999b3;44:2251-2262.

[44] Gokhale P, Hussein EM, K~ulkarni N. The use of beams eye view volumetrics in the
selection of non-coplanar radiation portals. Medical Ph;,.. 1994;23:153-163.

[45] Meedt G, Alber M, Niisslin F. Non-coplanar beam direction optimization for
intensity-modulated radiotherapy. Phys M~ed Biol 2003;48:2999-3019.

[46] Chen GT, Spelbring DR, Pelizzari CA, Balter JM, Myrianthopoulos LC, Vii .r kIinew1r
S, Halpern H. The use of beams eye view volumetrics in the selection of non-coplanar
radiation portals. Int J Radiat Oncol Biol Phys 1992;23:153-163.

[47] Cho BCJ, Roa HW, Robinson D, Murray B. The development of target-eye-view
maps for selection of coplanar or noncoplanar beams in conformal radiotherapy
treatment planning. Medical Ph;,.. 1999;26:2367-2372.

[48] Goitein M, Abrams M, Rowell D, Pollari H, Wiles J. Multi-dimensional treatment
planning: li. beams eye-view, back projection, and projection through CT sections.
Int J Radial Oncol Biol Phys 1983;9:789-97.

[49] Pugachev A, Xing L. Computer-assisted selection of coplanar beam orientations in
intensity-modulated radiation therapy. Phys M~ed Biol 2001;46:2467-2476.

[50] Pugachev A, Xing L. Pseudo beam's-eye-view as applied to beam orientation selection
in intensity-modulated radiation therapy. Int J Radiat Oncol Biol Phys 2001;
51:1361-1370.

[51] Holder A, Salter B. A tutorial on radiation oncology and optimization. In
H Greenberg, ed., Tutorials on Emerging M~ethodologies and Applications in Op-
erations Research. K~luwer Academic Press, Boston, MA, 2004.

[52] Morrill SM, Lane RG, Jacobson G, Rosen II. Treatment planning optimization using
constrained simulated annealing. Phys M~ed Biol 1991;36:1341-61.

[53] Oldham M, K~hoo V, Rowbottom CG, Bedford J, Webb S. A case study comparing
the relative benefit of optimising beam-weights, wedge-angles, beam orientations
and tomotherapy in stereotactic radiotherapy of the brain. Phys M~ed Biol 1998;
43:2123-46.

[54] Rowbottom CG, Webb S, Oldham M. Improvements in prostate radiotherapy from
the customization of beam directions. Medical Ph ;,;. ; 1998;25:1171-1179.










[55] Wang X, Zhang X, Dong L, Lie H, Wu Q. Alohan R. Development of methods for
beam angle optimization for IMRT using an accelerated exhaustive search strategy.
Int J Radial Oncol Biol Phy~s 2004;60:1:325-:37.

[56] Wang X, Zhang X, Dong L, Liu H, Gillin AI, Ahamad A, Ang K(, Mohan R.
Effectiveness of noncoplanar IMRT planning using a parallelized multiresolution
beam angle optimization method for paranasal sinus carcinoma. Int J Radial Oncol
Biol Phy~s 2005;6:3:594-601.

[57] Woudstra E, H. ~ils, Il BJM. Automated heam angle and weight selection in
radiotherapy treatment planning applied to pancreas tumors. Int J Radial Oncol
Biol Phy~s 2004;56:878-88.

[58] D'Souza WD, ?1. i-n r RR, Shi L. Selection of beam orientations in intensity-modulated
radiation therapy using single-beam indices and integer programming. Phy~s Afed Biol
2004;49::3465-:3481.

[59] Ehrgott AI, Johnston R. Optimisation of beam directions in intensity modulated
radiation therapy planning. OR Sp~ectrum 200:3;25:251-264.

[60] Lim J, Ferris AI, Shepard D, Wright S, Earl 3. An optimization framework for
conformal radiation treatment planning. INFORM~S Journal On C'omp~uting 2006.

[61] Wang C, Dai J, Hu Y. Optimization of beam orientations and heam weights for
conformal radiotherapy using mixed integer programming. Phy~s Afed Biol 200:3;
48:4065-4076.

[62] Fox C, Romeijn HE, Dempsey JF. Fast voxel and polygon r n, -tiracing algorithms for
IMRT treatment pI l ....ilr_ 2005. Submitted to Afedical Ph ; ..~

[6:3] Siddon R L. Prism representation: a :3d r n, -tiracing algorithm for radiotherapy
applications. Phy~s Afed Biol 1985;8:817-824.

[64] Siddon RL. Fast calculation of the exact radiological path for a three-dimensional CT
array. Afedical Ph;;-.. 1985;12:252-255.

[65] Jacobs F, Sundermann E, Sutter BD, C!~! -I I~ ..- Is Lemahieu I. A fast algorithm
to calculate the exact radiological path through a pixel or voxel space. Journal of
C'ovrl. .9,:l and Infortuation T,.~ In...J..,I;; (G'IT) 1998;6:89-94.

[66] Aleman DM, Romeijn HE, Dempsey JF. Beam orientation optimization methods
in intensity modulated radiation therapy treatment planning. IIE O'onference
Proceedings 2006.

[67] Aleman DM, Romeijn HE, Dempsey JF. A response surface approach to heam
orientation optimization in intensity modulated radiation therapy treatment planning.
In review 2006.










[68] Jones DR. A ::u~rsulrni- of global optimization methods based on response surfaces.
Journal of Global Op~timization 2001;21::345-38:3.

[69] Jones DR, Schonlau AI, Welch WJ. Efficient global optimization of expensive
black-box functions. Journal of Global Op~timization 1998;1:3:455-492.

[70] Csallner AE, Csendes T, Alarkot MC. Multisection in interval branch-and-hound
methods for global optimization i. theoretical results. Journal of Global Op~timization
2000;16::371-392.

[71] Lagouanelle J, Soubry G. Optimal multisections in interval branch-and-hound
methods of global optimization. Journal of Global Op~timization 2004;:30:2:3-38.

[72] Epperly TGW, Pistikopoulos EN. A reduced space branch and bound algorithm for
global optimization. Journal of Global Op~timization 1997;11:287-:311.

[7:3] Barrientos O, Correa R. An algorithm for global minimization of linearly constrained
quadratic functions. Journal of Global Op~timization 2000;16:77-9:3.

[74] Thoai NV. Convergence of duality bound method in partly convex programming.
Journal of Global Op~timization 2002;22:26:3-270.

[75] Tuy H. On solving nonconvex optimization problems by reducing the duality gap.
Journal of Global Op~timization 2005;:32::349-:365.

[76] Phong TQ. An LTH, Tao PD. Decomposition branch and bound method for globally
solving linearly constrained indefinite quadratic minimization problems. Operations
Research Letters 1995;17:215-220.

[77] Bomze I. Branch-and-hound approaches to standard quadratic optimization problems.
Journal of Global Op~timization 2002;2:17-:37.

[78] Cambini R, Sodini C. Decomposition methods for solving nonconvex quadratic
programs via branch and bound. Journal of Global Op~timization 2005;:33::31:3-336.

[79] Aleman DM, K~umar A, AlMll I RK(, Romeijn HE, Dempsey JF. Neighborhood search
approaches to heam orientation optimization in intensity modulated radiation therapy
treatment planning. in review 2007.

[80] K~umar A. Novel methods for 1. to1,;i;,-modulated radiation therapy treatment
plant:,..t:l Ph.D. thesis, University of Florida, 2005.

[81] Geman S, Geman D. Stochastic relaxation, gihhs distributions, and the 'I li- Im
restoration of images. IEEE Transactions on Pattern A,: ale; -. and M~achine Intelli-
gence 1984;6:721-741.










[82] Gelfand AE, Smith AFM. Sampling based approaches to calculating marginal
densities. Journal of the American Statistical Association 1990;85:398-409.

[83] Smith RL. A monte carlo procedure for the random generation of feasible solutions
to mathematical programming problems. Bulletin of the TIM~S/ORSA Joint National
Meeting 1980;:101.

[84] Bi61isle C JP, Romeijn HE, Smith RL. Hit-and-run algorithms for generating
multivariate distributions. Mathematics of Op~erations Research 1993;18:255-266.

[85] K~irkpatrick S, Gelatt CD. Optimization by simulated annealing. Science 1983;
220:671-680.

[86] Bomze I. Fast simulated annealing. Ph;,:. Letters 1987;122A:157-162.

[87] Bi61isle C JP. Convergence theorems for a class of simulated annealing algorithms on
R d. Journal of Applied P,~~rol~.,t.7.1;i 1992;29:885-895.

[88] Aleman DM, Glaser D, Romeijn HE, Dempsey JF. A primal-dual interior point
algorithm for fluence map optimization in intensity modulated radiation therapy
treatment planning. work in progress 2007.

[89] Jeledi U, Alber M. A finite size pencil beam algorithm for IMRT dose optimization:
density corrections. Ph ;,;:. in M~edicine and B: ..l. .i;, 2007;52:617-633.

[90] Jeledi U, Sohn M, Alber M. A finite size pencil beam for IMRT dose optimization.
Ph ;-: i- in M~edicine and B: ..l. .i;, 2005;50:1747-1766.

[91] Sempau J, Wilderman SJ, B;.1 li. w AF. Dpm, a fast, accurate monte carlo code
optimized for photon and electron radiotherapy treatment planning dose calculations.
Phys M~ed Biol 2000;45:2263-91.

[92] Bar6 J, Sempau J, Fernindez-Varea JM, Salvat F. Penelope: An algorithm for monte
carlo simulation of the penetration and energy loss of electrons and positrons in
matter. Nuclear Instruments and M~ethods 1995;B100:31-46.

[93] Sempau J, Bar6 J, Fernindez-Varea JM, Salvat F. An algorithm for monte carlo
simulation of coupled electron-photon showers. Nuclear Instruments and M~ethods
1997;B132:377-90.

[94] K~han FM. The Ph;,:. of Radiation Therap~y. Lippincott William and Wilkins, 1994.










BIOGRAPHICAL SKETCH

Dionne AI. Alenian completed her bachelor's degree in industrial and systems

engineering at the University of Florida. She went on to study intensity modulated

radiation therapy (IMRT) treatment planning optimization in the graduate program of the

Department of Industrial and Systems Engineering at the University of Florida. She will

receive her Doctor of Philosophy in Industrial and Systems Engineering in December of

2007, after which she will pursue a career in the Department of Mechanical and Industrial

Engineering at the University of Toronto. Dionne plans to continue her research in cancer

treatments, as well as other applications of operations research techniques to the medical

and healthcare industries.





PAGE 1

OPTIMIZATIONMETHODSININTENSITYMODULATEDRADIATIONTHERAPYTREATMENTPLANNINGByDIONNEM.ALEMANADISSERTATIONPRESENTEDTOTHEGRADUATESCHOOLOFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENTOFTHEREQUIREMENTSFORTHEDEGREEOFDOCTOROFPHILOSOPHYUNIVERSITYOFFLORIDA2007 1

PAGE 2

c2007DionneM.Aleman 2

PAGE 3

Tomyever-patientwifeNancy,andtomyfatherRoberto,who,ifnotfortheshortcomingsofcurrentcancertreatments,mightstillbewithustoday 3

PAGE 4

ACKNOWLEDGMENTSManythankstoNancyHuang,ChristopherFoxandBartLynchforsohelpfullyandhappilyexplainingthephysicsofmedicalphysicstomeonawiderangeoftopics,evenwhenthosetopicsarenotrelevanttomyownresearch.ThisworkwassupportedinpartbytheNSFAlliancesforGraduateEducationandtheProfessoriate,theNSFGraduateResearchFellowshipandNSFgrantDMI-0457394. 4

PAGE 5

TABLEOFCONTENTS page ACKNOWLEDGMENTS ................................. 4 LISTOFTABLES ..................................... 8 LISTOFFIGURES .................................... 9 ABSTRACT ........................................ 11 CHAPTER 1INTRODUCTION .................................. 12 1.1IntensityModulatedRadiationTherapyTreatmentPlanning ........ 12 1.2DissertationSummary ............................. 13 1.2.1FluenceMapOptimization ....................... 13 1.2.2BeamOrientationOptimization .................... 14 1.2.3Fractionation .............................. 15 1.2.4ModelingtheDoseDepositionofaBeam ............... 15 1.3ContributionSummary ............................. 16 1.3.1Fluencemapoptimization ....................... 16 1.3.2BeamOrientationOptimization .................... 17 1.3.3Fractionation .............................. 19 1.3.4ModelingtheDoseDepositionofaBeam ............... 19 2FLUENCEMAPOPTIMIZATION ......................... 21 2.1Introduction ................................... 21 2.2LiteratureReview ................................ 21 2.3ModelFormulation ............................... 22 2.4SpatialConsiderations ............................. 24 2.5APrimal-DualInteriorPointAlgorithmforFMO .............. 25 2.5.1Primal-DualInteriorPointAlgorithm ................. 28 2.5.2HessianApproximations ........................ 29 2.5.2.1SingleHessianApproximation ................ 29 2.5.2.2BFGSHessianUpdate .................... 30 2.5.3InsignicantBeamlets .......................... 30 2.5.4WarmStart ............................... 31 2.6Results ...................................... 32 2.6.1HowSmallofaDualityGapisNecessary? .............. 33 2.6.2ComputationalResults ......................... 34 2.6.3ClinicalResults ............................. 36 2.6.4SpatialCoecientResults ....................... 37 2.6.5WarmStartResults ........................... 38 2.7Conclusions ................................... 45 5

PAGE 6

3BEAMORIENTATIONOPTIMIZATION ..................... 46 3.1Introduction ................................... 46 3.2LiteratureReview ................................ 47 3.3ModelFormulation ............................... 48 3.4Mixed-IntegerModelFormulation ....................... 50 3.5BeamDataGeneration ............................. 52 3.6AResponseSurfaceApproachtoBOO .................... 54 3.6.1OverviewofResponseSurfaces ..................... 55 3.6.2DeterminingtheNextObservation ................... 58 3.6.2.1Maximizingtheexpectedimprovement ........... 59 3.6.2.2Obtaininganupperboundontheuncertainty ....... 59 3.6.2.3Branch-and-Bound ...................... 61 3.6.3MethodofObtainingtheNextObservation .............. 69 3.7NeighborhoodSearch .............................. 69 3.7.1Introduction ............................... 69 3.7.2NeighborhoodSearchApproaches ................... 70 3.7.3ADeterministicNeighborhoodSearchMethodforBOO ....... 70 3.7.3.1NeighborhoodDenition ................... 71 3.7.3.2NeighborSelection ...................... 72 3.7.3.3Implementation ........................ 72 3.7.4SimulatedAnnealing .......................... 73 3.7.4.1NeighborhoodDenition ................... 75 3.7.4.2NeighborSelection ...................... 75 3.7.4.3Implementation ........................ 75 3.7.4.4Convergence .......................... 76 3.7.5ANewNeighborhoodStructure .................... 77 3.8Results ...................................... 79 3.8.1EvaluatingPlanQuality ........................ 79 3.8.1.1Targetcoverage ........................ 79 3.8.1.2Criticalstructuresparing .................. 80 3.8.2ResponseSurfaceMethodResults ................... 81 3.8.2.1Proofofconcept ....................... 83 3.8.2.2Addinganon-coplanarbeamtoacoplanarsolution .... 84 3.8.2.3Clinicalresults ........................ 85 3.8.3NeighborhoodSearchMethodResults ................. 88 3.8.3.1Add/Dropalgorithmresults ................. 89 3.8.3.2SimulatedAnnealingresults ................. 89 3.8.3.3Clinicalresults ........................ 91 3.9ConclusionsandFutureDirections ...................... 92 3.9.1ResponseSurfaceConclusions ..................... 92 3.9.2NeighborhoodSearchConclusions ................... 95 6

PAGE 7

4FRACTIONATION .................................. 96 4.1Introduction ................................... 96 4.2ModelFormulation ............................... 97 4.3Results ...................................... 100 4.3.1ComputationalResults ......................... 101 4.3.2ClinicalResults ............................. 102 4.3.3SpatialCoecientResults ....................... 103 4.4ConclusionsandFutureDirections ...................... 111 5AMONTECARLOMETHODFORMODELINGDOSEDEPOSITION .... 120 5.1Introduction ................................... 120 5.2MonteCarloEngine .............................. 121 5.3DoseDistributionofaBeamlet ........................ 121 5.3.1Depth-DoseCurve ............................ 122 5.3.2LateralPenumbra ............................ 123 5.4MethodologytoModelaBeamlet ....................... 124 5.4.1ModelingtheDepth-DoseCurve .................... 125 5.4.2ModelingtheLateralPenumbra .................... 128 5.5Results ...................................... 132 5.6ConclusionsandFutureDirections ...................... 138 REFERENCES ....................................... 147 BIOGRAPHICALSKETCH ................................ 148 7

PAGE 8

LISTOFTABLES Table page 2-1Averageruntimesfor5-beamtreatmentplans ................... 36 2-2FMOvalueobtainedusing=0:001 ........................ 36 2-3Comparisonofdualitygaps ............................. 37 2-4Performancemeasuresofinteriorpointmethodwarmstarts ........... 43 2-5Performancemeasuresofprojectedgradientmethodwarmstarts ......... 44 3-1Sparingcriteriavariesforeachcriticalstructure .................. 80 3-2Sizesoftestcases ................................... 82 3-3MinimumFMOvalueobtainedandtimerequiredtoobtainit .......... 86 3-4Targetcoverageachievedbythetreatmentplans .................. 86 3-5Percentageofplansinwhichanorganisspared .................. 87 3-6Denitionsofimplementations ............................ 91 4-1Casesizesandruntimesusingidenticalalgorithmandweightingparameters .. 102 4-2Sparingcriteriavariesforeachcriticalstructure .................. 103 5-1ComputationtimesinminutesofMonteCarlosimulations ............ 132 5-2Computationtimesfordosedistributionts .................... 134 5-3Variationofts .................................... 138 8

PAGE 9

LISTOFFIGURES Figure page 2-1Progressionofdualitygap .............................. 34 2-2Dosereceivedbytargetsasafunctionofthedualitygap ............. 35 2-3Dosereceivedbysalivaglandsasafunctionofthedualitygap .......... 35 2-4QualityofDVHsforvariousdualitygaps ...................... 37 2-5Thespatialcoecientsusedfortwocases ..................... 38 2-6Comparisonofspatialandnon-spatialtreatmentplans .............. 39 2-7Comparisonofspatialandnon-spatialtreatmentplans .............. 40 3-1Alinearacceleratorandtheavailablemovements ................. 46 3-2FMOvalueasafunctionoftwoangles ....................... 51 3-3Initialregions ..................................... 64 3-4Partitioningaregionintosubregions ........................ 67 3-5Accountingforsymmetry .............................. 68 3-6Theipneighborhood ................................ 78 3-7SelectionprobabilitiesinNhandNFh .................... 78 3-8Proofofconceptresults ............................... 84 3-9Comparisonofresponsesurface,Add/Dropandequi-spacedtargets ....... 87 3-10Comparisonofresponsesurface,Add/Dropandequi-spacedtargets ....... 88 3-11Add/DropandsimulatedannealingcomparisonofFMOconvergence ...... 90 3-12ComparisonofAdd/Dropand7-beamequi-spacedplans ............. 93 3-13Comparisonofsimulatedannealingand7-beamequi-spacedplans ........ 93 4-1TargetDVHs,salivaDVHsandaxialslicesinFractions1and2 ......... 104 4-2TargetDVHs,salivaDVHsandaxialslicesinFractions1and2 ......... 105 4-3TargetDVHs,salivaDVHsandaxialslicesinFractions1and2 ......... 106 4-4TargetDVHs,salivaDVHsandaxialslicesinFractions1and2 ......... 107 4-5TargetDVHs,salivaDVHsandaxialslicesinFractions1and2 ......... 108 9

PAGE 10

4-6TargetDVHs,salivaDVHsandaxialslicesinFractions1and2 ......... 109 4-7TargetDVHs,salivaDVHsandaxialslicesinFractions1and2 ......... 110 4-8DVHsandaxialslicesinFractions1and2usingspatialcoecients ....... 112 4-9DVHsandaxialslicesinFractions1and2usingspatialcoecients ....... 113 4-10DVHsandaxialslicesinFractions1and2usingspatialcoecients ....... 114 4-11DVHsandaxialslicesinFractions1and2usingspatialcoecients ....... 115 4-12DVHsandaxialslicesinFractions1and2usingspatialcoecients ....... 116 4-13DVHsandaxialslicesinFractions1and2usingspatialcoecients ....... 117 4-14DVHsandaxialslicesinFractions1and2usingspatialcoecients ....... 118 5-1Dosedistributionofasinglebeamletinvarioustissues .............. 122 5-2Colorwashofthelateralpenumbraofanitesizedpencilbeam ......... 124 5-3Plotofthelateralpenumbraofanitesizedpencilbeam ............. 125 5-4Observeddepth-dosecurveinwaterforseveralhistories .............. 126 5-5Polynomialtsofseveralhistories .......................... 127 5-6Variationofpolynomialtasfunctionofdegree .................. 128 5-7Anerrorfunctionandanerrorfunctionpair .................... 129 5-8LateralpenumbraforseveralnumbersofMonteCarlohistories .......... 130 5-9Errorfunctiontsofseveralhistories ........................ 131 5-10Errorfunctionpairssummedtoapproximateabeamletinwater ......... 135 5-11Depth-dosecurvesinmuscletissue. ......................... 135 5-12Lateralpenumbracurvesinmuscletissue. ..................... 136 5-13Depth-dosecurvesinlungtissue. .......................... 136 5-14Lateralpenumbracurvesinlungtissue. ....................... 137 5-15Depth-dosecurvesinheterogeneousmuscleandlungtissue. ........... 138 5-16Variationoftsasafunctionofnumberofhistories ................ 139 10

PAGE 11

AbstractofDissertationPresentedtotheGraduateSchooloftheUniversityofFloridainPartialFulllmentoftheRequirementsfortheDegreeofDoctorofPhilosophyOPTIMIZATIONMETHODSININTENSITYMODULATEDRADIATIONTHERAPYTREATMENTPLANNINGByDionneM.AlemanDecember2007Chair:H.EdwinRomeijnMajor:IndustrialandSystemsEngineeringThedesignofatreatmentplanforintensitymodulatedradiationtherapyamathematicalprogrammingproblemwhichisnotyetsatisfactorilysolved.Currenttechniquesincludedividingtheproblemintoseveralsubproblems,whicharethensolvedsequentially.Myresearchaddressesseveralofthesesubproblems,particularly,beamorientationoptimizationBOO,uencemapoptimizationFMOandfractionation.TheintegrationoftheBOOandFMOsubproblemsisconsidered,aswellasimprovedtechniquestomodelthedosedepositionofabeamlet. 11

PAGE 12

CHAPTER1INTRODUCTION1.1IntensityModulatedRadiationTherapyTreatmentPlanningEveryyear,approximately1.4millionpeopleintheUnitedStatesalonearenewlydiagnosedwithcancerAmericanCancerSociety,[ 1 ].MorethanhalfofthesepatientswillreceivesomeformofradiationtherapyMurphyetal.[ 2 ],PerezandBrady[ 3 ],andapproximatelyhalfofthesepatientsmaysignicantlybenetfromconformalradiationtherapySteel[ 4 ].Duringthistherapy,beamsofradiationpassthroughapatient,therebykillingbothcancerousandnormalcells.Althoughsomepatientsdieoftheirdiseasedespitesophisticatedtreatmentmethods,manypatientsmaysuerunpleasantsideeectsasaresultoftheradiationtherapywhichmayseverelydetractfromthepatient'squalityoflife.Thus,theradiationtreatmentmustbecarefullyplannedsothataclinicallyprescribeddoseisdeliveredtotargetscontainingcancerouscellssothatthecancerwillbeeradicated.Simultaneously,asmallenoughdosemustbedeliveredtothenearbyorgansandtissuescalledcriticalstructuressothattheymaysurvivethetreatment.Thisisachievedbyirradiatingthepatientusingseveralbeamssentatdierentorientationsspacedaroundthepatientsothattheintersectionofthesebeamsincludesthetargets,whichthusreceivethehighestradiationdose,whereasthecriticalstructuresreceiveradiationfromsome,butnotall,beamsandmaythusbespared.Currently,atechniquecalledintensitymodulatedradiationtherapyIMRTisconsideredtobethemosteectiveradiationtherapyformanyformsofcancer.TheproblemofdesigninganIMRTtreatmentplanforanindividualpatientisalarge-scalemathematicalprogrammingproblemthatisnotyetsolvedsatisfactorily.Currenttreatmentplanningsystemsdecomposetheplanningproblemintoseveralstages,andthecorrespondingsubproblemsaresolvedsequentially.Thesesubproblemsincludedeterminingthenumberandorientationofthebeamsofradiation,theradiationdose 12

PAGE 13

distributionofeachbeamandthedecompositionofasingletreatmentplanintoseveralsmallerfractions.ThisworkaddressestheintegrationofthebeamorientationoptimizationBOOanduencemapoptimizationFMOsubproblemsbasedonaconvexformulationofthelatterandassociatedecientalgorithmsforsolvingit,anapproachwhichhasnotreceivedmuchattentioninpreviousstudies.Thefractionationproblem,theproblemofdividingasingletreatmentplanintothe35treatmentsfractionsthepatientwillactuallyreceive,isalsoaddressed.Also,theproblemofmodelingthedosedepositionofabeamisalsoconsidered.1.2DissertationSummaryInIMRT,eachbeamismodeledasacollectionofhundredsofsmallbeamlets,theuencesofwhichcanbecontrolledindividually.Theseuencevaluesareknownasauencemap,andoptimizationoftheseuencesgivenaxedsetofbeamsisknownasuencemapoptimization.TheoptimalsolutionvalueoftheFMOproblemquantiesthequalityofthetreatmentplan,wherequalitymeanstheabilityoftheplantodelivertheprescribedradiationdosetothespeciedtargetstructureswhilesparingcriticalstructuresbyensuringthattheyreceiveanacceptablylowamountofradiation.Thus,thequalityofasetofbeamscanbemeasuredbytheoptimalsolutionoftheFMOproblemperformedwiththosebeams.Thus,theproblemofselectingthebestdirectionsfromwhichtodeliverradiationtothepatienttheBOOproblemisbasedonthetreatmentplanqualityindicatedbytheoptimalsolutionvaluetothecorrespondingFMOproblem.1.2.1FluenceMapOptimizationOneofthemostpopularsubproblemsoftheintensitymodulatedradiationtherapyIMRTtreatmentplanningproblemistheuencemapoptimizationFMOproblem.InIMRT,eachbeamofradiationcanbediscretizedinhundredsofsmallerbeamlets,theradiationintensitiesuencesofwhichcanbemodulatedindependentlyoftheotherbeamlets.Foragivensetofbeams,thebeamletuencescangreatlyinuencethequalityofthetreatmentplan,thatis,theabilityofthetreatmenttodeposittheprescibedamount 13

PAGE 14

ofdosetocanceroustargetstructureswhilesimultaneouslydeliveringasmallenoughdosetocriticalstructuressothattheymaycontinuetofunctionafterthetreatment.Theseuencevaluesareknownasauencemap,andoptimizationoftheseuencesgivenaxedsetofbeamsisknownasuencemapoptimization.Becausetheuencesofthebeamletscandrasticallyaectthequalityofthetreatmentplanitiscriticaltoobtaingooduencemapsforradiationdelivery.AstheFMOproblemisoneofthemostpopularsubproblemsinIMRToptimization,ithasbeenextensivelystudiedintheliterature.Severalproblemstructuresandalgorithmstosolvevariousmodelsarepresentedinmanystudies.1.2.2BeamOrientationOptimizationInatypicalhead-and-necktreatmentplan,radiationbeamsaredeliveredfrom5-9nominally-spacedcoplanarorientationsaroundthepatient.Thesecoplanarorientationsareobtainedfromrotatingthegantryonly.Severalcomponentsofalinearacceleratorcanrotateandtranslatetoachievemoreorientationsthanthoseobtainedfromrotatingthegantry.Theavailableorientationsconsistoftheorientationsobtainedfromrotationofthegantry,collimatorandcouch,aswellasthethreetranslationdirectionsofthecouch.BeamorientationoptimizationBOOistheproblemofselectingfromtheavailablebeamorientationsthebestsettouseindeliveringatreatmentplan.Givenaxedsetofbeams,dierentuencemapsradiationintensitiesofbeamletsyieldtreatmentplanswithdierentqualities.Therefore,thequalityofanoptimizeduencemapshouldbeconsideredwhenselectingasetofbeamorientationstouseinatreatmentplan.OptimaluencemapsmaybediculttoobtaindependingontheFMOmodel.Thus,itiscommonintheliteratureforscoringapproximationsandotherheuristicstobeusedtoestimatethequalityofabeamsolution.RegardlessoftheobjectivefunctionusedintheBOOproblem,theproblemisfundamentallynonlinearasthephysicsofdosedepositionchangewithdirection.Becausenonlinearprogrammingproblemsarediculttosolve,mostapproachestotheBOO 14

PAGE 15

problemrelyonglobalsearchalgorithmstoobtainasolution,whichmayormaynotbeoptimal.1.2.3FractionationAnimportantsubproblemrelatedtotheFMOproblemwhichhasnotyetreceivedmuchattentionisthefractionationproblem.Ratherthandeliveranentiretreatmentplaninonesession,atreatmentplanisdividedintoseveralsessions,calledfractions.Thisisdonetotakeadvantageofthefactthatnormal,healthycellsrecoverfasterfromtheradiationthancancerouscells.Toobtainthetreatmentplansforthefractions,inpractice,asingleFMOtreatmentplanisdevelopedandthendividedintothedesirednumberoffractions,usuallyaround35.Thisdivisionofatreatmentplanisanon-trivialtask,asthetargetvoxels,geometriccubesoftissue,mustreceive1.8-2.0Gyofradiationineachfraction.WithasingleIMRTtreatmentplan,itispracticallyimpossibletodeviseaconstantdose-per-fractiondeliverytechniquebecauseonlyasingleFMOproblemissolvedtoobtainthetreatmentplan,whichisthensimplydividedintoanumberofdailyfractions.Ifasingleplanisoptimizedtodeliverdosestomultipletarget-doselevels,thenthedoseperfractiondeliveredtoeachtargetmustchangeintheratioofagivendoseleveltothemaximumdoselevel.Forexample,sayPTV1hasaprescriptiondoseof70Gy,PTV2hasaprescriptiondoseof50Gy,andthenumberoffractionsis35.Ifasingletreatmentplanisdividedamongthe35fractions,thenPTV1willreceive70=35=2:0Gyineachfraction,butPTV2willonlyreceive50=35=1:4Gy,andthusanycancerouscellsinPTV2maynotbeeradicatedbythetreatment.Similarly,ifonly25fractionsareusedinordertoensurethatPTV2receives2.0Gyperfraction,thenPTV1receives70=25=2:8Gyperfraction,wellabovethedesireddose.1.2.4ModelingtheDoseDepositionofaBeamTheFMOproblemisarguablythemostsignicantindeterminingthequalityofthetreatmentplan.TheFMOproblemdependsheavilyonthecalculationofdose 15

PAGE 16

receivedineachvoxelofapatient.Thisdoseistypicallyapproximatedbyassumingalinearrelationshipwiththeradiationintensitiesofthebeamletsdeliveringtheradiation.Althoughthisapproximationisacceptedassatisfactory,itisnottrulyaccurate.Thedoseinavoxelisdeterminedbythepathsthephotonsintheradiationbeamsfollowthroughthepatient.Somephotonsmaycollidewithparticlesinsidethepatientandscatterinanydirection,whileothersmaycollidewithparticlesandbeabsorbed.Stillotherphotonsmaypassentirelythroughthepatientwithnocollisions.Duetotheunpredictablenatureoftheradiationbeaminsidethepatient,thedosereceivedinavoxelcanonlybeaccuratelyobtainedthroughMonteCarlosimulations.AsimplelinearrelationshipisassumedbetweentotaldoseandbeamletuencesandiscommonlyacceptedasasatisfactorydoseapproximationinIMRToptimization.Errorsofasmuchas30%havebeenreportedforphotonbeamsneartissueinhomogeneitiesMaetal.[ 5 ].ForIMRToptimization,particularlywithadventofimage-guidedIMRTIGIMRT,or4DIMRT,theFMOproblemmustbesolvedextremelyquicklytocreatereal-timetreatmentplans.Thus,thespeedoftheFMOproblemisparamount.LengthyMonteCarlosimulationcanprovideanaccuratemeasureofthedosedepositedinavoxel,butthistechniqueistimeintensiveandimpracticalforclinicaluseandparticularlyfortreatmentplanningoptimization.1.3ContributionSummary1.3.1FluencemapoptimizationNonlinearfunctionstoapproximatebiologicalbehavioranddesireddosedistributionsarecommoninthepreviouslyproposedFMOmodelsintheliterature,asaremixed-integerprogrammingmodels.Thesemodelscanbedicultandcomputationallyexpensivetosolve.TomaketheFMOproblemmoretractable,weemployamodelwithaconvexobjectivefunctionandlinearconstraints.Thisdesirablestructureallowsourmodeltobesolvedquicklyandtooptimalitywiththeprimal-dualinteriorpointalgorithmwehavedevelopedspecicallyforthisproblem. 16

PAGE 17

Oneofthegreatestbenetsofaninteriorpointalgorithmisthatagloballyoptimalsolutioncanbefoundformanyproblemstructures,andinparticular,convexproblemstructures.AsourFMOmodelisconvex,theinteriorpointalgorithmcanlocatethegloballyoptimalsolutiontowithinaspecieddualitygap.Whilethereareotheralgorithmsthatcantheoreticallyreturnagloballyoptimalsolutiontoaconvexproblemandmanyalgorithmsthatcannot,interiorpointmethodshavetheadvantageofprovidingaknowndualitygapandgenerallyfastcomputationtimes.Becausethedualitygapisknownineachiteration,theusercanmakeknowledgeabletrade-osbetweencomputationtimeandsolutionoptimalitywithouthavingtoguesshowfarfromtheoptimumthenalsolutionmaybe.ThisallowsforascienticcomparisonofdierentIMRTdeliverytechniquesaswecansolvethedierentproblemstoaspecicdualitygap.Severalalterationstothestandardprimal-dualinteriorpointmethodweremadetoimproveitsperformance.BeamletsthatarelikelytohavelittleornocontributiontothetreatmentplanareremovedapriorianddierentapproximationstotheobjectivefunctionHessianaretestedtosavetimeincalculatingthetrueHessianineachiteration.Theuseofwarmstartstoinitializetheinteriorpointmethodisalsoexamined.Thesolutionsobtainedprovidequalitytreatmentplansinaclinicallyfeasibleamountoftime.TheincorporationofspatialinformationintotheFMOmodelisalsoconsidered.Theprobabilityoftumormetastasisincreaseswithproximitytogrosstumormass.Byusingthedistancesofvoxelsfromtargetstructures,thevoxelscanbeweightedaccordingtotheirimportanceinthetreatmentplan.Forexample,itshouldbelessimportanttosparesalivaglandvoxelsnearatargetstructurethanitshouldbetosparesalivaglandvoxelsfarfromatarget.Theuseofspatialcoecientswillhelpthemodelidentifyqualitytreatmentplansthatwillpreventfuturemetastasis.1.3.2BeamOrientationOptimizationForhead-and-neckcancers,typicalIMRTtreatmentplansuse5-9equi-spacedcoplanarbeams.Coplanarbeamsarethosebeamsobtainedfromtherotationofonly 17

PAGE 18

thegantryofthelinearaccelerator,themachinewhichdeliversradiationbeamstothepatient.Ifallothercomponentsofthelinearacceleratorarexed,therotationofthegantrysweepsoutasetofcoplanarbeams.Thecouchcanrotateandtranslateinthreedimensions,andtheheadofthegantrycanrotateindependently,creatinganevenlargersetofbeams.Beamsobtainedfromthemovementofmorethanonecomponentfromthelinearacceleratorareknownasnon-coplanarbeams.Intuitively,onemayexpectthatthenumberofbeamsrequiredforahigh-qualitytreatmentplancanbereduced,orthequalityofthetreatmentplanforagivennumberofbeamscanbeimproved,ifthebeamorientationsarechosenoptimallyand/orfromalargerset.Inparticular,weinvestigatetheeectofconsideringmorecoplanarornon-coplanarbeams.Atreatmentplanconsistingoffewerbeamsispreferablebecausethenumberofbeamsusedinaplandirectlyaectsthelengthoftheactualtreatment.Iffewerbeamsareusedtotreatapatient,theneachtreatmenttakeslesstimeandmorepatientscanbetreatedinaday,whichisbenecialfrombothaclinicalandeconomicperspective.Longertreatmenttimesalsoallowformoreerrorsduetopossiblepatientmotion.WeviewtheBOOprobleminIMRTtreatmentplanningasaglobaloptimizationproblemwithexpensiveobjectivefunctionevaluations,eachofwhichinvolvessolvingaFMOproblem.Weproposearesponsesurfacemethodthat,unlikeotherapproaches,allowsforthegenerationofproblemdataonlyforpromisingbeamorientationson-the-yasthealgorithmprogresses,enablingtheconsiderationoffarmorecandidateorientationsthaniscurrentlyfeasible.OurresponsesurfaceapproachtoBOOallowsustodevelophighqualityplansusingjustfourbeamsforhead-and-neckcases,incontrasttothecurrentpracticeofusing5-9beams.Theresponsesurfacemethodalsoprovidesforconvergencetothegloballyoptimalsolution.WehavedevelopedneighborhoodsearchmethodstosolveourBOOmodel.Onemethodissimulatedannealing,aproperglobaloptimizationalgorithm,andtheother 18

PAGE 19

isalocalsearchheuristicdesignedspecicallyfortheBOOproblem.Thelocalsearchheuristic,whichwecalltheAdd/Dropmethod,returnsalocallyoptimalsolutioninasmallamountoftime.Thesimulatedannealingalgorithmhastheabilitytoescapelocalminima,andistheoreticallyabletoreturnagloballyoptimalsolutiongivenenoughtime.Foreachofthesealgorithms,wehavedevisedanewneighborhoodstructurebasedonobservationsofknownoptimalBOOsolutionscomparedtothesimulatedannealingandAdd/DropBOOsolutions.Thisnewneighborhoodstructureprovidesfasterobjectivefunctionvalueconvergenceinbothalgorithms.1.3.3FractionationInpractice,asingleFMOtreatmentplanisdevelopedandthendividedintothenumberofdesiredfractions.DividingasingleFMOintomultipletreatmentsisanon-trivialtask,owingtotheneedofmaintainingaconstantdose-per-fractiontoeachthetargetstructures,whichmayhavedierentprescriptiondoses.Therefore,anydivisionofasingleFMOplanintomultiplefractionscanleadtosuboptimaltreatments.Weproposeanewmethodofformulatingthefractionationproblemwhichyieldsoptimaluencemapsforeachcanceroustargetstructure.Theseuencemapscanthenbeeasilydividedintooptimalfractions.Theproposedfractionationmodelissolvedusingthesameprimal-dualinteriorpointmethodpresentedfortheFMOproblem.Thesolutionsprovidehighqualityuencemapsforeachtarget,andinaclinicallyacceptableamountoftime.1.3.4ModelingtheDoseDepositionofaBeamWeproposeobtainingalimitednumberofMonteCarlohistoriestoobtainanoisydosedistributionwhichcanthenbetransformedintoaveryaccurate,smoothdosedistributionsuitableforoptimizationtechniquesinareasonableamountoftime.Becausetheparticlesinabeamletscatterinthreedimensionalspace,multipledosedistributionsmustbeconsideredtosatisfactorilymodelthebeamlet'saectonthepatient'stissue.Thesedistributionsarisefromtheamountofradiationthebeamlet 19

PAGE 20

depositsasafunctionofdepththedepth-dosecurve,andfromtheamountofradiationradiatingoutwardfromthecenterofthebeamletthelateralpenumbra.Thedepth-dosecurveismodeledusingahigh-degreepolynomialandthelateralpenumbraismodeledasthesumoferrorfunctions.TheparametersoftheerrorfunctionsaredeterminedusingaLevenberg-Marquardtquasi-Newtonminimizationmethod.Usingthesetechniques,dosedistributionswithsatisfactoryaccuracycanbeobtainedusingatleastafactorof10fewerMonteCarlohistoriesthanwouldotherwiseberequired.ThiscangreatlydecreasetheamountoftimerequiredtoobtaindosedataforbeamletsintheFMOproblemofIMRTtreatmentplanningwithoutsacricingaccuracy. 20

PAGE 21

CHAPTER2FLUENCEMAPOPTIMIZATION2.1IntroductionIMRTisdierentiatedfromconformalradiationtherapybythedosedistributionsthatcanbedeliveredbyeachbeam.Ratherthatjustdeliveringauniformradiationeldofradiation,thedosedistributionofabeamcanbeanydesireddistribution.Thisabilityallowsforgreaterexibilityandaccuracyintargetingthetargetstructureswhileavoidingthecriticalstructures.Thedosedistributionofabeamisachievedasfollows.InIMRT,eachbeamcanbethoughtofasconsistingofseveralhundredsmallerbeamlets,eachofwhichcanhaveitsownradiationintensityuenceindependentofitsneighbors.Bymodulatingtheintensitiesofthesebeamlets,anydosedistributioncanbeachieved.Givenaxedsetofbeams,theoptimizationoftheseintensitiesiscalleduencemapoptimization.2.2LiteratureReviewBecausetheFMOproblemisoneofthemoststudiedproblemsofIMRT,manydierentapproacheshavebeentakentoformulatetheFMOproblem,basedonbothphysical"Bortfeld[ 6 ]andbiological"AlberandNusslin[ 7 ],JonesandHoban[ 8 ],Kallmanetal.[ 9 ],Mavroidisetal.[ 10 ],Niemierkoetal.[ 11 ],Niemierko[ 12 ],Wuetal.[ 13 14 ]objectivefunctionsandconstraints.LinearprogrammingLP-basedmulti-criteriaoptimizationHamacherandKufer[ 15 ]andmixed-integerlinearprogrammingMILPBednarzetal.[ 16 ],Ferrisetal.[ 17 ],Langeretal.[ 18 19 ],Leeetal.[ 20 21 ],Shepardetal.[ 22 ]modelshavebeenproposedforFMO.ConstraintstoenforcevariousmeasuresoftreatmentqualityarealsotakenintoaccountindierentFMOmodels.HamacherandKufer[ 15 ]includethehomogeneityofthedosereceivedbythetargetsasaconstraintintheirFMOmodel.Full-volumeconstraints,whichrequirethatthedoseineveryvoxelofastructurebewithinpre-determinedupperandlowerbounds,arecommonforcontrollingthedoseineachstructure.Models 21

PAGE 22

employingfull-volumeconstraintsarefoundinBednarzetal.[ 16 ],HamacherandKufer[ 15 ],Leeetal.[ 20 21 ],Romeijnetal.[ 23 ]andmanyothers.Modelscontainingpartialvolumeconstraints,constraintsrequiringthatdoseinonlyasubsetofvoxelsbewithinpre-determinedupperand/orlowerbounds,arealsocommon.FormulationswithpartialvolumeconstraintsarefoundinLeeetal.[ 20 21 ],Romeijnetal.[ 23 24 ]andShepardetal.[ 22 ].Inadditiontovaryingconstraints,therearemanycompetingmethodsofformulatingtheFMOobjectivefunctiontoreectthequalityofthetreatmentplan.Shepardetal.[ 22 ]describeseveraldierentobjectiveformulations.Theseformulationsincludeminimizingthesumofdosesreceivedatallvoxels;minimizingaweightedcombinationofdosesreceivedateachvoxel,wheretheweightsdependonthestructureinwhichthevoxelresides;andminimizingthedeviationofthedoseineachvoxelfromtherecommendedprescription.Romeijnetal.[ 25 ]showedthatmostofthetreatmentplanevaluationcriteriaproposedinthemedicalphysicsliteratureareequivalenttoconvexpenaltyfunctioncriteriawhenviewedasamulticriteriaoptimizationproblem.Foreachsetoftreatmentplanevaluationcriteriafromaverylargeclass,thereexistsaclassofconvexpenaltyfunctionsthatproducesanidenticalParetoecientfrontier.Therefore,aconvexpenaltyfunction-basedapproachtoevaluatingtreatmentplansisusedtoinvestigatetheBOOproblem.Althoughthisapproachcouldbeusedinamulticriteriasetting,Romeijnetal.[ 23 26 ]suggestthatitispossibletoquantifyatrade-obetweenthedierentevaluationcriteriathatproduceshigh-qualitytreatmentplansforapopulationofpatients,eliminatingtheneedtosolvetheFMOproblemasamulticriteriaoptimizationproblemforeachindividualpatient.2.3ModelFormulationAconvexpenaltyfunction-basedapproachtotheFMOmodelasdescribedinRomeijnetal.[ 23 ]isemployedtoquantifythequalityofthetreatmentplanbyappropriately 22

PAGE 23

makingthetrade-obetweendeliveringtheprescribedradiationdosetothetargetstructureswhilesparingthecriticalstructures.Usingthisapproach,theFMOproblemcanformulatedasaquadraticprogrammingproblemwithlinearconstraintsasfollows.DenotethesetofallpotentialbeamorientationsasB.Thestructuresbothtargetsandcriticalstructuresareirradiatedusingapredeterminedsetofbeamangles,denoted,whereeachbeamh2B,h=1;:::;kandkisthenumberofbeamsin.Eachbeamisdecomposedintoarectangulargridofbeamletswithmrowsandncolumns,yieldingtypically100-400beamletsperbeam.Thepositionandintensityofallbeamletsinabeamcanberepresentedbyavectorofvaluesrepresentingthebeamletintensities,calledbixels.ThesetofallbixelsinbeamhisdenotedbyBh.ThecoretaskinIMRTtreatmentplanningisndingradiationintensitiesforallbeamlets.DenotethetotalnumberofstructuresbySandthenumberoftargetsbyT.Eachstructuresisdiscretizedintoanitenumbervsofvolumecubes,knownasvoxels.Typically,around350,000voxelsarerequiredtoaccuratelyrepresentthetargetsandsurroundingstructuresofahead-and-neckcancersite.Becauseabeamletmustpassthroughacertainamountoftissuetoreachavoxel,thedosereceivedinavoxelfromabeamletmaynotbethefulldeliveredintensity.DenoteDijsasthedosereceivedbyvoxeljinstructuresfrombeamletiatunitintensity.TheDijsvaluesareknownasdosedepositioncoecients.Letxidenotetheintensityofbixeli.Thisbringsustothefollowingexpressionforthedosezjsreceivedbyvoxeljinstructures:zjs=kXh=1Xi2BhDijsxij=1;:::;vs;s=1;:::;SAlthoughthegoalofIMRTtreatmentplanningistocontrolthedosereceivedbyeachstructure,ifhardconstraintsareimposedontheamountofdosereceivedbyeachstructurebecausesuchasolutionmaynotexist.Insomecases,itmaybenecessarytosacriceorgansinordertotreattargets,andifthatpossibilityisnotallowedinthemodel,thenafeasibleorasatisfactorysolutionmaynotexist.Thus,inourmodel,apenaltyis 23

PAGE 24

assignedtoeachvoxelbasedonthedoseitreceivesforagivensetofbeamletintensities.LetFjsdenoteaconvexpenaltyfunctionforvoxeljinstructuresofthefollwingform:Fjszjs=1 vsw sTs)]TJ/F23 11.955 Tf 11.956 0 Td[(zjs+p s+ wszjs)]TJ/F23 11.955 Tf 11.955 0 Td[(Ts+ ps;whereTsisthedosethresholdvalueforstructures,w sandp sareweightingfactorsforunderdosing,and wsand psareweightingfactorsforoverdosing.Theexpression+denotesmaxf0;g.Thefunctionisnormalizedoverthenumberofvoxelsinthestructureusingthecoecient1=vs.Bysettingw s, ws0and ps,p s1,convexityisensured.AbasicformulationoftheFMOproblemisthen:minimizeSXs=1vsXj=1Fjszjssubjecttozjs=kXh=1Xi2BhDijsxij=1;:::;vs;s=1;:::;Sxi0i2Bh;h=1;:::;kTheFMOproblemistheblack-boxfunctionFintheBOOmodeltoquantifythequalityofbeamvector.IncontrastwiththemethodspresentedbyallofthepreviouslycitedFMOstudiesexceptforDasandMarks[ 27 ],Haasetal.[ 28 ]andSchreibmann[ 29 ],thismeasureofbeamvectorqualityisanexactmeasureoftheFMOproblem,ratherthanusingheuristicmethodsorscoringapproacheswhichcannotaccuratelyoptimizethebeamorientations.2.4SpatialConsiderationsWithIMRToptimization,itispossibletogeneratetreatmentplanswithsimilarFMOobjectivefunctionvaluesbutverydierentlevelsofclinicaltreatmentquality.Chaoetal.2003[ 30 ]illustratethispossibilitywithtwotreatmentplansthathavenearlyidenticaltargetcoveragewhenplottedonadose-volumehistogram,butwhileoneplandeliversanacceptablehomogeneousdose,theotherplanresultsinsignicantunderdosingofthetargetstructure. 24

PAGE 25

Chaoetal.2003[ 30 ]showthattheprobabilityofmicroscopictumorextensiondecreaseslinearlywithdistancefromthegrosstumorvolume,implyingthatcoldspotslocatednearthegrosstumorvolumearefarmorelikelytoallowfortumormetastasisaftertreatment.Likewise,coldspotslocatedfarfromthegrosstumorvolumeareunlikelytoresultintumormetastasis.Toreducethelikelihoodofobtaininganunsatisfactoryplanwithagooddose-volumehistograms,spatialcoecientsareintroducedintotheFMOmodel.Foreachvoxel,weconsideritspositionrelativetotheprimarytargetasameasureofhowacceptable/unacceptableoverdosingorunderdosingmaybe.Voxelsfurtherfromthegrosstumorvolumearepenalizedmoreheavilythanvoxelsclosertothegrosstumorbecauseitislessacceptableforavoxelfarawayfromtheactualtumortoreceiveanoverdose,asthecancerouscellsareunlikelytospreadveryfarfromthetumorlocationChaoetal.[ 30 ].Thisadditionalpenalizationiscalledthespatialcoecient,andisdenotedcjsforvoxeljinstructures.Forvoxelsinsidethetargetstructures,theprobabilityofcancerspreadis1,ascanceralreadyexistsinthosevoxels.LetS0denotethesetofgrosstumorstructures.Letd`jsbetheminimumdistancefromvoxeljinstructurestostructure`.Thespatialcoecientcjsforvoxeljinstructuresiscjs=8><>:1j=1;:::;vs;s=2S0minn1;maxn0:001;PjS0j`=1[exp)]TJ/F23 11.955 Tf 9.299 0 Td[(`d`js+`d`js+`]ooj=1;:::;vs;s2S0;where`,`and`areweightingcoecients.TheobjectivefunctionfortheFMOproblembecomesFspatialx=SXs=1vsXj=1cjsFjszjs2.5APrimal-DualInteriorPointAlgorithmforFMOTosolvetheFMOandfractionatedFMOmodels,aprimal-dualinteriorpointmethodisemployed.ForaconvexproblemsuchastheFMOmodelpresentedintheprecedingsection,thismethodyieldsanoptimalsolutioninshortamountoftime. 25

PAGE 26

Theprimal-dualinteriorpointalgorithmmovesthroughtheinteriorofthesolutionspacealongacentralpathapaththroughtheinteriorofthesolutionspacetowardtheoptimalsolution.ThecentralpathisdenedbyperturbingtheKKTconditionsdescribedbelow.Theseconditionsensureprimalfeasibility,dualfeasibilityandcomplementaryslackness.Iftheseconditionsaresatisedforaconvexprogrammingproblemwithlinearlyindependentconstraints,theyyieldtheoptimalsolution.Thus,weonlyneedtosolvethissystemtoobtainanoptimalsolutiontoourFMOmodelwhichhasaconvexobjectivefunctionandlinear,linearlyindependentconstraints.TheKKTsystemcanbediculttosolve,sotheconditionsareperturbedinordertoobtainasolution.Thegeneralideaoftheprimal-dualinteriorpointalgorithmistostartfromaninitialfeasiblesolution,usetheperturbedKKTconditionstoobtainastepdirectionclosetothecentralpath,andthenmovethecurrentsolutionsomesteplengthalongthatdirection.TheamountofpertubationintheKKTconditionsisgraduallydecreasedsothatineachstep,thesolutionbecomesclosertotheoptimum.Theinteriorpointmethodallowsforthedualitygap,thegapbetweentheobjectivefunctionsoftheprimalanddualproblems,tobecalculated,thusprovidingameasureofhowclosethecurrentsolutionistotheoptimum.Foraproblemwithcontinuousvariables,whentheobjectivefunctionsoftheprimalanddualproblemsareequaldualitygapofzero,thesolutionisoptimal.Amathematicaldescriptionoftheprimal-dualinteriorpointmethodcanbefoundinNocedalandWright[ 31 ].Furtherexplanationisprovidedonlyasneededtodenevariablesinthealgorithm.IntheFMOproblem,Gx=)]TJ/F39 11.955 Tf 9.299 0 Td[(Ix,sotheKKTconditionsfortheFMOformulationareXs2S1 vsXj2VsDijF0jX`2ND`jx`!)]TJ/F23 11.955 Tf 11.955 0 Td[(si=0i2N: {1 sixi=0i2N: {2 si0i2N {3 xi0i2N; {4 26

PAGE 27

wheretheEquation 2{4 ensuresthatthesolutionisfeasible,astheonlyconstraintsintheFMOproblemarenonnegativity.Thecomplimentaryslacknessconstraint 2{2 forcesthesolutiontotheaboveconditionstobeontheboundaryofthesolutionspace.Sinceapointintheinteriorofthesolutionspaceisdesired,thecomplimentaryslacknessconstraintmustberelaxed.Thecomplimentaryslacknessconstraint 2{2 isrelaxedbychangingeachsixi=0tosixi=,where>0.This,alongwithrequiringthatx>0ands>0forfeasibility,ensuresthatasolutiontotheperturbedKKTconditionsisaninteriorpoint.Letnbethesizeofdecisionvariablevectorx.Asolutioniscloseenough"tothecentralpathifthedualitymeasureiniterationkisk=xk>s n{5andjjXkSk)]TJ/F23 11.955 Tf 12.58 0 Td[(kejjk,whereXkisamatrixwithxkivaluesasdiagonalsandzeroselsewhere,andSkisamatrixwithskivaluesasdiagonalsandzeroselsewhere.Asthealgorithmprogresses,isreducedtozerountilthesolutionissucientlyclosetooptimality.Toreduce,ineachiterationweset=,where2[0;1]iscalledthecenteringparameter.Ifthedualitygapisverylarge,canbereducedsothatisreducedfaster.Ineachiteration,thecurrentsolutionx;sismovedinadirectionx;sforsomesteplengthisgivenby264xk+1sk+1375=264xksk375+264xksk375LetXk=diagxk,Sk=diagsk,Hxk=r2xk.Thedirectionsxkandskcanbedeterminedbysolvingthefollowingequations:h)]TJ/F39 11.955 Tf 5.48 -9.684 Td[(Xk)]TJ/F22 7.97 Tf 6.587 0 Td[(1Sk+Hxkixk=)]TJ/F39 11.955 Tf 9.299 0 Td[(rDF)]TJ/F28 11.955 Tf 11.955 9.684 Td[()]TJ/F39 11.955 Tf 5.479 -9.684 Td[(Xk)]TJ/F22 7.97 Tf 6.587 0 Td[(1rxs {6 sk=)]TJ/F28 11.955 Tf 11.291 9.684 Td[()]TJ/F39 11.955 Tf 5.48 -9.684 Td[(Xk)]TJ/F22 7.97 Tf 6.587 0 Td[(1)]TJ/F39 11.955 Tf 5.479 -9.684 Td[(rxs+Skxk {7 27

PAGE 28

Inordertosolvethissystem,wemustobtainxkfromEquation 2{6 bytakingtheinverseof[Xk)]TJ/F22 7.97 Tf 6.587 0 Td[(1Sk+H].Becausecomputingtheinverseofsuchalargedensematrixisverytimeconsuming,aCholeskyfactorizationtosolvethissystemquickly.Theprimal-dualinteriorpointmethodrequiresafeasiblex;ssolutionineachstep.Thus,amaximumsteplengthmaxmustbeimposedoneachstepdirectiontoensurethatx0ands0:max=minmini=1;:::;nf)]TJ/F23 11.955 Tf 15.276 0 Td[(xi=xig;mini=1;:::;nf)]TJ/F23 11.955 Tf 15.276 0 Td[(si=sigBecausetheinverseofeachxiisrequiredtodeterminethestepdirections,itisundesirabletohaveanyxi=0,whichwouldresultfromusingsteplengthmax.Instead,onlyapercentage<1ofmaxisused:=minf1;maxg{8Thebenetofthisprimal-dualmethodisthatineachstep,wecancalculatetheobjectiveofthedualproblemsimplys>x,thusprovidingaboundonhowfarthecurrentsolutionisfromoptimality.2.5.1Primal-DualInteriorPointAlgorithmTheprimal-dualinteriorpointalgorithmisasfollows: Initialization 1. Selectinitialvaluesfor,andweuse=5,=0:01,and=0:95. 2. Setx0=0:05verycloseto0andcalculaterx0andHx0=r2x0. 3. Sets0=X0)]TJ/F22 7.97 Tf 6.587 0 Td[(1. 4. Set0=Pni=1rx0i=100. 5. Setk=0. Algorithm 1. Ifthedualitygapisverylargexk+1>sk+1>107,set=0:01. 2. Setk=k. 28

PAGE 29

3. Solveforthestepdirectionxk;skasdescribedinEquations 2{6 and 2{7 .NotethatthisinvolvescalculatingtheHessianHxk. 4. SolveforthesteplengthasdescribedinEquation 2{8 5. Setxk+1=xk+xandsk+1=sk+s. 6. Ifthedualitygapxk+1>sk+1<,stop.Otherwise,setk+1=xk+1>sk+1=nandkk+1andrepeat.2.5.2HessianApproximationsThemosttime-consumingstepintheprimal-dualinteriorpointalgorithmiscalculatingtheHessianoftheobjectivefunctionineachiteration.Forclarity,letPdenotePs2S1=vsPj2VsandF00jxdenoteF00jPl2NDljxl.TheHessianoftheFMOproblemisthengivenbyHx=266664PF00jxD21j:::PF00jxD1jDnj.........PF00jxDnjD1j:::PF00jxD2nj377775NotethatonlythepairwiseDijproductsdierineachelementoftheHessian.Byprecomputingthesecrossproducts,onlyPs2S1=vsPj2VsF00jPl2NDljxlhastoberecomputedineachiteration.ThematrixoftheDijproductsyieldsthesparsityordensitypatternoftheHessian,whichstaysconstantthroughoutthealgorithm.BecausetheHessianissymmetric,thematrixvaluesonlyneedtobecomputedforhalfofthematrix,furtherimprovingeciency.Despitetheseobservations,computingtheHessianisstillsoexpensivethatitrendersthealgorithmimpractical.MethodsofapproximatingtheHessianareimplementedtospeedupthealgorithm.2.5.2.1SingleHessianApproximationOnewayofspeedingupthealgorithmistocomputetheHessianjustonceduringinitializationtoobtainHx0,andthenratherthanre-computetheHessianineachiteration,useHx0asanapproximationtoHxk.WecallthistheSingleHessian 29

PAGE 30

approximation.Althoughtheconvergenceofsuchanapproximationhasnotyetbeenmathematicallyproven,testsrunonseveralhead-and-neckcasesfor5-beamand7-beamplansshowthattheSingleHessiandoesinfactconvergetotheknownoptimalsolution.2.5.2.2BFGSHessianUpdateAnotherHessianapproximationistheBroyden-Fletcher-Goldfarb-ShannoBFGSHessianupdate.TheapproximationtotheHessianiniterationkisBk,withB0=Hx0.TheupdatetotheapproximatedHessianineachiterationisBk+1=Bk+qkq>k q>kpk)]TJ/F39 11.955 Tf 13.15 8.088 Td[(Bkpkp>kBk p>kBkpk;wherepk=xk+1)]TJ/F39 11.955 Tf 11.955 0 Td[(xkqk=rxk+1)-222(rxkNotethatthisupdateensuresthatBkisalwayssymmetricandpositivedenite,sotheCholeskyfactorizationcanstillbeappliedtoobtainthestepdirection.Thisapproximationalsoempiricallyconvergestotheknownoptimalsolutionfor5-and7-beamhead-and-neckcases.2.5.3InsignicantBeamletsInsignicantbeamletsarethosethatbearlittlecontributiontothequalityoftheFMOplan.LettingddenotethediagonalelementsoftheinitialHessianHx0,thesetofinsignicantbeamletsBIisdenedasBI=i:jdij maxfjdjg<0:001ThesebeamletsareremovedbyremovingtheithrowandtheithcolumninHx0foreveryi2BI,andthenupdatingthenumberofbixelstothenumberofremainingbixels.Theinsignicantbeamletsmustbere-insertedintothesolutionxkinordertocalculate 30

PAGE 31

thevoxeldoses,objectivefunction,gradientandHessian,buttheinversionoftheHessianisdonetotheHessianwiththebadbeamletsremoved,providingsignicanttimesavings.2.5.4WarmStartForthesakeoftheoreticalaccuracy,atrulyoptimalsolutioncannothavethebadbeamletsdescribedinSection 2.5.3 removed.Withoutremovingthebadbeamletsapriori,theinteriorpointmethodmustberunforanimpracticalamountoftimetoobtainanear-optimalsolution,say,=0:001.Theinteriorpointmethodistypicallystartedwithadecisionvariablevectorxequaltoalmostzero.Ifthealgorithmweretobestartedatapointclosertothenalsolution,denotedxwarm,timesavingscouldbegained,allowingallbeamletstobeconsideredintheinteriorpointalgorithminareasonableamountoftime.Suchanapproachisacalledawarmstart".Onedicultyinusingawarmstartwiththeinteriorpointmethodisthatawarmstartsolutionmayhavesomexwarmi=0,whichisnotallowedbecausetheinverseofeachximustbetaken.Tocorrectthisproblem,anyxwarmi=0issimplyreplacedwithsomeverysmallvalue.Becausethesezero-valuedvariablesarelessimportanttotheproblemthannonzerovariables,shouldbelessthantheminimumnonzerovalueofxwarm.Let=mini=1;:::;nfxwarmi:xwarmi>0g.Then,=minf0:001;g.x0i=8><>:xwarmii=2BIi2BIAnadditionalproblemwithwarmstartsintheinteriorpointmethodisthattheKKTvariablevectorsisunknownatthewarmstartpoint.Dependingonthealgorithmusedtoobtainthewarmstart,someinformationaboutswarmandwarm,sandatthewarmstartpoint,respectively,maynotbeavailable.Ifnoinformationisavailableaboutsfromthewarmstart,thens0=0.Ifaninteriorpointalgorithmisusedtoobtainthewarmstart,thenswarmisavailable.Ifthewarmstartdidnotincludetheinsignicantbeamlets,somecorrectionsmustbemadetoaccountfortheinsignicantbeamletswhichwillbe 31

PAGE 32

optimizedinthenalsolution.Lets0betheinitialsusedintheinteriorpointmethodafterthewarmstarthasbeenobtained.Then,s0i=8><>:swarmii=2BIwarm=i2BI;wherethevaluechosenfors0icorrespondingtoinsigncantbeamletsarisesfromthegeneralinitializations=X0)]TJ/F22 7.97 Tf 6.587 0 Td[(1.2.6ResultsThetrueHessian,SingleHessianapproximation,andBFGSupdateimplementationsoftheprimal-dualinteriorpointalgorithmaretestedonsixcaseshead-and-neckcasestoobtaincoplanar,equi-spaced5-beamplans.Thetestsarerunona2.33GHzIntelCore2Duoprocessorwith2GBofRAM.Themethodistestedforbothleavinginandremovingtheinsignicantbeamlets,aswellastheproposedalternativetocomputingtheHessian.TheoptimalityoftheinteriorpointmethodsolutionsisveriedbycomparisontotheknownoptimalsolutionsobtainedbyJavawithCPLEXILOG.Anacceptabledualitygapmustbedeterminedinordertoimplementtheinteriorpointmethod.Whileweconsideradualitygapof=0:001tobeacceptablyclosetooptimal,itmaybeunnecessarytoachievesuchasmalldualitygaptoobtainaqualitysolution.Adualitygapof0.001maybesucientlysmalltoensureoptimalsolutionsgivenobjectivefunctionvaluesusingcertainweightingparameters,dependingontheparametersusedintheFMOobjectivefunction,thevalueoftheobjectivefunctionmayvarywidely.Becauseofthepotentialrangeofvalues,astoppingcriteriabasedonarelativedualitygapratherthananabsolutedualitygapispreferable.Saytheobjectivefunctionvalueinaniterationisf.Denetherelativedualitygapinaniterationtobe0==f.AnexaminationoftherelativedualitygapnecessaryispresentedinSection 2.6.1 .ComputationalresultsarepresentedinSection 2.6.2 andclinicalcomparisonsareprovidedinSection 2.6.3 32

PAGE 33

2.6.1HowSmallofaDualityGapisNecessary?Becausetheruntimeofthealgorithmisdependentontherequireddualitygap,itisdesirabletoonlyrequirethealgorithmtoachieveassmalladualitygapasnecessarytoensureaclincallygoodsolution.Thedualitygapdecreasesquicklyintherstfewiterations,andthensubsequentlydecreasesbyonlyasmallamountperiteration,asshowninFigure 2-1 A.Iftheseiterationswithonlymarginalimprovementsarefoundtobeunnecessaryintermsofclinicalquality,signicanttimecanbesavedbystoppingthealgorithmoncethedualitygapisreasonablysmall,asopposedtowaitinguntilthedualitygapisverysmall.Tochecktheimportanceofthedualitygap,theFMOvalueanddosedeliveredtothetargetsandthesalivaglandswereplottedagainstthedualitygapineachiterationusingthetrueHessianandwithoutremovinginsignicantbeamlets.Forarepresentativecase,theFMOvaluesperdualitygapareshowninFigure 2-1 B.Itisclearthatthedualitygapdecreasesrapidlyintherstfewiterations,butsubsequentiterationsyieldincreasinglysmallerdropsinthedualitygap.Similarly,theamountofdosereceivedbythetargetsandcriticalstructuresdoesnotchangesignicantlytowardtheendofthealgorithm.Figure 2-2 plotsthedosereceivedbythetwotargets,PTV1andPTV2,startingfromadualitygapof0.15%.Theprescriptiondosesare70GyforPTV1and50GyforPTV2,commondosevaluesusedinthecancerclinicatShandsHospitalattheUniversityofFlorida.Neitherthedosereceivedby95%ofthetargetsnorthesizeofthehotspotsandcoldspotschangessignicantlyinthisdualitygaprangeFigure 2-2 A.Thehotspotsaremeasuredbythepercentofthetargetreceiving110%and120%oftheprescriptiondose,whilethecoldspotsaremeasuredbythepercentofthetargetreceivingatleast93%oftheprescriptiondoseFigure 2-2 B.Figure 2-3 showsfortworepresentativecasestheamountofdosereceivedbythesalivaglandsstartingfromarelativedualitygapof0.15%.Bothcasesshowthatthe 33

PAGE 34

Figure2-1.Thedualitygapdropssharplyinearlyiterations,butveryslowlythereafter.Therelativedualitygapmonotonicallydecreasesafterseveraliterations. changeindosereceivedbythesalivaglandsasthedualitygapdecreasesisnotclinicallyrelevant.Fromthesegures,itappearsthatadualitygapaslargeas0.1%couldprovideclinicallyacceptableplans.Sincethealgorithmmayterminatewithadualitygaplessthantheonespeciedasthestoppingcriteria,adualitygaplargerthan0.1%willalsobetestedforacceptability.2.6.2ComputationalResultsTable 4-1 showstheaverageruntimesforeachoftheimplementationsofthealgorithm.Relativedualitygapsof0.15%,0.10%,0.05%and0.01%.arecompared.Thevalueofusedtodenethecentralpathis0.5.Asexpected,usingtheSingleApproximationHessianalternativewiththeinsignicantbeamletsremovedisthefastestmethod,whileusingthetrueHessianistheslowestmethod,regardlessofwhethertheinsignicantbeamletsareremoved.Interestingly,forlargedualitygaps,itisslightlyfastertoleavetheinsignicantbeamletsinthemodelwhenusingthetrueHessian.Otherwise,itisfastertoremovetheinsignicantbeamlets.ThenalFMOvaluesaredisplayedforeachofthetestedmethodsusingadualitygapof0.001,whichissucientlysmalltoensureoptimalsolutionsgiventypicalobjectivefunctionvaluesTable 2-2 .Foreachcase,thenalFMOvalueisnearlyidentical, 34

PAGE 35

ABFigure2-2.Dosereceivedbytargetsasafunctionofthedualitygap.ATheamountofdosereceivedbyatleast95%ofeachtargetisusedtoassesspropertargetcoverage.BThepercentofeachtargetreceiving110%and120%oftheprescriptiondoseindicateshotspots,while93%oftheprescriptiondoseindicatescoldspots. Figure2-3.Theamountofdosereceivedbyatleast50%ofeachsalivaglandremainsrelativelyconstantevenforlargedualitygaps.Tworepresentativecasesareshown. 35

PAGE 36

Table2-1.Averageruntimesfor5-beamtreatmentplans. Removeinsig.AverageruntimesHessiantypebeamlets?=0:0010=0:150=0:10=0:050=0:01 Trueno113.855.4855.4858.5871.75Trueyes105.655.2556.2959.0970.56BFGSno43.913.5914.1714.6616.67BFGSyes40.913.1913.6614.3015.88SingleApprox.no18.18.838.989.2910.13SingleApprox.yes16.88.698.849.149.90 Table2-2.FMOvaluefromusing=0:001. Removeinsig.Hessiantypebeamlets?Case1Case2Case3Case4Case5Case6 TrueHessianno2546.222200.702289.952566.385024.972585.40TrueHessianyes2546.222200.702289.952566.385024.972585.40BFGSupdateno2546.232200.702289.952566.395024.972585.40BFGSupdateyes2546.242200.702289.952566.395024.972585.40SingleApprox.no2546.382201.112290.402566.565025.062585.82SingleApprox.yes2546.382201.152290.442566.625025.142585.82 indicatingthattheHessianalternativesandtheremovaloftheinsignicantbeamletsstillprovideforconvergencetotheoptimalsolution.ThepercentageincreasesintheFMOvaluesusinganabsolutedualitygapof0.001andrelativedualitygapsof0.15%,0.10%,0.05%and0.01%areshowninTable 2-3 .2.6.3ClinicalResultsForeachofthedualitygapstested,theDVHsofthesolutionsobtainedusingtheSingleApproximationHessianwiththeinsignicantbeamletsremovedarecompared.Sincetheeachoftheinteriorpointimplementationsobtainsnearlyidenticalsolutions,itdoesnotmatterwhichimplementationisusedtoproducetheDVHs.Aspreviouslystated,theprescriptiondosesusedare70GyforPTV1and50GyforPTV2,markedbyaverticallineinFigure 2-4 A.Assalivaglandsarethemostdicultorganstospareinhead-and-neckcases,theonlycriticalstructuresshownarethesalivaglandsFigure 2-4 B.Allotherglandsaresparedineveryimplementation.Thesparingcriteriausedforsalivaglandsisthatnomorethan50%percentofthesalivaglandcan 36

PAGE 37

Table2-3.Percentincreaseinobjectivefunctionvaluefromvariousrelativedualitygapsasopposedtoanabsolutedualitygapof=0:001. Removeinsig.Avg.increaseinobj.fn.%Hessiantypebeamlets?0=0:150=0:10=0:050=0:01 Trueno0.580.580.270.05Trueyes0.580.480.250.06BFGSno0.990.540.300.05BFGSyes0.940.570.260.07SingleApprox.no1.260.890.600.19SingleApprox.yes1.210.870.570.16 ABFigure2-4.QualityofDVHsfordualitygaps0=0.01%,0.05%,0.1%and0.15%.AThetargetcoverageisnearlyidentical.BThesalivaglandsparingforthedierentdualitygapsissimilar,butthesolutionfor0=0.15%sacricesonesalivagland.Thesparingcriteriaismarkedbyastar. receivemorethan30Gyinordertobespared.ThispointismarkedbyastarinFigure 2-4 B.Eachofthedualitygapsachievesgoodtargetcoverage.Whiletheyeachprovidesimilarsalivaglanddosage,theplanobtainedusing0=0:15%slightlysurpassesthesparingcriteriausedforsalivaglands.2.6.4SpatialCoecientResultsToassessthepossibletreatmentplanimprovementaordedbyspatialcoecients,spatialparametersweretunedandthencomparedtotreatmentplansobtainedwithoutusingspatialinformation.Todemonstratethespatialcoecients,Figure 2-5 displaysthe 37

PAGE 38

Figure2-5.Thespatialcoecientsusedfortwocases. coecientsusedfortwocases.Inadditiontotuning,andtovaluesof1.07,-0.32and0.77,respectively,aminimumspatialcoecientof0.025wasalsosetfortargetvoxels.Bydenition,themaximumvalueofaspatialcoecientis1.Thesespatialparametersgenerallyproducetreatmentplansofnearlyidenticalqualitytothebestplansobtainedwithoutusingspatialinformation,thoughwiththeaddedbenetofpreventingmisleadingdose-volumehistograms.Insomecases,thespatialcoecientswereabletooutperformthenon-spatialplans.Figures 2-6 and 2-7 illustratestwosuchcases.InFigure 2-6 ,thespatialcoecientsyieldimprovedtargetcoverageandspareallsalivaglands,asopposedtothenon-spatialplanwhichonlysparesthreeofthefoursalivaglands.Thereislessdoseoutsidethedesiredtargetintheplanusingspatialcoecients.InFigure 2-7 ,thespatialcoecientsreducetheamountofoverdoseintheprimarytargets.Inthispatient,boththespatialandnon-spatialplansspareallsalivaglands.2.6.5WarmStartResultsWarmstartsolutionswereobtainedusingtheinteriorpointmethodandtheprojectedgradientalgorithmNocedalandWright[ 31 ].TheinteriorpointmethodwarmstartsweretestedwitheachHessianpossibilityandalargedualitygapof200,bothwithandwithoutinsignicantbeamletsremoved.Theprojectedgradientalgorithmwastestedusing 38

PAGE 39

ABFigure2-6.Comparisonofspatialandnon-spatialtreatmentplans.ANon-spatialparametersresultinslightlylowtargetdosageandfailtospareonesalivagland.BSpatialparametersallowforimprovedtargetcoverageandspareallsalivaglands. 39

PAGE 40

ABFigure2-7.ANon-spatialparametersresultinslightlylowtargetdosageandfailtospareonesalivagland.BSpatialparametersallowforimprovedtargetcoverageandspareallsalivaglands. 40

PAGE 41

severalstoppingcriteriaandwithoutinsignicantbeamletsremoved.Itwasobservedthattheprojectedgradientalgorithmisfastenoughthatthetimerequiredtoremoveandre-inserttheinsignicantbeamletsasnecessarycausedthealgorithmtoslowdown.Tobetheoreticallyclosetooptimal,theinteriorpointmethodusedafterthewarmstarthasdualitygapof0.001andnobeamletsremoved.Thedeterminethehowclosethewarmstartsolutionistothenalsolution,thepercentimprovementinobjectivefunctionvaluethenalsolutionobtainsoverthewarmstartismeasured.Toassesshowclosetooptimalitythenalsolutionsusingawarmstartare,thepercentagebywhichtheirobjectivefunctionvaluesaregreaterthantheobjectivefunctionvalueofanear-optimalsolutionismeasured.Lastly,thedecreaseinruntimesoverobtaininganear-optimalsolutionsareprovided.TheseresultsfortheinteriorpointandprojectedgradientwarmstartsaredisplayedinTables 2-4 and 2-5 ,respectively.FromTable 2-4 ,itisclearthatusinganinteriorpointwarmstartcanprovidesignicanttimesavingsoverthenear-optimalsolutiontimes.ThereisalsoasignicantincreaseintheFMOobjectivefunctionvalue.Fromtheamountofincreaseintheobjectivefunctionvalue,theinteriorpointwarmstartdoesnotappeartoconvergetotheoptimalsolution,andisunlikelytoprovideacceptablesolutions.Itisinterestingtonotethattheimprovementfromthewarmstartsolutiontothenalsolutionisverysmall.ThisindicatesthatKKTinformationobtainedfromthewarmstartandusedinthenalalgorithmwereunhelpfulinimprovingthesolution.Fortheprojectedgradientalgorithm,oncethereislessthanpercentdecreasesfromoneiterationtothenext,thealgorithmterminates.Severalvaluesaretested.Aswiththeinteriorpointwarmstarts,theprojectedgradientwarmstartsalsoprovidedsignicanttimesavings,asshowninTable 2-5 .Thenalsolutionsfromtheprojectedgradientwarmstartmethodsarenearlyidenticaltothenear-optimalsolutions.Thenalinteriorpointmethodalsosignicantlyimprovestheobjectivevalueofthewarmstartsolution.ThisimpliesthatdespitenothavingKKTinformationaboutthewarmstart,theinteriorpoint 41

PAGE 42

algorithmisstillabletoconvergetotheoptimal,orataleastnear-optimal,solutionusingtheKKTvalueapproximationsandadjustmentstothewarmstartvectordescribedinSection 2.5.4 42

PAGE 43

Table2-4.Performancemeasuresofinteriorpointmethodwarmstarts. interiorpointwarmstartnalinteriorpointalgorithmImprovementIncreaseRemoveinsig.Removeinsig.overwarmstartinnalAvg.timeHessiantypebeamlets?0Hessiantypebeamlets?0obj.fn.%obj.fn.%savingss Trueno5trueno0.010.004.4664.75Trueyes5trueno0.010.194.4865.20Trueno5BFGSno0.010.004.7927.94Trueyes5BFGSno0.010.204.8428.47Trueno5SingleApprox.no0.010.005.066.85Trueyes5SingleApprox.no0.010.764.496.93BFGSno5trueno0.010.004.4664.97BFGSyes5trueno0.010.194.4865.09BFGSno5BFGSno0.010.004.7927.83BFGSyes5BFGSno0.010.204.8428.55BFGSno5SingleApprox.no0.010.005.066.90BFGSyes5SingleApprox.no0.010.764.496.87SingleApprox.no5trueno0.010.004.4664.99SingleApprox.yes5trueno0.010.194.4865.00SingleApprox.no5BFGSno0.010.004.7927.95SingleApprox.yes5BFGSno0.010.204.8428.54SingleApprox.no5SingleApprox.no0.010.005.066.88SingleApprox.yes5SingleApprox.no0.010.764.496.88 43

PAGE 44

Table2-5.Performancemeasuresofprojectedgradientmethodwarmstarts. interiorpointwarmstartnalinteriorpointalgorithmImprovementIncreaseRemoveinsig.Removeinsig.overwarmstartinnalAvg.timebeamlets?Hessiantypebeamlets?0obj.fn.%obj.fn.%savingss no1Trueno0.0119.830.0036.63no5Trueno0.0131.780.0010.98no10Trueno0.0136.430.0019.16no100Trueno0.0156.590.0139.28no500Trueno0.0189.460.0956.88no1BFGSno0.0119.830.009.27no5BFGSno0.0131.780.0012.53no10BFGSno0.0136.430.0019.30no100BFGSno0.0156.590.0327.79no500BFGSno0.0189.460.1330.30no1SingleApprox.no0.0119.820.003.40no5SingleApprox.no0.0131.770.013.95no10SingleApprox.no0.0136.420.014.28no100SingleApprox.no0.0156.560.089.28no500SingleApprox.no0.0189.440.2710.04 44

PAGE 45

2.7ConclusionsTheprimal-dualinteriorpointmethodisaneectivealgorithmforobtaininguencemapsthatdeliverqualitytreatmentplans.TheproposedHessianalternativesappeartoconvergetotheoptimalsolution,evenwheninsignicantbeamletsareremoved.Theremovaloftheinsignicantbeamletsprovidessignicanttimesavingsinallinstances.Theinteriorpointmethodmayalsoberunwithadualitygapaslargeas20andstillachievequalitytreatmentplans,thusdecreasingtheamountoftimerequiredtorunthealgorithm.Oftheimplementationstested,thefastestmethodthatstillprovidesqualitysolutionswithoutusingawarmstartistousetheSingleApproximationHessianalternative,removeinsignicantbeamletsandemployarelativedualitygapof0.1%.Whentheinteriorpointmethodisstartedwithoneofthewarmstartsdiscussed,timesavingswereagainsignicant.Althoughtheinteriorpointwarmstartsgenerallyprovidedmoreimprovementincomputationtimethantheprojectgradientwarmstarts,thenalsolutionsusingtheprojectedgradientwarmstartsweremuchclosertooptimality.Thefastestandmosteectivewarmstartmethodistousetheprojectedgradientalgorithmwith=500,followedbytheinteriorpointmethodwith=0:1%andtheSingleApproximationHessian.Thiscombinationresultsinanear-optimalsolutionwithanaveragetotalcomputationtimeof8.32seconds. 45

PAGE 46

CHAPTER3BEAMORIENTATIONOPTIMIZATION3.1IntroductionInatypicalhead-and-necktreatmentplan,radiationbeamsaredeliveredfrom5-9nominally-spacedcoplanarorientationsaroundthepatient.Thesecoplanarorientationsareobtainedfromrotatingthegantryonly.AsshowninFigure 3-1 ,severalcomponentsofalinearacceleratorcanrotateandtranslatetoachievemoreorientationsthanthoseobtainedfromrotatingthegantry.Theavailableorientationsconsistoftheorientationsobtainedfromrotationofthegantry,collimatorandcouch,aswellasthethreetranslationdirectionsofthecouch. Figure3-1.Alinearacceleratorandtheavailablemovements;thegantryrotationishighlighted. BOOistheproblemofselectingfromtheavailablebeamorientationsthebestsettouseindeliveringatreatmentplan.Givenaxedsetofbeams,dierentuencemapsradiationintensitiesofbeamletsyieldtreatmentplanswithdierentqualities.Thus,thequalityofanoptimizeduencemapshouldbeconsideredwhenselectingasetofbeamorientationstouseinatreatmentplan. 46

PAGE 47

3.2LiteratureReviewManyapproacheshavebeentakentosolvetheBOOproblem.EvolutionaryalgorithmsSchreibmann[ 29 ]andvariantsofevolutionaryalgorithms,particularlygeneticalgorithmsEzzell[ 32 ],Haasetal.[ 28 ],Lietal.[ 33 ]havebeenemployed.Lietal.[ 34 ]useaparticleswarmoptimizationmethod,whichisconceptuallybasedonevolutionaryalgorithms.BortfeldandSchlegel[ 35 ],Djajaputraetal.[ 36 ],Luetal.[ 37 ],PugachevandXing[ 38 ],Rowbottometal.[ 39 ]andSteinetal.[ 40 ]haveallemployedvariationsofsimulatedannealingtodetermineabeamsolution.SoderstromandBrahme[ 41 ]selectedcoplanarbeamorientationsusingtwomeasures,entropyandtheintegralofthelowfrequencypartoftheFouriertransformoftheoptimalbeamproles,bothofwhicharebasedonthesizeandshapeofthetargetstructure.SoderstromandBrahme[ 42 ]alsouseaniterativetechniquetodeterminetheoptimalnumberofcoplanarbeamsrequiredusingBOO.DasandMarks[ 27 ]useaquasi-Newtonmethod.Rowbottometal.[ 43 ]usearticialneuralnetworkalgorithmstoselectbeamorietations.Gokhaleetal.[ 44 ]useameasureofeachbeam'spathofleastresistance"fromthepatientsurfacetothetargetlocationtodeterminethebestbeamdirections.Meedtetal.[ 45 ]useafastexhaustivesearchtoobtainanon-coplanarsolution.Theconceptofbeam's-eyeviewBEVhasalsobeencommonlyusedtoapproachtheBOOproblemChenetal.[ 46 ],Choetal.[ 47 ],Goiteinetal.[ 48 ],Luetal.[ 37 ],PugachevandXing[ 38 49 50 ].Despitethevaryingtechniquestoquantifythequalityofabeamsolution,itiswidelyacceptedthattheoptimalsolutiontotheFMOproblempresentsthemostrelevantmeasureBortfeldandSchlegel[ 35 ],Djajaputraetal.[ 36 ],HolderandSalter[ 51 ],Leeetal.[ 20 21 ],Lietal.[ 33 34 ],Meedtetal.[ 45 ],Morrilletal.[ 52 ],Oldhametal.[ 53 ],Rowbottometal.[ 39 43 54 ],Schreibmannetal.[ 29 ],SoderstromandBrahme[ 41 ],Steinetal.[ 40 ],Wangetal.[ 55 56 ],WoudstraandHeijman[ 57 ].Giventhisacceptedmeasureoftreatmentquality,theshortcomingofthepreviousworksistwofold.First,theypredominantlyonlyconsidercoplanarangles,andnotnecessarilyeventheentire 47

PAGE 48

coplanarsolutionspace,whilethosethatdoconsidernon-coplanarbeamsonlyconsiderahand-selectedsubsetoftheavailableorientations.Second,themajorityofthepreviousstudiesdonotselectbeamsolutionsusingtheFMOproblemasamodelfordeterminingquality;instead,thebeamsolutionsarechosenbasedonscoringmethodse.g.,BEV,pathofleastresistanceorapproximationstotheFMO.BynotoptimizingthebeamsolutionwithrespecttotheexactFMOproblem,theBOOmethodscannotguaranteeconvergencetoanoptimalsolution.Ofthepreviouslycitedworks,onlyDasandMarks[ 27 ],Gokhaleetal.[ 44 ],Meedtetal.[ 45 ],Luetal.[ 37 ],Rowbottometal.[ 39 ]andWangetal.[ 56 ]considernon-coplanarorientations.Thisislikelyduetothecomputationaldicultiesassociatedwiththeinclusionofnon-coplanarorientationsaswellasthewidespreadbeliefthatnon-coplanarorientationsdonotimprovethequalityofatreatmentplan.Also,ofthoseworksthataddressednon-coplanarbeams,DasandMarks[ 27 ]requirethatthebeamdistancesbemaximized,essentiallyrequiringthatbeamsolutionsmustbeequi-distantandthusrestrictingthesizeofthesolutionspace;Meedtetal.[ 45 ]onlyconsider3,500beamsaminutesubsetoforientationsavailablebyrotationofthecouchandthegantry;andWangetal.[ 56 ]useonlyninepre-selectednon-coplanarbeams.WiththeexceptionofDasandMarks[ 27 ],Haasetal.[ 28 ]andSchreibmann[ 29 ],thepreviousstudieshavebasedtheirBOOapproachesnotonabeamsolution'soptimalsolutiontotheFMOproblem,butonlocallyoptimalFMOsolutionsoronvariousscoringtechniques.WithoutbasingBOOontheoptimalFMOsolutions,theresultingbeamsolutionshavenoguaranteeofoptimality,orevenoflocaloptimality.3.3ModelFormulationThegoalofradiationtherapytreatmentplanningistodesignatreatmentplanthatdeliversaprescribedlevelofradiationdosetothetargetswhilesimultaneouslysparingcriticalstructuresbyensuringthatthelevelofradiationdosereceivedbythesestructuresislessthanastructure-specicradiationdose.Thesetwogoalsarecontradictoryifthe 48

PAGE 49

targetsarelocatednearcriticalstructures.Thisisespeciallyproblematicforcertaincancers,suchastumorsinthehead-and-neckarea,whichareoftenlocatedverycloseto,forinstance,thespinalcord,brainstemandsalivaryglands.InordertomodeltheBOOproblem,aquantitativemeasurethatappropriatelymakestrade-osbetweenthesecontradictorygoalsmustbedeveloped.LetFbeablack-boxfunctionthatquantiesthequalityofthetreatmentplanifradiationisdeliveredfrombeamvector=1;:::;k,wherekistheuser-speciednumberoforientationsthatmaybeused.Fisformulatedinsuchawaythattheoptimalplanyieldstheminimumfunctionvalue.Forkbeamsorientationstobeoptimizedinthetreatmentplan,thevectorofdecisionvariablesrepresentingthebeamorientationsisdenedas=1;:::;kT.Thedecisionvectorisusedasinputintotheblack-boxfunctionFtodeterminetheabilityofthebeamvectortodelivertheprescribedtreatmentwithoutundulydamagingnormaltissueandcriticalstructures.TheBOOproblemisthenformulatedasminFsubjecttoh2Bh=1;:::;k;whereBisthesetofcandidatebeams.Thecandidatesetofbeamscanbeselectedaccordingtoanyuser-speciedcriteria;forexample,thebeamscanbecoplanarornon-coplanar,continuousordiscrete,oronlyrepresentasubsetoftheavailablebeams.Itisalsopossibletoxsomebeamsandonlyoptimizeasubsetofthetotalnumberofbeamstobeused.Theoretically,thelinearacceleratorisabletocaptureacontinuoussetoforientations,butduetomachinetolerances,theactualbeamsdeliveredmaynotbeexactlythedesiredbeams.Therefore,itiscommontoonlyconsideradiscretizedsetofbeamorientations.InourBOOmodel,theblack-boxfunctionFistheconvexFMOproblemdescribedinSection 2.3 ,thusensuringanexactmeasureofthequalityofeachbeamvector.EventhoughFisconvex,thisformulationoftheBOOproblemisfundamentally 49

PAGE 50

nonlinearbecausethephysicsofdosedepositionchangewitheachbeamorientation;thatis,theeectofabeamoneachpatientcanbedrasticallydierentthantheeectofaneighboringbeam.Toillustratethenonlinearityoftheproblem,Figure 3-2 showstheFMOproblemasafunctionofjusttwocoplanarbeamangles.Fromthisillustration,itisevidentthattheFMOfunction,particularlyinhigher,morerealisticdimensions,islikelytoalsobemulti-modal.AlthoughtheFMOproblemitselfcanbesolvedquicklyusingtheconvexmodelpresentedinSection 2.3 ,inordertoperformtheFMO,lengthycalculationsmustbemadeinordertodetermineeachcandidatebeam'seectonthepatient.Thesecalculations,describedinSection 3.5 ,require13minutesperbeamtocalculate,andthusmakeeachevaluationoftheFMOproblemexpensive.Despitethetimerequiredforeachfunctionevaluation,thelimitingfactorinbeamorientationoptimizationistheharddrivespacerequiredtostorethebeamdataforeachcandidatebeam.Ifthecandidatesetofbeamsissmall,thisdatacanbepre-computedandstored,allowingtheFMOproblemtobesolvedquicklyintheBOOproblem.But,ifthecandidatesetofbeamsislarge|forexample,consistingofnon-coplanarorientations|thenthedatacannotbepre-computedduetostoragerequirements.BecauseofthesedicultieswiththeBOOproblem,previousstudieshavebeenlargelyunabletoconsidertheentiresolutionspaceofavailablebeams.Byusingtheresponsemethod,whichisspecicallydesignedtomodelexpensivenonlinearblack-boxfunctions,wecaniterativelyidentifypromisingbeamvectorsolutionsandgeneratebeamdataforthesesolutionson-the-y,thuscircumventingtheissueofstoragespaceandallowingfortheconsiderationofalldeliverablebeamorientations.3.4Mixed-IntegerModelFormulationAsanalternativetotheBOOmodelgiveninSection 3.3 ,ifthesetofbeamorientationsBisnite,theBOOandFMOproblemscanbeformulatedtogetherandsolvedsimultaneouslyasamixed-integerlinearornonlinearprogramD'Souzaetal.[ 58 ], 50

PAGE 51

Figure3-2.FMOvalueasafunctionoftwoangles. EhrgottandJohnston[ 59 ],Ferrisetal.[ 17 ],Leeetal.[ 20 21 ],Limetal.[ 60 ],Shepardetal.[ 22 ],Wangetal.[ 61 ].TheFMOformulationcanbecombinedwithBOOinthefollowingmodel.Letybeabinaryvariableindicatingwhetherornotbeam2Bisused.Ifbeamisusedinthetreatmentplan,thenallthebeamletsin,B,areturnedon";thatis,theycanhavepositiveuencesuptosomepre-determinedmaximumintensityM.ThesimultaneousBOO+FMOMIPmodelisthenminimizeFzsubjecttozjs=kXh=1Xi2BkDijsxij=1;:::;vs;s=1;:::;SxiMyi2B;2BX2Bykxi0i2B;2By2f0;1g2B 51

PAGE 52

Inordertosolvesuchaproblem,allbeamdatamustbepre-computedforeverybeamorientation.AsdescribedinSection 3.5 ,beamdatarequiresatremendousamountoftimeandspacetocomputeandstore.Becauseofthisrequirement,onlyasmallsubsetofallpossiblebeamorientationscanbeconsideredduetotimeandspaceconstraintsforaBOO+FMOMIPformulation.3.5BeamDataGenerationForeachbeamorientationthatisconsidered,lengthycalculationsmustbemadetodeterminethebeam'seectonthepatient'stissueandorgans.Thisincludesdetermininginwhichstructureeachvoxellies,whichvoxelsarehitbywhichbeamletsandtheamountofintensityofeachbeamletisdepositedineachvoxelthroughwhichitpasses.BeamletdosecomputationmodelsusedinIMRTrelyheavilyonray-tracingalgorithmsforvoxelclassicationanddeterminationoftheradiologicalpathFoxetal.[ 62 ].VoxelclassicationSiddon[ 63 ]establisheswhethervoxelsareinsideoroutsidethepathofaradiationbeamandclassiesvoxelcentersasinsideoroutsideofsegmentedtargetsandcriticalstructures.Theradiologicalpathistheeectivedistancetraveledbyabeamletwhentheeectoftravelingthroughtissuesofdierentdensitiesisconsidered.TheexactradiologicalpathofabeamletthroughthepatientisrequiredtocorrectfortissueheterogeneitiesindeterminingthedosedepositioncoecientsSiddon[ 64 ].Siddon'sray-tracingalgorithmsSiddon[ 63 64 ]havebeenthestandardmethodsusedforray-tracinginradiotherapysincethe1980s.InSiddon'spolygonandvoxelray-tracingalgorithmsforvoxelclassicationpoint-in-polygontesting,structuresarerepresentedas3Dpolygonalobjects,knownasSiddonPrisms,andthesignsofcross-productsofrayspassingthroughthepolygonsareusedtodeterminewhetheravoxelliesinsideoroutsideastructure.Despiteitsoverwhelminguse,Siddon'salgorithmforpolygonray-tracingbecomesverycostlyduetothenumberofvoxelsinapatient.Foxetal.[ 62 ]developedanovelapproachtopolygonray-tracingthatcircumventstheneedforcross-productsbytranslatingthepolygonstructureontoacoordinatesystem,replacing 52

PAGE 53

theneedforacross-productbythesignofthesecondcoordinateofeachvoxelinthecoordinatesystem.InSiddon'salgorithmfordeterminingradiologicalpathsSiddon[ 64 ],theradiologicalpathmustbedeterminedforeachvoxelforeverybeamlet.Thisinvolvescomputationsformillionsofbeamlet-voxelcombinations.AsreportedbyJacobsetal.[ 65 ]asignicantamountofcomputationaltimeisrequiredfortheserepeatedcalculations.Foxetal.[ 62 ]combinetheincrementalvoxelray-tracingalgorithmpresentedbyJacobsetal.[ 65 ]withamethodofvirtualstereographicprojectiontosignicantlyreducethecomputationalcostofobtainingradiologicalpathlengths.Usingtheirpolygontranslationandincrementalray-tracingalgorithms,Foxetal.[ 62 ]achievea100-300foldimprovementincomputationtimeoverSiddon'spoint-in-polygonalgorithm.Becauseofthesignicantreductionincomputationtime,thesemethodsareusedtogeneratebeamdata.Becausethesebeamdatacalculationsmustbeperformedforeachofmillionsofbeamlet-voxelcombinations,beamdatagenerationisalengthyprocess,requiring13minutesperbeamusingthealgorithmsdescribedbyFoxetal.[ 62 ].InatypicalFMOformulation,thebeamvectorispre-determinedandthebeamdataforthebeamvectoriscalculatedonceandstoredapriori.Fora5-beamcase,thisrequires150MBofspacetostore.AswithatypicalFMOproblem,inasimultaneousFMO+BOOmixed-integerprogrammingMIPformulation,beamdataforeachofthecandidatebeamsinBmustbegeneratedapriori.Ifcandidatebeamsareconsideredonlyforcoplanaranglesona10grid,thatis,onlyevery10thangle,beamdatawouldhavetobecomputedfor36beams,whichrequires5hourstocomputeand800MBofspacetostore.Ifwealsowantedtoconsiderthepossibilityofrotatingthecouchona10gridinadditiontothegantry,beamdatawouldthenhavetobecomputedfor362beams,whichwouldrequire170hoursand60GBofspaceforjustoneplan. 53

PAGE 54

Clearly,thestoragespacerequirementsforeachbeamrestrictsthenumberofbeamsthatcanbeconsideredinasimultaneousFMO+BOOMIPformulation.ThisissueistypicallyaddressedbysimplyrestrictingthenumberofcandidatebeamsinB.Leeetal.[ 20 ]restrictthesetBtoonlycontain18pre-selectedbeamorientations,whichcanbecoplanarornon-coplanar.Ifonlygantryandcouchrotationsareallowedona10grid,abeamsetof18beamscomprisesonlyasmallpercentoftheavailablebeamorientations.Asmorerangesofmotionareallowed,thispercentagefallsevenfurther.Theinclusionofallpossiblebeamorientationssignicantlyincreasesthesizeofthesolutionspaceandcouldpossiblyallowforimprovedtreatmentplans,butthebeamdataforallorientationscannotbepre-computed.Inordertoconsidertheseorientations,weuseamethodthatallowsustogeneratethebeamdataon-the-yonlyasnecessary.3.6AResponseSurfaceApproachtoBOOTheshortcomingofthepreviousworksonBOOistwofold.First,theypredominantlyonlyconsidercoplanarangles,andnotnecessarilyeventheentirecoplanarsolutionspace,whilethosethatdoconsidernon-coplanarbeamsonlyconsiderahand-selectedsubsetoftheavailableorientations.Second,themajorityofthepreviousstudiesdonotselectbeamsolutionsusingtheFMOproblemasamodelfordeterminingquality;instead,thebeamsolutionsarechosenbasedonscoringmethodse.g.,BEV,pathofleastresistanceorapproximationstotheFMO.BynotoptimizingthebeamsolutionwithrespecttotheexactFMOproblem,theBOOmethodscannotguaranteeconvergencetoanoptimalsolution.Ofthepreviouslycitedworks,onlyDasandMarks[ 27 ],Gokhaleetal.[ 44 ],Meedtetal.[ 45 ],Luetal.[ 37 ],Rowbottometal.[ 39 ]andWangetal.[ 56 ]considernon-coplanarorientations.Oftheseworks,DasandMarks[ 27 ]requirethatthebeamdistancesbemaximized,essentiallyrequiringthatbeamsolutionsmustbeequi-distantandthusrestrictingthesizeofthesolutionspace;Meedtetal.[ 45 ]onlyconsider3,500beamsa 54

PAGE 55

minutesubsetoforientationsavailablebyrotationofthecouchandthegantry;andWangetal.[ 56 ]useonlyninepre-selectednon-coplanarbeams.WiththeexceptionofDasandMarks[ 27 ],Haasetal.[ 28 ]andSchreibmann[ 29 ],thepreviousstudieshavebasedtheirBOOapproachesnotonabeamsolution'soptimalsolutiontotheFMOproblem,butonlocallyoptimalFMOsolutionsoronvariousscoringtechniques.WithoutbasingBOOontheoptimalFMOsolutions,theresultingbeamsolutionshavenoguaranteeofoptimality,orevenoflocaloptimality.Becausebeamdatagenerationiscostly,amethodthatiterativelyidentiesonlypromisingbeamorientationsisrequired.TheresponsesurfaceRSmethodissuchanalgorithm.Incontrasttothepreviousstudies,ourapproachtotheBOOproblemallowsfortheinclusionofallpossiblebeamorientationswhicharemeasuredaccordingtotheexactFMOproblem,thusensuringconvergencetooptimalityduetothepropertiesoftheresponsesurfacemethod.TheRSmethodisdesignedtoecientlymodelexpensiveblack-boxfunctions.Inthisapplication,theFMOsolverisourblackboxandthesetofbeamstobeusedistheinput.AsinAlemanetal.[ 66 67 ],weemploytheresponsesurfacemethodasdetailedinJones[ 68 ]andJonesetal.[ 69 ].3.6.1OverviewofResponseSurfacesTheresponsesurfacemethodidentiespromisingsolutionsbasedontheperformanceofprevioussolutions.Thefunctionvalueandexpectedimprovementoverthecurrentbestsolutionofacertainpointisestimatedbasedonthefunctionbehaviorlearnedfrompreviouslysampledpointsandtheircalculatedobjectivefunctionvalues.Thefunctionvaluesofpointsarerelatedbycorrelationfunctionsthatdependoneachpoint'sdistancefromthepreviouslysampledpoints.Fromthecorrelationfunctions,thealgorithmpredictstheprobabilitythatthebestsolutionwillimproveatunexploredpointsinthesolutionspace.Usingthisprobability,apromisingsolutionisidentied.FortheBOOproblem, 55

PAGE 56

beamdataonlyneedstobegeneratedforthesepromisingsolutions,thussavingbothcomputationtimeandstoragespace.TheresponsesurfacemethodmodelstheobjectivefunctionasastochasticprocessoftheformF=+;{1whereisaconstantrepresentinganaverageofthefunctionFandisarandomerrortermassociatedwiththepoint.Inthegeneralcase,theerrortermsbetweentwopoints,sayand,arecorrelatedbyCorr;=exph)]TJ/F23 11.955 Tf 9.299 0 Td[(d;i;{2whered;isaweighteddistancemeasurebetweenand.Intuitively,iftwopointsareveryclosetogether,thecorrelationbetweenthemwillbeclosetoone;similarly,iftwopointsareveryfarapart,thecorrelationbetweenthemwillapproachzero.Jonesetal.[ 69 ]proposethefollowingweighteddistancemeasureingeneral:d;=kXh=1chh)]TJ/F23 11.955 Tf 11.955 0 Td[(hph;wheretheparameterschandphareweightingfactorscorrespondingtotheimportanceofeachvariablehandthesmoothnessofthefunctionFinthedirectionofvariableh,respectively.IfsmallchangesinvariablehcauselargechangesinthefunctionF,thenchshouldbelargetoreectthattwopointswithrelativelysmalldierencesinthevalueofvariablehshouldbefar"apartduetothelargedierenceintheirfunctionvalues,andthushavealowcorrelation.Theparameterchcantakeonanyvalue,whereas1ph2,withph=2correspondingtoobjectivefunctionsmoothnessandph=1correspondingtolessobjectivefunctionsmoothness.IntheapplicationtoBOO,=1;:::;kisthevectorofkanglesfromwhichradiationwillbedelivered.Becausenobeamismoreimportantthananotherbeam,eachbeamorientationhcontributesequallytotheFMOfunction,soch=candph=pfor 56

PAGE 57

allh=1;:::;k.Tomaintaintractabilityofthesubproblemsdescribedinthefollowingsections,theanglesaretreatedasthoughtheyarepointsonalineratherthanpointsonacircleandsoaEuclideandistancemetricisusedtodeterminethedistancebetweentwopoints.TheweighteddistancemeasureforBOOisthend;=c)]TJ/F42 11.955 Tf 11.956 0 Td[(pp;{3wherekkpdenotesthe`p-norm.ToensuretractabilityofthesubproblemsdescribedinSection 3.6.2 ,thevaluep=2isused.TheideaoftheRSmethodistoiterativelyevaluatethetruefunctionFatcertainbeamvectors,andthenconstructtheconditionalstochasticprocessgiventhesefunctionvalues.ThisconditionalstochasticprocessisthenusedtodecidewheretoevaluatethefunctionFnext.DuetothetimeandspacerequiredtogeneratethebeamdatanecessarytoevaluatethefunctionF,itisdesirabletoonlyevaluatepointsthatwilleitherimprovethebestsolutionwithasignicantprobabilityorsignicantlyincreaseourknowledgeofthefunction.TheoptimizationmodelstodeterminethenextobservationaredescribedinSection 3.6.2 .Let;:::;nbenpreviouslysampledpoints.Rnisthematrixofcorrelationsbetweenthepreviouslysampledpoints,ynisthevectoroffunctionvaluesFiofthepreviouslysampledpointsand^nand^nbeestimatorsoftheaverageandvarianceofthefunctionF,respectively.Theresponsesurfacealgorithmisgivenby: Initialization: 1. Choosevaluesfortheparameterscandp. 2. Chooseaninitialsamplesize,n,andasetofanglesi,i=1;:::;n.EvaluatethefunctionFateachofthesepoints,yieldingthevaluesyi,i=1;:::;n. Iteration: 1. ComputeorupdatethevaluesofRn,R)]TJ/F22 7.97 Tf 6.586 0 Td[(1n,^n,^n,andF n,theminimumobservedobjectivefunctionvalue. 57

PAGE 58

2. DeterminethenextpointtoobserveusingoneofthemethodsdescribedinSection 3.6.2 andcallthispointn+1. 3. Findthevalueyn+1=Fn+1,setnn+1,andrepeat.3.6.2DeterminingtheNextObservationBecausethefunctionFisexpensivetoevaluate,wewanttosampleasfewpointsaspossible.Thus,ineachiteration,anoptimizationproblemissolvedthatdeterminesthebest"nextpointatwhichtoobservethetruefunctionF.Someoftheoptimizationproblemsthathavebeenproposedintheliteraturedependontheuncertaintyofthepredictorasafunctionof,aswellastheexpectedimprovementoverthecurrentbestsolutionJones[ 68 ],Jonesetal.[ 69 ].Letrnbethevectorcorrelationsbetweenandthenpreviouslysampledpoints.Theuncertaintyisthengivenbys2n=^2n"1)]TJ/F39 11.955 Tf 11.955 0 Td[(rn>R)]TJ/F22 7.97 Tf 6.587 0 Td[(1nrn+1)]TJ/F39 11.955 Tf 11.956 0 Td[(1>R)]TJ/F22 7.97 Tf 6.586 0 Td[(1nrn2 1>R)]TJ/F22 7.97 Tf 6.587 0 Td[(1n1#;where^2n=1 nyn)]TJ/F39 11.955 Tf 11.955 0 Td[(1^n>R)]TJ/F22 7.97 Tf 6.587 0 Td[(1nyn)]TJ/F39 11.955 Tf 11.955 0 Td[(1^nistheestimatorofthevariance2nbasedonthenobservations.Theexpectedimprovement,denotedIn,isgivenbyIn=sn[zz+z]{4wherez=F n)]TJ/F15 11.955 Tf 14.606 3.022 Td[(^Fn sn!{5andF n=minfy1;:::;yngisthecurrentbestsolutionand^Fnistheestimatedfunctionvalueofgiventhenpreviouslysampledpoints.andarethec.d.f.andp.d.f.ofastandardnormalrandomvariable,respectively. 58

PAGE 59

Theselectionofthenextpointwillbebasedonselectingthepointthatmaximizeseithertheuncertaintyortheexpectedimprovement,oracombinationofboth.Denotethebeamvectortobechosenasthevector.3.6.2.1MaximizingtheexpectedimprovementJones[ 68 ]andJonesetal.[ 69 ]recommendselectingthenextpointtosampleasthepointforwhichtheexpectedimprovementoverthecurrentbestsolutionvalue,In,islargest.Thiscorrespondstosolvingthefollowingoptimizationproblem:maxInsubjecttoh2Bh=1;:::;kAlthoughthisisadicultoptimizationproblem,itcanbesolvedusingabranch-and-boundtechnique,butinordertodoso,anupperboundonInmustbeobtained.Thiscanbedonebysolvingfortheexpectedimprovementinequation 3{4 whilesubstitutinganupperboundontheuncertaintyandalowerboundon^Fn,usedinequation 3{5 todeterminethevaluez.Themethodofbounding^FnistakendirectlyfromJones[ 68 ]andJonesetal.[ 69 ]andisnotdiscussedfurtherhere.Themethodofboundings2nisimprovedfromtheoriginalformulationinJonesetal.[ 69 ]toovercomenumericalinstabilities,andispresentedinSection 3.6.2.2 .Thebranch-and-boundalgorithmusedtomaximizeInisdescribedinSection 3.6.2.3 .3.6.2.2ObtaininganupperboundontheuncertaintyDuetothecomplexityofthes2nfunction,maximizingtheuncertaintyisadicultproblemtosolve.ItcanberelaxedintoalinearlyconstrainedquadraticprogrammingproblemasfollowsJonesetal.[ 69 ].TheresultingsolutiontotherelaxeduncertaintymaximizationproblemisanupperboundontheuncertaintythatcanbeusedindetermininganupperboundonInasdescribedinSection 3.6.2.1 .Letr=fr1;:::;rng,whererisavectorofdecisionvariablesindependentof.Bytreatingbothrandasdecisionvariables,aquadraticobjectivefunctionisobtained. 59

PAGE 60

Becauserisnowadecisionvariableindependentof,anequalityconstraintmustbeaddedtotheproblemtoensurethatrassumesthecorrectcorrelationvaluesaccordingtothecorrelationdenitioninequation 3{2 .Thisconstraintisnonlinear,butitcanberelaxedbyexpressingthesingleequalityastwoinequalitiesandandthenreplacingthenonlineartermsgeneratedbylnriandck)]TJ/F42 11.955 Tf 12.881 0 Td[(ik22withlinearunderestimatorsai+biriandpi;h+qi;hh,respectively.Thedierenttypesoflinearestimatorsrequiredierentvaluesforai,bi,pi;handqi;h,andaredierentiatedbyasuperscriptcforthechordunderestimatorsandasuperscripttforthetangentlineunderestimatorsinthemodelformulation,denotedProblems2-UB.Unfortunately,thisrelaxationprovidedbyJonesetal.[ 69 ]canbecomenumericallyunstableiftwosampledpointsareveryclosetogether.Ifsuchasituationarises,theboundsofthecorrespondingcorrelationvaluecanbecomesoclosethatduetoround-oerror,thelowerboundrLicanbecomeslightlylargerthantheupperboundrUi,resultingininfeasibility.Toavoidsuchaninstability,insteadofboundingriusingconstraints,theamountbywhichriisoutsideofitsfeasiblerangeispenalizedbyaddingpenalizationtermswLi=minf0;ri)]TJ/F23 11.955 Tf 12.239 0 Td[(rLigandwUi=minf0;rUi)]TJ/F23 11.955 Tf 12.239 0 Td[(rig.ThisnalformulationisgiveninProblems2-UB.Thisformulationhasonlytwomorevariablesandtwomoreconstraintsforeachsampledpoint,sotheincreasedproblemsizedoesnotsignicantlyincreasetheamountoftimerequiredtosolvetheproblem. 60

PAGE 61

PROBLEMs2-UB:Chooserandtomin)]TJ/F15 11.955 Tf 12.57 0 Td[(^2n"1)]TJ/F39 11.955 Tf 11.955 0 Td[(r>R)]TJ/F22 7.97 Tf 6.586 0 Td[(1nr+1)]TJ/F39 11.955 Tf 11.955 0 Td[(1>R)]TJ/F22 7.97 Tf 6.586 0 Td[(1nr2 1>R)]TJ/F22 7.97 Tf 6.586 0 Td[(1n1#+nXi=1)]TJ/F23 11.955 Tf 5.479 -9.684 Td[(wLi2+nXi=1)]TJ/F23 11.955 Tf 5.48 -9.684 Td[(wUi2subjecttoaci+bciri+ckXh=1)]TJ/F23 11.955 Tf 5.48 -9.684 Td[(pti;h+qti;hh0i=1;:::;n)]TJ/F23 11.955 Tf 5.48 -9.684 Td[(ati+btiri+ckXh=1)]TJ/F23 11.955 Tf 5.48 -9.684 Td[(pci;h+qci;hh0i=1;:::;nwLi0i=1;:::;nwLiri)]TJ/F23 11.955 Tf 11.955 0 Td[(rLii=1;:::;nwUi0i=1;:::;nwUirUi)]TJ/F23 11.955 Tf 11.955 0 Td[(rii=1;:::;nlhhuhh=1;:::;kUsingtheupperboundontheuncertaintyprovidedbyProblems2-UB,thepointyieldingthemaximumuncertaintyisobtainedbyusingthesamebranch-and-boundmethoddescribedin 3.6.2.3 ,exceptthats2nismaximizedratherthanIn.Alternatively,anotherapproachwouldbetochoosethenextpointbasedonmaximizinguncertaintyratherthantheexpectedimprovement.Thebranch-and-boundapproachdescribedinSection 3.6.2.3 canbeadaptedtosolvethatproblemratherthanmaximizingtheexpectedimprovement.3.6.2.3Branch-and-BoundAbranch-and-boundmethodisusedtodeterminethemaximumexpectedimprovementineachiteration.Atsomepointinthealgorithm,npoints,;:::;n,havealreadybeenobserved.Thesolutionspaceisdividedintoregionsbasedonthesepreviouslysampledpointsandconsidereachregionasaseparatesubproblem.Eachofthesesubproblemsissolvedusingbranchandbound.First,theupperboundontheuncertaintyisdeterminedasdescribedinSection 3.6.2.2 usingthesubregion's 61

PAGE 62

lowerandupperboundson.Next,thelowerbound^FLon^FnisdeterminedusingthemethodinJones[ 68 ]andJonesetal.[ 69 ].Theupperboundons2nandlowerboundon^FarenowusedtodetermineanupperboundonInoverthecurrentsubregionbysolvingforInsubstituting^Fn=^FLandsn=sUasdescribedinJones[ 68 ]andJonesetal.[ 69 ].Inaddition,thethatyieldedthemaximumuncertaintycanbeusedtoevaluatethefunctionIn,yieldingalowerboundonInovertheintervallhhuh,h=1;:::;k.Thisvalueisusedtoupdatethecurrentbestlowerboundfoundi.e.,ifthecurrentbestlowerboundislessthanthenewlowerboundfound,thecurrentbestlowerboundisreplacedbythenewone;otherwise,thecurrentbestlowerboundisunchanged.Iftheupperboundislessthanthecurrentbestlowerbound,thesubregionisdiscardedasnotinteresting.Ifthelowerandupperboundareveryclose,wesaythatwehavefoundtheoptimumoverthecurrentsubregion.Otherwise,theupperboundissignicantlylargerthanthecurrentlowerbound,sothesubregionisfurtherdividedintosubregionsasdescribedbelowandtheprocedureisrepeatedforeachofthenewregions.Thisisthebranchingstep.Atsomepoint,therearenomoresubregionstoconsider,aswehaveeitherdecidedtheyarenotinterestingorhavefoundtheoptimalsolutionforthatsubregion.Then,thealgorithmterminatesandthecurrentbestlowerboundistheoptimalsolutionforInoverthecurrentregion.Thisbranch-and-boundprocedureisappliedtoeachoftheregions,andtheoveralllargestInvalueisthenthemaximumIn,andthecorrespondingisthenextpointatwhichtoevaluatetheFMOfunction.Selectingthesubregions.Animportantcomponentofthebranch-and-boundalgorithmisthemethodofselectingthesubregions.Thedenitionofthesesubregions,aswellasltheorderinwhichtheyareexplored,canhavesignicantimpactonboththeamountoftimeandmemoryrequiredtoperformthealgorithm.Asourimplementation 62

PAGE 63

ofthebranch-and-boundmethodrequiresthattheentiresolutionspacebedividedintosubregionsbeforethebranch-and-boundalgorithmbegins,theselectionoftheseinitialregionsmayalsoaectthespeedofthealgorithm.Initialregions.Beforebeginningthebranch-and-boundprocess,thesolutionspaceofthedecisionvariables,h2[0;360]forallh=1;:::;k,isdividedintoasetofinitialregions.Ifrepresentsnon-coplanarorientations,weconsidertwowaysofselectingtheregionsdenedbythenon-coplanarorientations.First,weconsidertheentiresolutionspaceastheonlyregion,thatis,insteadofdividingthesolutionspaceintoseveralsubregions,weonlyconsideronesubregionthatencompassestheentiresolutionspaceseeFigure 3-4 A.Second,denoteasubsetofvariableindicesHf1;:::;kg.Foreachindexh2H,orderthenpreviouslysampledpointsincreasinglybyh.Foreachpreviouslysampledpointi=1;:::;n)]TJ/F15 11.955 Tf 12.47 0 Td[(1,considertheregionsdenedbylh=0anduh=360forh=2H,andlh=ihanduh=i+1h.Alsoconsidertheregiondenedbylh=0anduh=360forh=2H,andlh=0anduh=h.Similarly,considertheregiondenedbylh=0anduh=360forh=2H,andlh=nhanduh=360.Figures 3-4 A3-4 DillustratetheinitialregionsfordierentHvalueswherek=2.DenotetheinitialregionsetwhereH=;asB0Figure 3-4 A,H=f1gasB1Figure 3-4 B,H=f2gasB2Figure 3-4 CandH=f1;2gasB2Figure 3-4 D.Notethatinthecoplanarcase,itisonlynecessarytotesttheinitialregionschemeforoneanglebecausetheanglesareinterchangeable.Boundsfordiscreteandcontinuousvariables.Ifisdiscrete,thepointsontheboundarybetweenbetweenthetwosubregionswillbecontainedinbothsubregions,thuscreatinganineciency.ThiscanbeseeninFigure??,wherebisthepointatwhichwebranchandthebluelinerepresentsthedivisionoftheregionintotwosubregions.Theboundarylineiscontainedinboththetopintervalandthebottominterval.Thisoverlapcanbeavoidedwhenisintegralbyadjustingtheboundsbetweensubregionsinsucha 63

PAGE 64

AB CDFigure3-3.Initialregionsinthebranch-and-boundalgorithm.AInitialregionswithH=;B0.BInitialregionswithH=f1gB1.CInitialregionswithH=f2gB2.DInitialregionswithH=f1;2gB3. 64

PAGE 65

wayastopreventoverlappingbetweenanysubregions.Ifthelowerboundlhonhinasubregionisfractional,thenwediscardthenon-integralsolutionsbysettinglh=dlhe.Similarly,iftheupperbounduhonhinasubregionisfractional,thenuh=buhc.Ifthelhanduhboundsareintegralandlh=uh,overlappingisavoidedbysettinglh=lh)]TJ/F15 11.955 Tf 12.408 0 Td[(1seeFigure??.Ifiscontinuous,theboundscannotbeadjusted.Branchingscheme.Thebasicprincipleofthebranch-and-boundmethodistodecomposeregionsintosmallersubregionsinsuchawaythatasmanysubregionsaspossiblecanbediscardedasuninteresting,leavingareducednumberofsubregionsthatmustactuallybesearched.Thebranch-and-boundmethodisawellstudiedproblem,andassuch,therearenumerousmethodsofselectingthesubregions.Regionsmaybedividedintotwoequalsubregionsbisection,ormoregenerally,intomultiplesubregionswhichmayormaynotbeequalinsizemultisectionCsallneretal.[ 70 ],LagouanelleandSoubry[ 71 ].SomeothercommonmethodsincludeselectingonlyasubsetofvariablesonwhichtobranchEpperlyetal.[ 72 ],usingLangrangiandualitytoobtainlowerboundsBarrientosandCorrea[ 73 ],Thoai[ 74 ],Tuy[ 75 ]andapplyingdecompositionalgorithmsPhongetal.[ 76 ],Bomze[ 77 ],CambiniandSodini[ 78 ].Inourbranchingstep,weformthesubregionsbasedonsomepointintheregion.Theregionisdividedatthispointalongoneoftheindices.InFigure 3-4 A,bisthepointatwhichwebranch.Webranchbydividingtheregionhorizontallyintotwosubregionsatb,takingintoaccounttheadjustmentstotheboundsdescribedabovesoastoavoidoverlappingregions.Fork=2,ineachbranchingstep,wealternatelydividetheregionhorizontallyalongindex2andverticallyalongindex1asshowninFigures 3-4 B{ 3-4 D.AfterbranchinghorizontallyonceatbasshowninFigure 3-4 B,weexaminethetopregionandselectbasthepointatwhichwebranch.Wethenbranchbydividingthissubregionverticallyatb.Weproceedinthesamemannerforb,wherewebranchhorizontally,andsoonuntiltheconvergencecriteriaismet. 65

PAGE 66

Inthegeneralcase,wedividetheregionintotwosubregionsalongthebranchingindexwhilecyclingthrougheachoftheindicesh=1;:::;ksequentially.Forthebranchingindexh,theboundsforonenewsubregionarelh=lhanduh=b;h)]TJ/F15 11.955 Tf 12.871 0 Td[(1,andtheboundsfortheothernewsubregionarelh=b;handuh=uh.Thelowerandupperboundsontheregionfortheremainingindicesareunchangedforbothnewsubregions,i.e.lh=lhanduh=uhforh6=h.Inthenon-coplanarcase,abeaminmayberepresentedbymorethanoneindex.Forexample,ifasinglenon-coplanarbeamconsistingofcouchandgantryrotationisoptimized,thevectorconsistsof1representingthegantryangleand2representingthecouchangle.Thebranchingindexh2f1;2grepresentsbranchingoneitherthegantryangleoronthecouchangle.Iftwosuchnon-coplanarbeamsareoptimized,thenconsistsof1and2representingthegantryandcouchanglesoftherstbeam,respectively,and3and4representingthegantryandcouchanglesofthesecondbeam,respectively.Thebranchingindexh2f1;2;3;4gthenrepresentsbranchingonasinglecomponentofasinglebeam.Accountingforsymmetry.Inthecasewhererepresentsasetofcoplanarbeamangles,theorderingofthevariablesinisirreleventtotheFMOvalueobtainedat.Forexample,if=;20;30;40and=;30;40;10,thenF=F.Thus,itisredundanttoconsiderbothand,andeliminationoftheseredundantregionscangreatlydecreasethesizeofthesolutionspace.Forexample,ifweconsiderthetwo-dimensionalcasek=2,thesolutionspaceisasquareregionwith01360and02360.Thepointsabovetheline12areequivalenttothepointsbelowtheline,soweonlyneedtoconsideroneoftheseregions.Saywebranchbysplittingtheregionintofourequalquadrants,asshowninFigure 3-5 A.Ifwearbitrarilychoosetoonlyexaminethepointsabovetheline12,thenquadrant4canbeeliminated. 66

PAGE 67

AB CDFigure3-4.Partitioningaregionintosubregions.APartitioningaregionintosubregionswithoutaccountingforoverlap.BPreventingoverlappingregions.CRegionsaftertwobranches.DRegionsafterthreebranches. 67

PAGE 68

ABFigure3-5.Accountingforsymmetry.AAccountingforsymmetryin2D.BAccountingforsymmetryin3D. Inthreedimensions,thesolutionspaceisacube.Ifwebranchbysplittingthecubeintoeightequalcubes,theregiontobeexaminedisshowninFigure 3-5 B,wheretheoriginisthebackbottomleftcornerofthecube.Fromthisgure,wecanseethatasizableportionofthesolutionspacecanbediscarded.Inregionswheretherearebothviableandredundantsolutionsforexample,quadrants2and3inFigure 3-5 A,theadditionofconstraintsrequiringthat1:::kintheproblemofmaximizingtheexpectedimprovementensurethatonlytheuniqueportionoftheregionisconsidered.Ifmorethanonenon-coplanarorientationisoptimized,asimilarsymmetrytothemultiplecoplanarorientationsymmetryexists.Consideranimplementationwheretwonon-coplanarbeamorientationsareoptimized,andtheseorientationsareobtainedfromrotatingboththegantryandthecouch.Eachbeamisrepresentedbytwovariablesinthesolutionvector:onevariableindicatingthedegreeofgantryrotation,andonevariableindicatingthedegreeofcouchrotation.Let1and2bethegantryrotationandcouchrotationoftherstbeam,respectively,and3and4bethegantryrotationandcouchrotationofthesecondbeam,respectively.Then,thesolutionvectorf1;2;3;4g 68

PAGE 69

isidenticaltothesolutionvectorf3;4;1;2g.Becausethecouchangleselectedisdependentonthegantryangleandviceversa,thissymmetrycanbeexploitedbyonlyremovingredundantsolutionsfromoneofthebeamvariables,thatis,byrequiringthat13removingredundancyfromthegantryanglesor24removingredundancyfromthecouchangles.Ingeneral,ifddegreesofmotionareusedtoobtainmbeamorientations,andthelinearacceleratormotionvariablesareinthesameorderforeachbeam,thenkk+dk+2d:::k+m)]TJ/F22 7.97 Tf 6.586 0 Td[(1dforsomek2f1;:::;dg.3.6.3MethodofObtainingtheNextObservationTheRSalgorithmallowsfortwomethodsofselectingthenextpointtoobserve:bymaximizingtheexpectedimprovementorbymaximizingtheuncertainty.Inthesetests,thepointtoobserveisobtainedbyrstselectingthepointthatmaximizestheexpectedimprovementuntilthemaximumexpectedimprovementfallsbelowacertainthreshold,andthenswitchingtothepointthatmaximizestheuncertainty.Oncethemaximumuncertaintyalsofallsbelowacertainthreshold,thealgorithmterminates.Byrstselectingaccordingtotheexpectedimprovement,themethodquicklyobtainsagoodsolution.Bythenselectingaccordingtouncertainty,theoreticalconvergencetotheglobalminimumisensured.3.7NeighborhoodSearch3.7.1IntroductionFromAlemanetal.[ 79 ],wetestthesimulatedannealingalgorithmontheBOOproblem,aswellasexistingandnewvariantsofagreedyneighborhoodsearchheuristiccalledtheAdd/DropalgorithmseeKumar[ 80 ]toobtainagoodsolutiontotheBOOproblem.IneachstepoftheAdd/Dropalgorithm,abeaminthecurrentbeamsetisreplacedbyaneighboringbeamthatyieldsanimprovingsolution.Aswiththesimulatedannealingimplementation,wealsoapplyournewneighborhoodtotheAdd/Dropalgorithmandcompareitsperformancetoacommonlyusedneighborhoodstructure. 69

PAGE 70

3.7.2NeighborhoodSearchApproachesNeighborhoodsearchapproachesarecommonmethodsofobtainingsolutionstoglobaloptimizationproblems.Foravectorofdecisionvariables,aneighborisobtainedbyperturbingoneormoreofthedecisionvariables.Aneighborhoodforaparticularvectorofdecisionvariablesisthesetofallitsneighborsforagivenmethodofperturbatingthedecisionvariablevector.Asolutionisconsideredtobelocallyoptimaliftherearenoimprovingsolutionsinitsneighborhood.Bothdeterministicandstochasticneighborhoodsearchalgorithmshavebeenappliedtoawidevarietyofoptimizationproblems.Adeterministicneighborhoodsearchalgorithmisoneinwhichtheentireneighborhood,orapre-denedsubsetoftheneighborhood,isenumeratedineachiterationtondanimprovingsolution.Stochasticversionsofneighborhoodsearchapproaches,forexample,simulatedannealing,randomlyselectneighboringsolutionsinanattempttondanimprovingsolutionineachiteration.FortheBOOproblem,weconsidertwoneighborhoodsearchmethods.Therstisadeterministicneighborhoodsearchalgorithmthatndsalocallyoptimalsolution,andthesecondisthesimulatedannealingalgorithm,which,althoughbasedonneighborhoodsearches,provablyconvergestothegloballyoptimalsolutionforcertainneighborhoodstructures.3.7.3ADeterministicNeighborhoodSearchMethodforBOODeterministicneighborhoodsearchmethodsareoptimizationalgorithmsthatstartfromagivensolutionandtheniterativelyselectthebestpointinthecurrentneighborhoodasthenextiterate.Thebestpointintheneighborhoodcanbefoundbycompleteenumerationiftheneighborhoodissmall,orbyoptimizationistheneighborhoodislargeorifobjectivefunctionevaluationsareexpensive.DuetothecomplexityoftheBOOproblem,evenwhenonlyasubsetofavailableorientationsisconsidered,wewillfocusonsmallerneighborhoodsanduseenumeration.Theneighborhoodcouldalternativelybesearchedheuristically,forexamplebysearchingtheneighborhooduntil 70

PAGE 71

therstimprovingsolutionisfound,ratherthanthebestimprovingsolution.Ifnoimprovedsolutioncanbefoundthecurrentsolutionisalocaloptimum.InourimplementationoftheAdd/Dropalgorithm,asmallneighborhoodisdesiredforenumerationpurposes.Ineachiteration,aneighborhoodforjustasinglebeamisconsidered.Sayabeamsetconsistingofkbeamsisdesired.LettingtheneighborhoodofasinglebeamhinbedenotedasNh,theAdd/Dropalgorithmisasfollows: Initialization: 1. Chooseaninitialstartingsolution. 2. Set=andi=0. Iteration: 1. Selecth2f1;:::;kg,thengenerate2Nhi. 2. IfF
PAGE 72

3.7.3.2NeighborSelectionTheprocessofselectinganeighboringpointineachiterationconsistsoftwosteps:selectingtheindexhtochangeandthenselectinganimprovingangleinNhtoreplaceh.Ifhisselectedasimodk+1,thealgorithmwillcyclethrougheachindexsequentially,similartoaGibbsSamplersee,forexample,GemanandGeman[ 81 ]andGelfandandSmith[ 82 ].TheGibbsSampleralsousesasimilartwo-stepapproachtogeneratinganewpointbysequentiallygeneratinganewvalueforeachvariableinturn.Ifhisselectedrandomlyineachiteration,theresultingalgorithmissimilartoaHit-and-Runmethodsee,forexample,Smith[ 83 ]andBelisle[ 84 ],inwhichavariabletobechangedisselectedrandomly,andthenanewvalueforthatvariableisalsoselectedrandomlywithinaneighborhood.Oncehisselected,thenewvalueforhcanbegeneratedbyenumerationorbyaheuristicmethod.TheAdd/Dropalgorithmcomparesthequalityofthenewsolutiontothecurrentsolution,andthenonlyacceptsimprovingsolutions.Thisgreedyapproachresultsinalocallyoptimalsolution.3.7.3.3ImplementationTheindexofbeamangletobechangedineachiteration,hinStep1ofthealgorithminSection 3.7.3 ,ischosenash=imodk+1tocyclethrougheachindexinasequentialmanner.IntheAdd/Dropimplementation,oncehisdetermined,initerationiischosenas=argmin2NhifFg.Byreplacingeachbeambythemostimprovingneighbor,theAdd/Dropalgorithmisagreedyheuristicwhichterminateswhenthereisnoimprovingneighborforanybeam.Amulti-startaspectisaddedbyrepeatingthealgorithmwithmultipleinitialstartingpoints.Forexample,onestrategytoselectstartingpointswouldbetoselectarandomstartingpointaccordingtoaparticulardistribution.Anotherstrategywouldbetoselectanequi-spacedsolutionandrotateitaxednumberoftimestoobtainnewstartingpointsuntiltheinitialequi-spacedsolutionisrepeated.Equi-spacedbeamsolutionsarecommon 72

PAGE 73

inclinicalpracticeforanoddnumberofbeams.Thereasonthatsuchamethodisnotgenerallyusedinpracticeforeven-numberedbeamsisthattheresultingbeamsetwouldcontainparallel-opposedbeamsbeamsthatlieonthesameline,whicharenotusedbyconventionasitisbelievedthattheeectofaparallel-opposedbeamisverysimilartosimplydoublingtheradiationdeliveredfromabeam.Ifanequi-spacedsolutionisnotpossiblegivenabeamsetofkbeamsandthediscretizationlevelofthecandidatebeamsetB,thenthesolutioncanberoundedsothath2B,h=1;:::;k.3.7.4SimulatedAnnealingThesimulatedannealingalgorithmusedissimilartotheclassicalsimulatedannealingapproachproposedinKirkpatricketal.[ 85 ].ThesimulatedannealingalgorithmisbasedontheMetropolisalgorithm,whereinaneighboringsolutiontothecurrentiterateisgenerated,andifitisanimprovingpoint,itbecomesthecurrentiterate.Otherwise,itbecomesthecurrentiteratewithprobabilityexpfF=Tg,whereFisthedierenceinFMOvaluebetweenthecurrentiterateandthenewlygeneratedpointandTisthetemperature,ameasureoftherandomnessofthealgorithm.IfT=0,thenonlyimprovingpointsareselected.IfTisverylarge,thenanymoveisaccepted,whichisessentiallyarandomsearch.ThesimulatedannealingalgorithmstartswithaninitialtemperatureT0andperformsanumberofiterationsoftheMetropolisalgorithmusingT=T0.Then,thetemperatureisdecreasedaccordingtosomecoolingschedulesuchthatfTig!0.ObviousparallelscanbedrawnbetweenthesimulatedannealingalgorithmandtheAdd/DropneighborhoodsearchmethoddescribedinSection 3.7.3 .WhiletheAdd/Dropalgorithmdeterministicallysearchestheneighborhoodforimprovingsolutions,thesimulatedannealingalgorithmrandomlyselectsneighboringsolutions.Ratherthanbeinglimitedbytheabilitytoonlymovetoimprovingsolutions,thesimulatedannealingalgorithmmaystillmovetoanon-improvingsolutionwithacertainprobability,thus 73

PAGE 74

allowingfortheescapefromlocalminima.TheAdd/Dropalgorithm,ontheotherhand,isagreedyalgorithmthatisspecicallydesignedtondlocalminima.ThesimulatedannealingalgorithmisessentiallyarandomizationoftheAdd/Dropalgorithm.Inadditiontotheaddedrandomness,thepossibilityofchangingmorethanonebeamineachiterationisallowedbyselectingasetofindicesHf1;:::;kgtochange,ratherthanjustselectingasingleindexh.Thesimulatedannealingalgorithmisasfollows: Initialization: 1. ChooseaninitialbeamsetandcalculateitsFMOobjectivefunctionvalueF0. 2. Set^=,^F=F0,i=0. Iteration: 1. SelectHf1;:::;kg,generate2[h2HNhi,andcalculateitsFMOobjectivefunctionvalueF. 2. IfF<^F,set^F=F,Fi+1=F,i+1=and^=.Otherwise,setFi+1=Fandi+1=withprobabilityexpfFi)]TJ/F23 11.955 Tf 11.955 0 Td[(F=Tig. 3. Setii+1andrepeatStep1.ThesimulatedannealingalgorithmhasbeenpreviouslyappliedtotheBOOproblem.BortfeldandSchlegel[ 35 ]usethefast"simulatedannealingalgorithmdescribedbySzuandHartley[ 86 ]whichemploysaCauchydistributioningeneratingneighboringpoints.Steinetal.[ 40 ],Rowbottom[ 39 ]andDjajaputraetal.[ 36 ]alsouseaCauchydistributioningeneratingneighoringsolutions.Luetal.[ 37 ]randomlyselectnewpointssatisfyingBEVandconventionalwisdomcriteriaandPugachevandXing[ 38 ]randomlygeneratenewpointsandthenvarythemaccordingtoanexponentialdistribution.Allacceptimprovingsolutions,andwiththeexceptionofRowbottometal.[ 39 ]whoonlyacceptimprovingsolutionsessentiallyTi=0foralli,allacceptnon-improvingsolutionswitha 74

PAGE 75

Boltzmannprobability.NoneofthepreviousBOOstudiesemployingsimulatedannealingusetheexactFMOasameasureofthequalityofabeamset.3.7.4.1NeighborhoodDenitionTwoneighborhoodstructuresareexplored.TherstneighborhoodissimilartothatdescribedinSection 3.7.3.1 inthataneighborhoodNhisconsideredforonlyasinglebeamindexh2f1;:::;kg,justasintheAdd/Dropmethod.Asanextensiontochanginingasingleangleineachiteration,wealsoconsideraneighborhoodthatinvolveschangingallbeamsineachiteration,correspondingtoH=f1;:::;kginStep1ofthesimulatedannealingalgorithminSection 3.7.4 .ThisneighborhoodisdenedasN=[kh=1Nh.Again,theneighborhoodsfortheindividualbeamsaredenedasintherstmethod,withboundsofdegrees.3.7.4.2NeighborSelectionThemethodofselectinganeighbordependsontheneighborhoodstructureasdescribedinSection 3.7.4.1 .Intherstmethodwhereonlyonebeamischangedatatime,aneighborisselectedusingtherandomizedapproachdescribedinSection 3.7.3.2 .Oncehisselected,theprobabilityofselectingaparticularsolutioninNhwherethenewisddegreesfromhisPfD=dg,whereDistherealizationofarandomvariableofsomeprobabilitydistributiondenedontheinterval[)]TJ/F23 11.955 Tf 9.298 0 Td[(;)]TJ/F23 11.955 Tf 9.298 0 Td[(+1;:::;].FortheneighborhoodNwhereallbeamsarechangedinaniteration,thenewvalueforeachbeamh2f1;:::;kgisgeneratedfromNhinthesamemannerdescribedabove.3.7.4.3ImplementationInadditiontobasingouralgorithmsontheexactFMOsolutionratherthanonheuristicsorscoringmeasures,oursimulatedannealingapproachdiersfromthepreviousstudiesinthedistributionusedtogenerateneighbors,thedenitionoftheneighborhood,thecoolingscheduleandthenumberofiterations/restartsused.Notonlydoweuseanewneighborhoodstructure,butalsoageometricprobabilitydistributionratherthana 75

PAGE 76

uniformorCauchydistributionontheneighborhood.ThegeometricdistributionissimilarinshapetotheCauchydistributioninthattheybothcanhavefattailsdependingonthechoiceofprobabilityparameters.Thefattailsofthesedistributionsallowforpointsfarawayfromthecurrentsolutiontobeselectedassuccessiveiterates,whichpotentiallyincreasesthelikelihoodofndingagloballyoptimalsolution.Thegeometricdistributionhastheaddedattractivenessofproducingdiscretesolutions,whichisdesirablefortheBOOprobleminwhichdiscretesolutionsarepreferred.ByusingthecoolingscheduleTi+1=Tiwith<1,thesequenceoftemperaturesfTigconvergestozeroasthenumberofiterationsincreases.Inourapproach,theneighborhoodofabeamforboththeNhandNneighborhoodsisdenedusing=180,thatis,Nh=B.Bydeningtheneighborhoodofeachbeamtobetheentiresingle-beamsolutionspace,thesimulatedannealingalgorithmconvergestotheglobaloptimumwhenusingtheneighborhoodNdenedinSection 3.7.4.1 .ThoughNhislarge,eachbeaminNhisassignedaprobabilitysothatonlythebeamsclosesttohhaveasignicantprobabilityofbeingselected.Figure 3-7 Ashowstheprobabilityofreplacinghwithbeamsatvaryingdistancesusingprobabilityp=0:25forthegeometricdistribution.Notethatthecurrentbeamcannotbeselectedasareplacement.AswiththeAdd/Dropmethod,amulti-startaspectisaddedtothesimulatedannealingalgorithmbyrepeatingthealgorithmusingseveraldierentstartingpoints.3.7.4.4ConvergenceUnlikemanypreviouslyproposedsimulatedannealingalgorithms,ouralgorithmconvergestothegloballyoptimalsolutiontotheBOOproblemundermildconditions.Thefollowingtheoremsummarizestheseconditions. Theorem3.7.1. Supposethat H=f1;:::;kg limi!1Ti=0 =180 76

PAGE 77

Thereisapositiveprobabilityofgeneratinganysolutionintheneighborhood.Thenoursimulatedannealingalgorithmconvergestotheglobaloptimumsolutioninthesensethatlimi!1Fi=FinprobabilitywhereFistheglobaloptimumvalueoftheBOOproblem. Proof. ThisfollowsfromTheorem1inBelisleetal.[ 87 ]. 3.7.5ANewNeighborhoodStructureFortheBOOproblem,theneighborhoodstructurethatistypicallyusedforavectorofbeamorientationsissimplythecollectionofbeamvectorsobtainedfromchangingoneormoreofthebeamstoaneighboringbeam,whereeachbeamhasitsownneighborhoodNh.InadditiontoNh,weconsideranewneighborhoodwhichwecallaip"neighborhood.TheipneighborhoodofabeamindexhconsistsofNhplusaneighborhoodaroundtheparallelopposedbeamofh.Theparallelopposedbeamisthebeam180away,thatis,h0=h+180mod360TheipneighborhoodcanbedenedasNFh=n1;:::;h)]TJ/F22 7.97 Tf 6.586 0 Td[(1;mod360;h+1;:::;k2Bk:2[h)]TJ/F23 11.955 Tf 11.955 0 Td[(;h+][h+180)]TJ/F23 11.955 Tf 11.955 0 Td[(F;+180+FoNotethatthevaluesandFmaybedierent.Figure 3-6 depictsaipneighborhoodforabeamlocatedat0degrees,thecenterofthetopshadedwedgerepresentingNh,whereh=0.Themotivationfortheipneighborhoodsarisesfromtheobservationthatmanyofthe3-beamsimulatedannealingplansgeneratedusingtheregularneighborhoodcontainedtwobeamsveryclosetotwobeamsintheoptimalsolutionobtainedbyexplicit 77

PAGE 78

enumeration,whilethethirdbeamwasveryclosetotheparallelopposedbeamofthethirdbeamintheoptimalsolution.Giventhisobservation,itisintuitivethattheinclusionoftheneighborhoodaroundtheparallelbeamshouldprovideimprovedsolutions.TheneighborhoodsNhandNFhwithvaryingFvaluesareappliedtoboththeAdd/Dropandthesimulatedannealingframeworks.Forthegeometricprobabilitydistributionusedinthesimulatedannealingmethod,Figure 3-7 Bshowstheprobabilityofselectingbeamsatdierentdistancesusingaipneighborhoodwithprobabilityp=0:25.Notethatthecurrentbeamcannotbeselectedasitsownneighbor. Figure3-6.NhtopshadedareaandNFhtopandbottomshadedareasforh=0. ABFigure3-7.Selectionprobabilities.ANh.BNFh. 78

PAGE 79

3.8ResultsInadditiontojudgingtheBOOalgorithmsaccordingtotheircomputationaltime,theplansmustalsobeevaluatedforclinicalviability.AllcriteriausedarethoseemployedattheDavisCancerCenteratShandsHospitalattheUniversityofFlorida.3.8.1EvaluatingPlanQualityInordertoformulateanoptimizationproblem,aquantitativemeasureofthetreatmentplanqualityisneeded.Thismeasure,theFMOfunctionvalue,needstoappropriatelymakethetrade-obetweenthecontradictorygoalsofcoveringtargetsandsparingcriticalstructures.Typically,agoodplanensuresthatatleastacertainpercentofeachtargetreceivestheprescriptiondose.Acoldspotoccurswherelessthanacertainpercentofthetargetreceivestheprescriptiondose.Similarly,ahotspotoccursifasignicantpercentageofthetargetreceivesmorethantheprescriptiondose.3.8.1.1TargetcoverageEachoftheplanscontainstwotargetstructures,orplanningtumorvolumesPTV:oneisthetumormassobservedfromimagingscans,whichwewillcallPTV2,andtheotheristhePTV2plussomemarginspeciedbythephysician,whichwewillcallPTV1.ThePTV1structureisusedbyphysiciansincasethereareelementsofthetumormassthatcannotbeseenfromtheimagingscans.ThedoseprescribedforPTV1islessthanthedoseprescribedforPTV2.Fortargetstructures,werequirethatatleast95%ofthetargetreceivesthefullprescriptiondose,sothedosethatisreceivedbyatleast95%ofeachofthetargetsismeasured.Wewanttorestricttheamountofthetargetthatreceivesmorethantheprescriptiondose.BecausePTV2iscontainedinsidePTV1,PTV2willnecessarilyhaveasizable,butlessimportant,areareceivinganoverdose.Thus,weareonlyconcernedwithPTV2overdose.Toevaluatethesizeofthehotspot,wecheckthepercentvolumeofPTV2thatreceivesmorethan110%oftheprescriptiondose.Toevaluatethecoldspots,wecheck 79

PAGE 80

Table3-1.Sparingcriteriavariesforeachcriticalstructure StructurePercent%DoseGy brainstem10055eyes5030mandible10070opticchiasm10055opticnerves10050parotidglands5030skin10060spinalcord10045submandibularglands5030 thepercentvolumeofbothPTV1andPTV2thatreceivesatleast93%oftheprescriptiondose.Theprescriptiondosesaresetto54GyforPTV1and73.8GyforPTV2,whicharethedosevaluesusedatShandsHospitalattheUniversityofFlorida.3.8.1.2CriticalstructuresparingThecriticalstructuresinvolvedineachcasevary,dependingontheirproximitytothetumor.Thecriticalstructurescanbeclassiedintotwogeneralgroupsaccordingtotheirabilitytosurviveradiationdose.Parallelstructures,e.g.,salivaglands,willcontinuetofunctionaslongacertainpercentageoftheorganreceiveslessthanacertainamountofdose.Serialstructures,ontheotherhand,willceasetofunctionifanyoftheorganreceivesoveracertainamountofdose.Thespinalcordisoneexampleofaserialstructure|ifitreceivestoomuchdose,theeectisequivalenttocuttingitinhalf,leavingthepatientparalyzed.Thesparingcriteriaforeachofthecommoncriticalstructuresinhead-and-neckcasesarelistedinTable 4-2 .Thecriticalstructuresinvolvedineachcasevary,dependingontheirproximitytothetumor.Therearefoursalivaglands:onesubmandibularandoneparotidglandoneachoftherightandleftsides.Thesalivaglandsareofparticularimportancebecausetheirlosscangreatlydecreasethepatient'squalityoflife,butbecauseoftheirlocationrelativetotheusualtumorpositions,theycanbediculttospare.Studiesshowthatapatientcanleadarelativelynormallifewiththreeofthefourglandsspared.Thelossofother 80

PAGE 81

organs,especiallythespinalcord,willalsogreatlyaectthepatient'squalityoflife,buthead-and-necktumorsareusuallysituatedinsuchawaythatotherorganscanbeeasilysparedintheFMOoptimzation.Thus,theresultspresentedplaceparticularemphasisonthesparingofsalivaglands.RatherthanrelyingstrictlyonFMOvalue,atoolcommonlyusedbyphysicianstojudgethequalityofatreatmentplanisthedose-volumehistogramDVH.Thishistogramisameasureofthecumulativedosereceivedbyagivenstructure.Itspeciesthefractionofeachstructure'svolumethatreceivesatleastacertainamountofdose.Althoughthereareseveralcriticalstructurestobeconsideredinhead-and-neckcases,thesalivaglandsarenotoriouslythemostdiculttospareduetotheirproximitytocommontumorlocations.Thus,forclarity,theDVHresultsprovidedincludeonlytargetstructuresandsalivaglands.EachofthetreatmentplanssparesallorgansnotshownintheDVHs.IntheDVHresultsprovided,verticallinesindicatetargetprescriptiondoses,andasterisksmarkthesparingcriteriaforthesalivaglands.3.8.2ResponseSurfaceMethodResultsTheresponsesurfacemethodwastestedonsixhead-and-neckcasesusingaWindowsXPcomputerwitha3.2GHzPentiumIVprocessorand2GBofRAM.ThesizesofthetestcasesforplanswiththreebeamsareshowninTable 3-2 .Eachalgorithmwasallowedtorunfor12hours,whichisnotanunreasonablerunlengthbecauseBOOwillnotbeperformedonaday-to-daybasis.ItisanticipatedthatBOOwillbeperformedonceovernightbetweenthetimethepatientisimagedandthetimethepatientbeginsradiationtherapy.Agoodbeamvectorchosenbeforetreatmentbeginsshouldcontinuetoprovidequalitytreatmentplansthroughoutthepatient'streatment,whichistypically35days.Thebeamorientationsfromwhichlinearacceleratorsarecapableofdeliveringradiationarenotrestrictedtointegervaluedegrees.Inthisstudy,integralbeamorientationsaredesiredtoaccountforsetuptolerances.Forthesamereasons,beamorientationsareconsideredona10grid.Toobtainintegralsolutions,inthesubproblem 81

PAGE 82

Table3-2.Sizesoftestcases. Case#bixels#voxels 1514345,6292546352,2843613347,2334549268,8235423271,1566585389,565 Avg.538329,115Low423268,823High613389,565 ofmaximizingI,theintegerconstraintisrelaxedintheproblemofdetermininganupperboundons2,andtheresultingsolutionisroundedtointegervalues.ThebranchingschemeusedtreatstheroundedsolutionasintegralandbranchessoastoavoidoverlappingsubregionsasdescribedinSection 3.6.2.3 .Resultsareprovidedforeachpossibleinitialregionscheme.Thepointatwhichbranchingisperformedineachregion,binSection 3.6.2.3 ,ischosenasthemidpointoftheregion.Also,riandhintheunderestimatingtermsinProblems2-UBinSection 3.6.2.2 aretakentobethemidpointsoftheirrespectiveintervals.Itisanticipatedthattheweighteddistancemeasureinequation 3{3 willhaveansignicantimpactonthealgorithm'sperformance.Intuitively,asmallweighteddistancecorrespondstoasmallcorrelationbetweenpoints,whichwillcausethealgorithmtobehavelocally.Inordertoinducethealgorithmtobehaveglobally,thealgorithmmustassumelesscorrelationbetweentwopoints.Ifthepointsarelesscorrelated,thealgorithmwillbelesslikelytostayintheneighborhoodofpreviouslysampledpoints.Thecorrelationbetweentwopointscanbedecreasedbyincreasingtheweighteddistancebetweenthepoints,whichcanbedonebyincreasingcorp.Ifcbecomessucientlylarge,thecorrelationbetweenpointswillbeeectivelyzero,thusyieldinganeectivelyrandomsearchalgorithm.Totesttheseexpectations,cwastestedwithvaluesof10.0,100.0and 82

PAGE 83

500.0.Ineachtest,verandomlyselectedstartingpointswereusedtoinitializetheRSalgorithm.Toevaluatethealgorithm'sperformanceacrossallofthetestedcases,therelativeimprovementsinFMOvalueovera5-beamequispacedplandenoted5equi,a7-beamequispacedplandenoted7equiandalocallyoptimal3-beamcoplanarplanobtainedusingalocalsearchalgorithmcalledtheAdd/DroplocalsearchheuristicintroducedbyKumar[ 80 ]anddenoted3A/Darecompared.3.8.2.1ProofofconceptTotesttheaccuracyoftheRSmethod,asinglecasewastestedwhereintheproblemofaddingasinglecoplanarbeamtoanequi-spaced,coplanar3-beamsolutionovera1gridwasconsidered.Thealgorithmwasinitializedwithtworandomlyselectedstartingpoints.Byconsideringsuchasmallscaleproblem,thesolutionspaceineachiterationcanbeexplicitlyenumeratedinordertoexactlyobtainthenextbestpointtosample.TheabilitytoenumeratethesolutionspacewillalsoallowustodeterminehowaccuratelytheRSmethodmodelstheFMOobjectivefunction.Ateachpointthathasbeensampled,boththeuncertaintyandtheexpectedimprovementwillbezero.Thisresultisnotonlytheoreticallytrue,butalsointuitivebecauseoncetheFMOvalueatacertainpointisknown,therewillbenoimprovementoverthecurrentbestFMOvaluebysamplingthatpointagain.Itisalsoexpectedthatasthealgorithmprogresses,theapproximationoftheFMOfunctionwillbecomeincreasinglyaccurate,withtheapproximationobtainingtheexactFMOvaluesatsampledpoints.Figures 3-8 A3-8 DdemonstratehowtheRSmethodbehavedaspredictedatdierentpointsintheRSalgorithm.TheexpectedvalueiszeroatsampledpointsandtheapproximationoftheFMOfunctionalmostperfectlytsthetrueFMOfunctionbythetimethealgorithmterminates.Theimportanceofthestartingpoints,thepointssampledbeforethealgorithmbeginstogivethemethodsomebaselineinformationabouttheFMOfunction,wasalsotested. 83

PAGE 84

AB CDFigure3-8.ProofofconceptresultsatvariousstagesoftheRSalgorithm.AAftertwopoints.BAfter20points.CAfter80points.DAfter148points,whenthealgorithmterminates. TheRSmethodwasrunwith100randomlygeneratedsetsofstartingpoints,andtheRSmethodobtainedtheglobaloptimumin90.6%oftrials,indicatingthattheperformanceofthealgorithmisnotsignicantlydependentonthestartingpoints.3.8.2.2Addinganon-coplanarbeamtoacoplanarsolutionNext,theproblemofaddinganon-coplanarbeamtoa3-beamlocallyoptimalcoplanarsolutionwasconsidered.ThelocallyoptimalsolutionisobtainedusingtheAdd/Dropalgorithm.Thebeamdataforthenon-coplanarbeambeingoptimizedisgeneratedon-the-y,andconsistsofgantryandcouchrotations,wherethebothgantryandcouchareallowedtorotateafull360ona10grid.Asthenalsolutionofthenon-coplanarRSplanwillbea4-beamplan,theresultsfromtheresponsesurfacesolution 84

PAGE 85

arecomparedtothelocallyoptimalcoplanar4-beamAdd/Dropplan,denoted4A/D.Theplanswillalsobecomparedtoanequi-spaced,coplanar7-beamplan,denoted7equi,whichiscommonlyusedinpracticetotreathead-and-neckcancers.Thereisrelativelylittledeviationinthenalsolutionsbetweenthedierentparameterchoicesandinitialregionsschemes,asshownbyTable 3-3 .Theresultsalsoindicatethatthestartingpointschosendonotsignicantlyaecttheoutcomeofthealgorithm.Thisimpliesthattheresponsesurfacealgorithmisrobustwithrespecttovaryingimplementations.Althoughthe4RSsolutionsobtainedanaverageof5.44%decreaseinFMOvaluefromthe7equiplans,the4RSsolutionsdidinfactobtainanaverageof16.12%improvementinFMOvalueoverthe4ADsolutions.DespitethedierencesinFMOvalue,alltreatmentplansexaminedweresimilarinclinicalquality,asdiscussedinSection 3.8.2.3 .Althoughthealgorithmwasallowedtorunfor12hoursineachscenario,theminimumFMOvalueobtainedbytheRSmethodwasfoundearlyon.Onaverage,thebestFMOvaluefoundwasobtainedin6.15hoursaftersampling27-40points.ForeachoftheRSmethodvariationstested,boththenumberofpointssampledandtherelativeimprovementsinFMOvaluearenearlyidentical.Thisindicatesthatthealgorithmisrobustwithrespecttoparameterandimplementationchanges.Thetimespentgeneratingbeamdatacomprisesapproximately84%ofthealgorithm'sruntime,whiletheresponsesurfaceportiononaverageaccountsforonly13%.Thus,itisexpectedthatchangestotheRSmethod,includingimprovementstothebranch-and-boundroutine,willnothaveaverystrongimpactonthenumberofpointsthealgorithmwillsampleinitsallottedruntime.3.8.2.3ClinicalresultsThetargetcoverageachievedbythedierenttreatmentplansaredisplayedinTable 3-4 .Onaverage,the7equiplanwasabletodeliverthemostamountofdosetoPTV2, 85

PAGE 86

Table3-3.MinimumFMOvalueobtainedandtimerequiredtoobtainit. Min.FMOvalueTimehrsCaseAvg.St.Dev.Avg.St.Dev. 1565.248.825.355.072570.5112.837.493.783927.3420.607.052.214710.927.726.543.395512.2220.046.963.336799.9534.073.483.40 Table3-4.Targetcoverageachievedbythetreatmentplans 4RS4A/D7equi PTV2doseat95%volume73.16Gy72.56Gy73.81GyPTV2%receiving>110%ofRx23.18%15.07%31.63%PTV2%receiving>93%ofRx98.87%98.67%99.57%PTV1doseat95%volume54.71Gy54.41Gy55.09GyPTV1%receiving>93%ofRx97.95%98.01%97.46% butthe4RSplanisveryclose.Bothofthe4-beamplansobtainsmallerhotspotsandbetterPTV1targetcoveragethanthe7equiplan.The4A/DplanonaverageunderdosesPTV2,whichcouldleadtorecurrenceofthecancer.Thisunderdosagecouldalsoaccountforthesmallerhotspotinthe4A/Dplans.Figures 3-10 and??illustratetworepresentativecaseswherethe4RS,4A/Dand7equiplanseachhaveclinicallyacceptabletargetcoverage.Theverticallineat73.8GyindicatestheprescriptiondoseforPTV2.TheabilityofeachofthetreatmentplanstosparetheorgansinthecasestestedisshowninTable 3-5 .Surprisingly,boththe4RSandthe4A/Dplanareequivalenttooroutperformthe7equiplanintermsoforgansparing.Inthe4-beamplans,theleftsubmandibularglandissparedin83%ofthetreatmentplansdeveloped,whereasitisonlysparedin67%ofthe7equiplans.OnecaseillustratingequivalentorgansparingisshowninFigure 3-10 ,andonecasedemonstratingimprovedorgansparingoverthe7equiplanisshowninFigure??.JustasPTV2underdosageinthe4A/Dplanslikelycontributedtothesmallerhotspots,itispossiblethattheimprovedorgansparinginthe4A/Dplansisalsoaresultoftheunderdosage. 86

PAGE 87

Table3-5.Percentageofplansinwhichanorganisspared. Structure4RS4A/D7equi brainstem100%100%100%mandible100%100%100%leftopticnerve100%100%100%rightopticnerve100%100%100%lefteye100%100%100%righteye100%100%100%opticchiasm100%100%100%leftparotidgland100%100%100%rightparotidgland67%67%67%leftSMBgland83%83%67%rightSMBgland50%50%50%spinalcord100%100%100%skin100%100%100% ABFigure3-9.7-beamequi-spaceddotted,4-beamAdd/Dropdashedand4-beamRSnon-coplanarsolidtargetandselectsalivaglandDVHs.ATargetcoverageisnearlyidentical.BThetumorsurroundstherightsubmandibulargland,sotheFMOsolverrecognizesthatitcannotbesparedandallowsittoreceiveasmuchdoseasnecessarytoensuregoodtargetcoverageinallplans.Allothersalivaglandsaresparedinallplans. 87

PAGE 88

ABFigure3-10.7-beamequi-spaceddotted,4-beamAdd/Dropdashedand4-beamRSnon-coplanarsolidtargetandselectsalivaglandDVHs.ATargetcoverageisnearlyidentical.BTheleftsubmandibularglandissparedbythetwo4-beamplans,butnotbythe7-beamplan.Allothersalivaglandsaresparedinallplans. 3.8.3NeighborhoodSearchMethodResultsThesimulatedannealingmethodwastestedonsixhead-and-neckcasesusingaWindowsXPcomputerwitha2.13GHzPentiumMprocessorand2GBofRAM.Onaverage,340FMOswerecalculatedin30-minuteruntimeallowedforthesimulatedannealingandAdd/Dropalgorithms.Beamswereselectedona5-degreegrid,yielding72candidatecoplanarbeams.ThesimulatedannealingandAdd/Dropalgorithmswereusedtoobtain4-beamcoplanarplansusingregularandipneighborhoods.Inordertocomparethequalityofthetreatmentacrossdierentplans,theplansarecomparedintermsofthepercentageimprovementsofeachplan'sFMOvalueimprovementovertheFMOvalueofthelocallyoptimal3-beamplanobtainedfromtheAdd/DroplocalsearchheuristicdescribedbyKumar[ 80 ].TheAdd/Dropplansaredenoted3A/Dand4A/Dforthe3-and4-beamplans,respectively.The4-beamplansgeneratedbythesimulatedannealingandAdd/Dropalgorithmsarecomparedtothetypical5-and7-beamequi-spacedplans, 88

PAGE 89

denoted5equiand7equi,respectively.Thesimulatedannealingplansaredenotedbytheimplementationnumbers,whichrefertotheparametersused,giveninTable 3-6 .Figure 3-11 demonstratestheimprovedconvergencetimespossibleusingtheipneighborhood.3.8.3.1Add/DropalgorithmresultsTheAdd/Dropalgorithmwasallowedtorunfor30minutestogeneratea4-beamplan.TheNhneighborhoodwith=20andtheNFhwithF=0andF=20neighborhoodsaretestedfortheAdd/Dropalgorithm.Thevalue=20ischosentoapproximatetheneighborhoodsizethatisexpectedfromthesimulatedannealingimplementationusingalargeipneighborhood,whereF=180.MoredetailsonthesimulatedannealingimplementationsareprovidedinSection 3.8.3.2 .UsingNh,the4-beamAdd/Dropsolutionisnearlyidenticalidenticaltothe7-beamequi-spacedplan,whiletheipneighborhoodsallowtheAdd/Dropalgorithmtond4-beamsolutionsthatexceedthequalityofthe7-beamplans.Figure 3-12 demonstratesthequalityofthesolutions,whileFigure 3-11 aillustratesthattheipneighborhoodsprovidefasterFMOconvergencethanthatofNh.3.8.3.2SimulatedAnnealingresultsSeveralparametersetsweretestedforthesimulatedannealingalgorithm.Forsimplicity,eachoftheparametersetsandmethodsofgeneratinganeighboringsolutionarenumberedaccordingtoTable 3-6 .Eachimplementationcontainsatotalof500iterations,i.e.,500sampledpoints,thusyieldingafaircomparisonbetweentheparameters.Toensureclinicalpracticality,thealgorithmwasallowedtorunforamaximumof30minutesor500iterations,whichevercamerst.Forthecoolingschedule,weupdatethetemperatureaccordingtoanexponentialcoolingschedule,Ti+1=Ti,where<1.Duetotherandomnatureofthealgorithm,thealgorithmisrestartedvetimes,eachtimewithadierentinitialstartingpoint.Therstinitialstartingpointisanequi-spacedsolution,andeachsubsequentstartingpointis 89

PAGE 90

ABFigure3-11.ComparisonofFMOconvergence.AAdd/Drop.BSimulatedannealing. thepreviousinitialsolutionrotatedbyddegrees,wherecandidateanglesareconsideredonad-degreegrid,thatis,everydthangleisconsidered.ThenumberofsimulatedannealingandMetropolisiterationsarechosensuchthatthetotalnumberofiterationsis500.TheinitialtemperaturevaluestestedareT0=0andT0=75.T0=0resultsintheacceptanceofonlyimprovingsolutions,whiletheinitialtemperaturevalue75wasselectedasthevaluethatwouldapproximatelyyielda50percentprobabilityofselectinganon-improvingsolutionfortheinitialiterationsofthealgorithm.ForboththeNhandNFhneighborhoods,=F=180isusedsothattheentiresolutionspaceisconsideredasaneighborhood.AsshowninFigure 3-7 A,theprobabilityofselectingabeam20awayusingtheNhneighborhoodwithgeometricdistributionwithp=0:25isonly0.39%ona5grid.Weconsiderthissucientlysmalltonotconsiderneighborhoodslargerthan=20forNhandF=20forNFhintheAdd/Dropalgorithm.JustasintheAdd/Dropimplementation,theneighborhoodNFhwithF=0isalsoconsidered. 90

PAGE 91

Table3-6.Denitionsofimplementations. NumbernmNT0 1100110.902101010.903100110.9904101010.9905100110.9756101010.9757100110.99758101010.997591001all0.90101010all0.90111001all0.990121010all0.990131001all0.975141010all0.975151001all0.9975161010all0.9975 Figure 3-11 bshowsthattheipneighborhoodsconvergeinFMOvaluesignicantlyfasterthandoestheNhneighborhood,whileFigure 3-13 showsthattheipneighborhoodsprovidecomparablesolutionqualitytoboththenon-ipsimulatedannealingand7-beamequi-spacedsolutions.3.8.3.3ClinicalresultsBecausethereisnofundamentalwayofquantifyingatreatmentplan,atoolcommonlyusedbyphysicianstojudgethequalityofatreatmentplanisthedose-volumehistogramDVH.ADVHisagraphicalmeasureofthecumulativedosereceivedbyagivenstructure.Itspeciesthepercentageofeachstructure'svolumethatreceivesatleastacertainamountofdose,thusprovidinganintuitivemeansofassessingthequalityofatreatmentplan.Theplanstestedplanseachcontaintwotargetstructures.ThegrosstumorvolumeGTVisthetumormassobservedfromimagingscans.TheclinicaltumorvolumeCTVistheGTVplussomemarginspeciedbythephysician.TheCTVisusedbyphysicians 91

PAGE 92

incasethereareelementsofthetumormassthatcannotbeseenfromtheimagingscans,andthedoseprescribedfortheCTVislessthanthedoseprescribedbytheGTV.DVHsforarepresentativecasecomparingthe7-beamequi-spacedplanwiththesimulatedannealingplansobtainedusingaregularneighborhoodandipneighborhoodswithF=0andF=180areshowninFigure 3-13 .Comparisonofthe7-beamequi-spacedplanandtheAdd/DropplansusingaregularneighborhoodandipneighborhoodswithF=0andF=20areshowninFigure 3-12 .Thesparingcriteriausedforthesalivaglands,nomorethan50%oftheglandreceiving30Gy,ismarkedbythestarinFigures 3-13 and 3-12 .TheprescriptiondosefortheGTVis73.8Gy,whichismarkedbytheverticallineinFigures 3-13 and 3-12 .Aspreviouslystated,fortargetstructures,werequirethatatleast95%ofthetargetreceivesthefullprescriptiondose.Figure 3-13 revealsthatthe7-beamequi-spacedplanactuallyoverdosesthetargetandhasalargerhotspotthanthe4-beamsimulatedannealingplans.The7-beamequi-spacedplanonlysparesthreeofthefoursalivaglands,whereasthe4-beamsimulatedannealingplanssparethreeormoresalivaglands.Thesimulatedannealingplansobtainedusingtheipneighborhoodsspareallfoursalivaglands,whiletheplanobtainedhowtheNhneighborhoodonlysparesthreesalivaglands,indicatingthattheipneighborhoodsdoinfactndsuperiorsolutionsintermsofclinicalquality.Figure 3-12 showsthatthe4-beamAdd/Dropplansobtainnearlyidenticalsolutionswhencomparedtothe7-beamequi-spacedDVHs.Theipneighborhoodsperformclinicallycomparablytotheregularneighborhoodplans,andalloftheAdd/Dropplansarecomparabletothe7-beamequi-spacedplanintermsofsalivaglandsparingandtargetcoverage.3.9ConclusionsandFutureDirections3.9.1ResponseSurfaceConclusionsWehaveshownthatforhead-and-neckcases,qualityplanswithfewerbeamsthanastandardtreatmentplancanbeobtainedifBOOisapplied.Theresponsesurface 92

PAGE 93

ABFigure3-12.ComparisonofAdd/Dropand7-beamequi-spacedplans.ATheAdd/Dropplansachievenearlyidenticaltargetcoveragewhencomparedtothe7-beamequi-spacedplan.BThesalivaglandsparingintheAdd/Dropplansandthe7-beamequi-spacedplanisclinicallyequivalent. ABFigure3-13.ComparisonofAdd/Dropand7-beamequi-spacedplans.AUnlikethe7-beamequi-spacedplan,the4-beamsimulatedannealingplansdonotoverdosethetarget.BThesimulatedannealingplansarealsocapableofsparingmoresalivaglandsthanthe7-beamequi-spacedplan. 93

PAGE 94

algorithmoperatesinaclinicallyreasonabletimeframe,andisgenerallysuccessfulinselectingnon-coplanarbeamorientationstoimprovetheFMOvalueoverthatoflocallyoptimalcoplanarsolutions.TheFMOvalueofthe4-beamresponsesurfaceplanswasalsoonlyslightlylargerthanthatofthe7-beamequi-spacedcoplanartreatmentplans,indicatingcomparabletreatmentplansdespitethedecreaseinthenumberofbeamsused.Intermsofclinicalresults,themostsignicantbenetofthenon-coplanarsolutionsoverthelocallyoptimalcoplanarsolutionswastheabilitytodeliverahigheramountofdosetothetargetstructures.Boththenon-coplanarandlocallyoptimalcoplanarsolutionswereabletoobtaintreatmentplanswithorgansparingthatiscomparabletoorimproveduponthe7-beamequi-spacedcoplanartreatmentplans.Whiletheinclusionofnon-coplanarorientationsinBOOisusefulintermsofFMOvalueandtargetcoverage,theresultingimprovementsinthetreatmentplanmaynotalwaysbeclinicallysignicant.Withbetterparametertuningorneighborhoodstructure,itispossiblethattheAdd/Dropalgorithmcanobtaincoplanartreatmentplanswithmoredesirabletargetcoverage,thusmakingtheresponsesurfaceplansandtheAdd/Dropplansclinicallyequivalent.Thissuggeststhattheinclusionofnon-coplanarbeamorientationsdoesnotsignicantlyimprovethequalityofatreatmentplan.AlthoughmostBOOresearchisrestrictedtocoplanarorientations,therehasnotyetbeenastudyassessingthesolutionqualityofcoplanarversusnon-coplanarsolutions.Withthisstudyasevidence,bothresearchersandpractionersnowhaveabasisforrestrictingthesolutionspacetothesmaller,moretractablesetofcoplanarbeamsforhead-and-neckbeamoptimization.Thepatientcasesinthisworkwereallhead-and-neckcases.Dierenttumorsites,e.g.,breast,lungandprostate,couldalsobenetfromBOO,andperhapsmayexperiencegreaterimprovementsintreatmentplanquality.Infuturework,thesesiteswillbetestedtoassessthegeneralclinicalusefulnessofnon-coplanarorientationsandtheresponsesurfacemethod. 94

PAGE 95

3.9.2NeighborhoodSearchConclusionsWehaveshownthatforhead-and-neckcases,qualityplanswithfewerbeamsthanastandardtreatmentplancanbeobtainedifBOOisapplied.ThesimulatedannealingandAdd/Dropalgorithmsbothregularlyobtainedqualitytreatmentplanswithasfewasfourbeamsinonly30minutes.TheuseoftheipneighborhoodimprovestherateofFMOconvergenceinbothalgorithms,andevenhastheabilitytoimproveorgansparingasshowninthesimulatedannealingresults.ThesimulatedannealingandAdd/Dropalgorithmsperformedcomparablytoeachother,withneitheralgorithmindicatingasignicantbenetovertheother.ItispossibletoincorporateipneighborhoodsintootherBOOalgorithmsthatrelyonneighborhoodsearchestoyieldimprovedtreatmentplansinclinicallyacceptabletimeframes. 95

PAGE 96

CHAPTER4FRACTIONATION4.1IntroductionTypically,head-and-necktreatmentplanseachcontaintwotargetstructures,orplanningtumorvolumesPTV:PTV1andPTV2.LetPTV1bethetumormassobservedfromimagingscans,andletPTV2bePTV1plussomemarginspeciedbythephysician.Ratherthandeliveranentiretreatmentplaninonesession,atreatmentplanisdividedintoseveralsessions,calledfractions.Thisisdonetotakeadvantageofthefactthatnormal,healthycellsrecoverfasterfromtheradiationthancancerouscells.Toobtainthetreatmentplansforthefractions,inpractice,asingleFMOtreatmentplanisdevelopedandthendividedintothedesirednumberoffractions,usuallyaround35.Thisdivisionofatreatmentplanisanon-trivialtask,asthetargetvoxelsmustreceive1.8-2.0Gyofradiationineachfraction.WithasingleIMRTtreatmentplan,itispracticallyimpossibletodeviseaconstantdose-per-fractiondeliverytechniquebecauseonlyasingleFMOproblemissolvedtoobtainthetreatmentplan,whichisthensimplydividedintoanumberofdailyfractions.Ifasingleplanisoptimizedtodeliverdosestomultipletarget-doselevels,thenthedoseperfractiondeliveredtoeachtargetmustchangeintheratioofagivendoseleveltothemaximumdoselevel.Forexample,sayPTV1hasaprescriptiondoseof70Gy,PTV2hasaprescriptiondoseof50Gy,andthenumberoffractionsis35.Ifasingletreatmentplanisdividedamongthe35fractions,thenPTV1willreceive70=35=2:0Gyineachfraction,butPTV2willonlyreceive50=35=1:4Gy,andthusanycancerouscellsinPTV2maynotbeeradicatedbythetreatment.Similarly,ifonly25fractionsareusedinordertoensurethatPTV2receives2.0Gyperfraction,thenPTV1receives70=25=2:8Gyperfraction,wellabovethedesireddose.WeproposeanewmethodofapproachingthefractionationsubproblemwhereinanFMOtreatmentplanisdevelopedforeachtargetstructure,ratherthandevelopinga 96

PAGE 97

singletreatmentplanforalltargetstructures.Theindividualtreatmentplanscanthenbeeasilydividedintooptimalfractions.Theprimal-dualinteriorpointalgorithmpresentedbyAlemanetal.[ 88 ]isusedtosolvetheFMOandfractionationmodelstooptimality.4.2ModelFormulationThefractionationmodelbuildsontheFMOmodeldescribedinChapter 2 .Tosolvethefractionationproblem,weconsiderdevelopinganindividualuencemapsolutionforeachtarget.Foracasewithtwotargets,twoplansmustbedeveloped:1aplanthatdeliverstheprescriptiondosetoPTV1andPTV2,andaplanthatboosts"thedosereceivedbyPTV1toreachtheprescribeddoselevel.Thesetwouencemapscanthenbedividedintotheappropriatenumberoffractionseasily.Fortheexampleof50Gyand70GyprescriptiondosesforPTV2andPTV1,respectively,thiswouldyield25fractionsoftreatingbothPTV1andPTV2to50=25=2:0Gy,andanother10treatmentsoftreatingjustPTV1to0)]TJ/F15 11.955 Tf 11.955 0 Td[(50=10=2:0Gy.Forsimplicity,wecalltheseindividualuencemapsfractions",ratherthanusingthetermtodescribethedailytreatments.Thedevelopmentoftheseuencemapsseparatelywouldresultinsuboptimalsolutions.Tooptimizetheseuencemapsetssimultaneously,weconsidereachbixelineachfractionasanindividualdecisionvariable.AstherenumberoffractionsisequaltothenumberoftargetsT,thisresultsinauencemapdevelopedforeachtarget.InthesingleFMOformulation,dosetovoxeljinstructuresisdenedaszjs=PNi=1Dijxi,s=1;:::;T,andthepenaltyassociatedwithitasFszjs.Becausethefractionationmodelwillbeconcernedwithdose-per-fractionaswellascumulativedose,newvariablesmustbedenedtoexpressthesevalues. 97

PAGE 98

Denexfi,f=1;:::;T,astheuenceofbeamletiinfractionf.Theamountofdosereceivedbyavoxeljinstructuresinfractionfisdenedaszfjs=NXi=1Dijxfi;j=1;:::;vs;s=1;:::;T;f=1;:::;T{1Criticalstructuresarethoughttobeaectedbyonlythecumulativedosereceivedfromalltreatments,ratherthanthejustthedoseinanyoneparticularfraction.Thiscumulativedosereceivedbyavoxeliszjs=TXf=1NXi=1Dijxfi;j=1;:::;vs;s=T+1;:::;S{2CriticalstructuresarepenalizedinthesamemannerasintheoriginalFMOmodel,thatis,Fszjs,s=T+1;:::;S.Targetsrequireamorecomplextreatmentinthefractionationmodel.Ineachfraction,weareprimarilyconcernedwithdosereceivedbythetargetsinthatparticularfraction.Thus,newvariablesareneededtoexpresstheamountofdoseperfractionreceivedbyavoxelzfjsinEquation 4{1 .Sincewemustalsoensurethatthecumulativedosereceivedbyeachtargetreachestheprescribeddose,variablestoexpressthecumulativedosereceivedbyavoxelarerequired.Intuitively,thiscumulativedoseshouldbethesumofallthedosesreceivedinallfractions.Ifthecumulativedosefortargetsisdenedthisway,thenover/underdosinginonefractioncanresultinunder/overdosinginanothertocompensate,whichisundesirable.Topreventsuchascenario,anothernewvariablecalledthearticialdoseisrequiredzjsinEquation 4{3 .Ratherthansimplysummingupthedosereceivedineachfraction,wewillassumethatinthepreviousfraction,thetargetvoxelreceivedexactlythecorrectprescriptiondoseforthepreviousfraction.Thus,nocompensatingwillbenecessary.Thearticialdoseisjusttheprescriptiondosefromthepreviousfraction 98

PAGE 99

Pf)]TJ/F22 7.97 Tf 6.587 0 Td[(1plusthedosereceivedinthecurrentfraction:zjs=Pf)]TJ/F22 7.97 Tf 6.586 0 Td[(1+zfjsj=1;:::;vs;s=1;:::;T;f=1;:::;T{3Sinceeachofthetargetvoxelsbeingirradiatedinfractionfistreatedastargetf,thepenaltyfunctionsforthesevoxelsisTXs=fXj2VsFfzfjsOnceatargethasreceiveditsprescriptiondose,ideally,itshouldnotreceiveanyfurtherdose.Astargetfistreatedinfractionf,forallfractionsafterf,targetfshouldbetreatedasnormaltissue.Specically,targetsthatnolongerrequiredosewillbetreatedasskin,denotedstructureS.Therefore,thesetargetvoxels,alongwithactualskinvoxels,willbepenalizedwithpenaltyfunctionFS.ThedosereceivedbythesetargetvoxelsistheprescriptiondoseofthevoxelPsplusthedosereceivedinallsubsequentfractionsPT`=s+1z`js.Thisleadstothefollowingpenaltyfunctionsforvoxelspenalizedasnormaltissueinfractionf:f)]TJ/F22 7.97 Tf 6.587 0 Td[(1Xs=1Xj2VsFSPs+TX`=s+1z`js!+Xj2VSFSzjSAswiththetraditionalFMOmodel,penaltyfunctionsarenormalizedaccordingtothenumberofvoxelsinthestructure.Forcriticalstructures,thisnormalizationfactorisstill1=vssincetherearealwaysvsvoxelsbeingtreatedascriticalstructures.Ineachfraction,thenumberoftargetvoxelsdependsonwhichtargetsstillneedtobetreated.Eachuencemapsetwillonlysee"thetargetvoxelsthatareincludedinitsprescriptiondoselevel.Thus,denethenumberoftargetvoxelstreatedinuencemapfasvf=TXs=fvsf=1;:::;T 99

PAGE 100

Thenumberofvoxelstreatedasskinineachiterationcanbeexpressedbyv1)]TJ/F15 11.955 Tf 12.785 0 Td[(vf+vS,wherev1)]TJ/F15 11.955 Tf 12.512 0 Td[(vfisthenumberoftargetvoxelsbeingtreatedasskinandvSisthenumberofactualskin/unspeciedtissuevoxels.IdenticaltothetraditionalFMO,thecriticalstructuresarenormalizedandpenalizedbyS)]TJ/F22 7.97 Tf 6.586 0 Td[(1Xs=T+11 vsXj2VsFszjsLetzbeavectorofallzjs,zfjsandzfjsvariables.Theobjectivefunctionisobtainedbysummingthenormalizedpenaltyfunctions:Ffracz=TXf=11 v1)]TJ/F15 11.955 Tf 12.398 0 Td[(vf+vS"f)]TJ/F22 7.97 Tf 6.587 0 Td[(1Xs=1Xj2VsFSPs+TX`=s+1z`js!+Xj2VSFSzjS#+1 vfTXs=fXj2VsFfzfjs+S)]TJ/F22 7.97 Tf 6.587 0 Td[(1Xs=T+11 vsXj2VsFszjsThefractionationmodelisthenformulatedasminimizeFfraczsubjecttozfjs=NXi=1Dijxfij=1;:::;vs;s=1;:::;T;f=1;:::;Tzjs=TXf=1NXi=1Dijxfij=1;:::;vs;s=T+1;:::;Szjs=Pf)]TJ/F22 7.97 Tf 6.586 0 Td[(1+zfjsj=1;:::;vs;s=1;:::;T;f=1;:::;Tx0Astheobjectivefunctionisthesumofquadraticfunctionsandtheconstraintsarealllinear,thefractionationformulation,justlikethebasicFMOformulation.4.3ResultsThefractionationmodelistestedusingtheprimal-dualinteriorpointalgorithminAlemanetal.[ 88 ].Onesignicantbenetofemployingaprimal-dualinteriorpointalgorithmisthatthesolutiongeneratedisguaranteedtobeoptimaltowithinacertain 100

PAGE 101

tolerancethatcanbespeciedbytheuser.Thirteenhead-and-neckcasesusingveequi-spacedbeamsaretested.Eachtestcaseconsistsoftwotargets,PTV1andPTV2,withprescriptiondoselevelsof70Gyand50Gy,respectively.AccordingthesuggestionsmadeonalgorithmparametersinAlemanetal.[ 88 ],theprimal-dualinteriorpointalgorithmwasimplementedwithaSingleApproximationHessianandastoppingcriteriaofarelativedualitygapof0.1%.Althoughitwasalsorecommendtoremoveinsignicant"beamlets,theseremovalofthesebeamletsactuallyincreasesruntimeinthefractionationmodel.Thus,insignicantbeamletsareleftinthefractionationmodel.4.3.1ComputationalResultsThetestsareruninMatlabMathWorks,Inc.ona2.33GHzIntelCore2Duoprocessorwith2GBofRAM.Table 4-1 showsthesizesofeachcaseintermsofthenumberofdecisionvariablesthenumberofbixelsandthesizeofthepatientareabeingtreatedthenumberofvoxels.ThecomputationtimesobtainedaredisplayinTable 4-1 .Onaverage,thefractionationmodelwassolvedin22.03seconds.Withthesamealgorithmparametersandweightingparameters,asingleFMOtreatmentplancanbedeterminedinanaverageof16.28seconds,thusthereisonlya35%increaseincomputationtimerequiredtodeveloptwoFMOplansforthefractionationmodel.Thisrelativelysmallincreaseintimecouldbeaccountedforbythefactthattheweightingparametersusedintheobjectivefunctionwerespecicallytunedforthefractionationmodel.Usingparametersspecicallytunedtothesingle-FMOmodel,thesingle-FMOmodelcanbesolvedonaveragein9.36seconds.Comparedtothisaverageruntime,theFMOmodelrequires2.4timesasmuchcomputationtimetodeveloptwomodelsasopposedtoone,whichisamoreintuitiveexpectationoftheinteriorpointmethod'sperformance. 101

PAGE 102

Table4-1.Casesizesandruntimesusingidenticalalgorithmandweightingparameters. SingleFMOFractionationCaseBixelsVoxelsIterationsTimesIterationsTimes 181385,017168.391619.6021320189,23410382.691455.34393586,2552411.751118.79469258,636156.871111.4751044102,2621413.161229.706100584,3691310.311225.58782271,873179.141418.88880292,3075922.921420.19991165,5411810.841726.121064266,634257.941612.441127956,847292.75142.991299496,1051712.301227.131382372,7293312.551418.15 Average85286,7552916.281422.03 4.3.2ClinicalResultsBecausethereisnofundamentalwayofquantifyingatreatmentplan,DVHsareexaminedinadditiontoobjectivefunctionvaluestoassessthequalityofatreatmentplan..Theprescriptiondosesusedare70GyforPTV1and50GyforPTV2.ThesearecommonprescriptionsusedinthecancercenteratShandsHospitalattheUniversityofFlorida.Figures 4-1 4-7 showbothdosevolumehistogramsDVHsandaxialslicesforseveralcases.TheDVHsshowthatintherstfraction,bothPTV1andPTV2aretreatedto50Gy,andinthesecondfraction,onlyPTV1istreatedtoanadditional20Gy.Theprescriptiondoseforthefractionismarkedbyaverticalline.Theamountofdosereceivedbyeachtargetineachfractionisclinicallyacceptable.Asthisstudyfocusesonhead-and-neckcaseswherethemostconictliesintreatingthetargetswhilesparingthesalivaglands,onlyDVHsofthesalivaglandsareshown.Allotherorgans,includingskin/unspeciedtissue,receivealowenoughamountofdosetobesparedinthetreatment.Thesparingcriteriaforeachofthecommoncriticalstructuresinhead-and-neckcasesarelistedinTable 4-2 .Thecriticalstructuresinvolvedineachcase 102

PAGE 103

Table4-2.Sparingcriteriavariesforeachcriticalstructure StructurePercent%DoseGy brainstem10055eyes5030mandible10070opticchiasm10055opticnerves10050parotidglands5030skin10060spinalcord10045submandibularglands5030 vary,dependingontheirproximitytothetumor,andthusDVHsforsomecasesdonotincludeallsalivaglands.DVHsofthesalivaglanddosesinFraction1showthatthesalivaglandsreceivethemajorityofdoseintherstfraction.Becausethecumulativeamountofdosereceiveddetermineswhetherornotcriticalstructurescanbespared,theDVHsforFraction2depictthecumulativedoseoftheseorgans.Thesparingcriteriausedforsalivaglandsisthatnomorethan50%oftheglandcanreceivemorethan30Gy.Thispointismarkedasastar.Formostcases,allofthesalivaglandsarespared.Figures 4-1 4-7 alsoshowthedosereceivedineachfractionasacolorwashofasliceofthepatient.Fraction1deliversahomogeneousdoseof50GytobothPTV1andPTV2whilegenerallyavoidingoverdosinganyofthemarkedcriticalstructures.InFraction2,thedosetoPTV1isboostedby20Gywithoutdeliveringanyunnecessarydose.4.3.3SpatialCoecientResultsTheconceptofemployingspatialinformationasdescribedinSection 2.4 isalsoappliedtothefractionationmodel.Onesetofspatialcoecientsisusedtoobtainbothfractions.Forthefractionationtreatmentplans,thespatialcoecientsare=1:02,=)]TJ/F15 11.955 Tf 9.299 0 Td[(0:92,=0:97andtheminimumcoecientfortargetvoxelsis0.6.Generally,theDVHsforbothtargetsandcriticalstructuresusingspatialcoecientsaresimilartothoseobtainedwithoutusingspatialcoecients.Infact,inthecasestested, 103

PAGE 104

Figure4-1.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2right. 104

PAGE 105

Figure4-2.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2right. 105

PAGE 106

Figure4-3.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2right. 106

PAGE 107

Figure4-4.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2right. 107

PAGE 108

Figure4-5.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2right. 108

PAGE 109

Figure4-6.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2right. 109

PAGE 110

Figure4-7.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2right. 110

PAGE 111

therewerenoinstancesofeitherthespatialtreatmentplansorthenon-spatialtreatmentplansyieldingclinicallysignicantchangesintheDVHs.Theslicesshowthatthereisimprovedhomogeneityinthetargetdoseswhenspatialcoecientsareused.Theslicesalsoindicatethatsincetheuseofspatialcoecientsresultsinthetargetvoxelsweighingmoreheavilythanothervoxels,themodelismorewillingtodeliverdosetocriticalstructuresratherthanoverdoseorunderdosethetarget.Thishelpsprovideauniformdoseinthetarget,andshouldstillbeacceptableasthecumulativedoseforallcriticalstructuresremainswithinacceptablelevelsandtherearenoinstancesofsacricingorgansthatwerenotalreadysacricedinthenon-spatialplan.Becausemorecriticalstructurevoxelsreceivedoseinthespatialplans,thedosedepositedinthetargetstructuresismorespreadout,andthusthemaximumdosereceivedbythecriticalstructurevoxelsislessthaninthenon-spatialplans.Thisofcoursemeansthatmorevoxelsareexposedtoradiation,butthelevelsarelowerandtheamountofradiationstillfallswithinclinicallyacceptablelimits.Theresultingimprovementinhomogeneityisevidentforeachofthecases,buttheeectofthemorespreadoutdoseisbestillustratedinthesecondfractionofeachcase.Figures 4-8 { 4-14 showtheDVHsandslicesforsomeofthetestedcases.Inparticular,Figures 4-9 4-10 4-11 and 4-14 demonstratethatthespatialcoecientsreducetheamountofdosedeliveredoutsideofthetargetswhencomparedtotheirrespectivenon-spatialplansinFigures 4-2 4-3 4-4 and 4-7 .4.4ConclusionsandFutureDirectionsThefractionationmodelpresentedallowsforthecreationofguaranteedoptimaluencemapsforeachfractionofapatient'streatment.Theseuencemapscanbeeasilydividedintotheappropriatenumberoffractionswithoutsacricingoptimality.Usingtheprimaldualinteriorpointmethod,thefractionationmodelobtainsuencemapsforeachtargetinaclinicallyfeasibleamountoftime.Asexpected,thecomputationtimerequiredtogeneratetwouencemapsforatwo-targetcaseismorethanthetimenecessaryto 111

PAGE 112

Figure4-8.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2rightusingspatialcoecients. 112

PAGE 113

Figure4-9.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2rightusingspatialcoecients. 113

PAGE 114

Figure4-10.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2rightusingspatialcoecients. 114

PAGE 115

Figure4-11.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2rightusingspatialcoecients. 115

PAGE 116

Figure4-12.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2rightusingspatialcoecients. 116

PAGE 117

Figure4-13.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2rightusingspatialcoecients. 117

PAGE 118

Figure4-14.TargetDVHs,salivaDVHsandaxialslicesinFractions1leftand2rightusingspatialcoecients. 118

PAGE 119

generateasingleFMOplan,butthecomputationtimesarestillacceptable.Furtherparametertuningcouldpossiblyyieldbetterresults.Theadditionofspatialcoecientsinthemodelallowsforimprovedhomogeneity,butdoesnotseemlikelytoprovideadditionalorgansparing.Theimprovedhomogeneityaloneisenoughtowarranttheinclusionofspatialinformationinthemodel.Themodelissensitivetothechangesinthespatialcoecients,sofurtherparametertuningwillhavetobeperformedinsmallincrementalchanges.Currently,themodelassumesthatpriortoeachfraction,eachtargetvoxelhasreceivedexactlytheprescribedamountofdoseuptothatpointintime.Whilewehaveassumedthatover/underdoseinonefractionshouldnotbecompensatedbyunder/overdoseinanotherfraction,itmayinfactbeadvantageoustoallowforsomedegreeofcompensation.Thefractionationformulationproposedaordsenoughexibilitytomodelsuchascenario.Forexample,sayaphysicianwouldliketoallowunderdoseintargetsinpreviousfractionstobecompensatedbyuptoGyofoverdoseinthecurrentfraction.Then,fortargetstructures,thePstermintheobjectivefunctionwouldbereplacedbytheexpressionmaxfzjs;Ps)]TJ/F23 11.955 Tf 12.03 0 Td[(g.Asthistypeofdiscontinuityalreadyexistsinthemodel,thestructureofthemodelwouldnotbealteredbymakingthismodication.Otherfutureresearchpossibilitiesincludefurtherparametertestingtoemploythemodelonothercancersitetreatments,forexample,lungandprostatecancers. 119

PAGE 120

CHAPTER5AMONTECARLOMETHODFORMODELINGDOSEDEPOSITION5.1IntroductionTheFMOproblemreliesonthecalculationoftheamountoftotalradiationdosereceivedineachvoxel.Thedoseinavoxelisdeterminedbythepathsthephotonsintheradiationbeamsfollowthroughthepatient.Somephotonsmaycollidewithparticlesinsidethepatientandscatterinanydirection,whileothersmaycollidewithparticlesandbeabsorbed.Stillotherphotonsmaypassentirelythroughthepatientwithnocollisions.Duetotheunpredictablenatureoftheradiationbeaminsidethepatient,thedosereceivedinavoxelcanonlybeaccuratelyobtainedthroughMonteCarlosimulations.AsimplelinearrelationshipisassumedbetweentotaldoseandbeamletuencesandiscommonlyacceptedasasatisfactorydoseapproximationinIMRToptimization.Errorsofasmuchas30%havebeenreportedforphotonbeamsneartissueinhomogeneitiesMaetal.[ 5 ].ForIMRToptimization,particularlywithadventofimage-guidedIMRTIGIMRT,or4DIMRT,theFMOproblemmustbesolvedextremelyquicklytocreatereal-timetreatmentplans.Thus,thespeedoftheFMOproblemisparamount.WhileMonteCarlosimulationmayprovidethemostaccuratemeasureofdose,thelengthycomputationtimerendersthemethodimpracticalforclinicaluse.WeproposeaMonteCarlomethodthatperformsalimitednumberofhistoriestoobtainanoisyapproximationofthedosedistributionofeachbeamlettowhichasmoothingfunctioncanbeappliedinordertodetermineanaccuratedosedistribution.Theanticipationisthatfewhistorieswillberequired,andthatthisapproachcanbeclinicallyfeasible.Recently,asimilarapproachhasbeentakenbyJelenandAlber[ 89 ]andJelenetal.[ 90 ]withgoodresults.Jelenetal.[ 90 ]acknowledgethatthereissomelossofaccuracyatthebeam'sedgeduetoalackoflateraldensitycorrectionandtheeectsarisingfromMLCsystems,forexample,tongue-and-grooveandinter-leafscatter.JelenandAlber 120

PAGE 121

[ 89 ]pursuetheissueofdensityscaling,buttheMLCeectshavenotyetbeenaddressed.Section 5.6 proposessomepossiblemethodsofaccountingforsuchMLCeects.5.2MonteCarloEngineTheDosePlanningMethod"DPMSempauetal.[ 91 ]programwillbeusedtoperformtheMonteCarlosimulations.DPMisdesignedtosimulatethetransportofphotonsinradiotherapyclassproblems.DPMisprimarilybasedonthepublicdomaincodePENELOPEBaroetal.[ 92 ],Sempauetal.[ 93 ].Thisstudyfocusesonmodelinganitesizedpencilbeamemanatingfroma6MVlinearaccelerator.Anitesizedpencilbeamisabeamofnitesizedthatisparalleltothepointsourceofradiation.Todetermineareasonablyaccuratemeasureofthedoseofasinglebeamletinagiventissue,approximatelyonebillionhistoriesareruninDPM.Asfewerhistoriesarerun,theinaccuraciesofthedoseresultingfromtheMonteCarloexperimentgrow.Figure 5-4 showshowthenoiseinthedepth-dosecurveofthebeamletbecomesincreasinglypronouncedinrelationtothenumberofhistories.AsshownbyTable 5-1 ,theamountoftimerequiredtoruneachexperimentisapproximatelylinearinthenumberofhistoriesrecorded.Thus,itisimpracticaltorunthenumberofhistoriesnecessaryforacceptableaccuracy.5.3DoseDistributionofaBeamletTheaccuracyofatreatmentplaniscontingentupontheaccuracyofthecalculateddosedepositedbyeachbeamletintheplan.Becausetheparticlesinabeamletscatterinthreedimensionalspace,multipledosedistributionsmustbeconsideredtosatisfactorilymodelthebeamlet'saectonthepatient'stissue.Thesedistributionsarisefromtheamountofradiationthebeamletdepositsasafunctionofdepththedepth-dosecurve,andfromtheamountofradiationradiatingoutwardfromthecenterofthebeamletthelateralpenumbra. 121

PAGE 122

5.3.1Depth-DoseCurveThedepth-dosecurverepresentstheradiationintensitydepositedbythebeamletinthetissuethroughwhichitpassesasfunctionofdepth.Figure 5-1 showsthedosedistributionofasingle6MVbeamletinvarioustissuesobtainedfromtheDPMsimulations.Thedosedistributionofabeamletinwaterisempiricallyknown,andtheresultsfromtheDPMsimulationinwatercanbeeasilyveriedtobecorrect.Muscle,whichhasnearlyidenticaldensityaswaterthedensitiesofmuscleandwaterare1.04g/cm3and1.00g/cm3,respectively,hasnearlythesamedepth-dosedistributionaswater.Asexpected,abeamletpassingthroughlungtissue,whichissignicantlylessdensethanwater,doesnotloseitsintensityasquicklyasittravelsthroughthelessdensetissue.Lastly,asimulationwithinhomogeneoustissueisconsidered.Asimulationofmusclewitha10-cmthicklayeroflunglocatedatadepthof10cmshowsadosedistributionthatwhenthebeamletreachestheless-densesegmentoflung,itsdepth-dosecurvebecomeslesssteep,indicatingthatlessradiationintensityislostthroughthelungthanthroughthemuscle.Oncethedeeperlayerofmuscleisreached,thesteepnessofthedepth-dosecurveincreasesagain. Figure5-1.Dosedistributionofasinglebeamletinvarioustissues. 122

PAGE 123

Althoughitmayseemunintuitivethatthedepth-dosecurveincreasesatshallowdepths,thisbehavioriscalledthebuild-upcurve,andisexplainedbythelikelihoodofelectronsscatteringoutofthetissueandintoairatshallowdepths.Becausethedensityofairisextremelysmall,anelectronthatreachesairislikelytotravelveryfarawayfromthetissue,andthereforeunlikelytoreturntothetissueanddepositradiationdose.Oncethedepthincreasespassesacertainpoint,theelectronscannotleavethetissueandtheamountofdosereceivedinthetissueincreases.Oncethatpointisreached,theamountofradiationdeliveredbythebeamletdecreasesmonotonicallyindepth,aswouldbeexpected.5.3.2LateralPenumbraInadditiontothedosedistributionoccuringasthebeamletpenetratesthetissue,thereisadosedistributionspreadingawayfromthebeamlet.Justaslightemanatingfromaashlightinadarkroomdoesnothaveadiscreteboundarybetweenlightanddark,theradiationdeliveredbyabeamletalsodoesnothaveadiscreteboundarybetweenwhatisandisnotirradiated.Withacircularashlightbeamshownontoaatsurface,itisapparentfromthedistributionoftheilluminatedportionofthesurfacethatsomeofthelightisdiusedintothesurroundingdarknessasaresultofscatter.Ifthedistributionoflightinthecircularprojectionoftheashlightbeamisplotted,abell-shapedcurvedescribesthebrightestpointinthecenteroftheilluminateddiscdecreasinginbrightnessastheedgeoftheilluminateddiscisapproached,eventuallyreachingcompletedarkness.Thisbehaviorisparalleltothebehaviorofabeamletpassingthroughanymedium.FromThePhysicsofRadiationTherapy[ 94 ],thepenumbraofabeamistheregionattheedgeofaradiationbeam,overwhichthedoseratechangesrapidlyasafunctionofdistancefromthebeamaxis.Hence,thedistributionofradiationdoseoriginatingfromthebeamletdescribedaboveiscalledthelateralpenumbra.Figure 5-2 showsthecolorwashofdosedistributionconsistitutingthelateralpenumbra,whileFigure 5-3 showsthedose 123

PAGE 124

Figure5-2.Colorwashofthelateralpenumbraofanitesizedpencilbeam distributionofthelateralpenumbraataxeddepthinonedimensionobtainedfromonebillionMonteCarlohistoriesofa5-cmnitesizedpencilbeaminwater.5.4MethodologytoModelaBeamletModelingthedosedistributionofabeamletisrelativelystraightforwardforabeamletinasinglemedium.Thedicultyariseswhenmultiplemediumsaretraversedbythebeamletbecausethevaryingdensitiesaecttheparticlescatteringofthebeam,thusaectingboththedepth-dosecurveandthelateralpenumbra.Aspreviouslystated,errorsofasmuchas30%havebeenreportedforphotonbeamsneartissueinhomogeneitiesMaetal.[ 5 ].Becausetherearenumerousinhomogeneitiesinmostcancertreatmentsites,theseinhomogeneitiesareofparticularinterest.Thebeamlet'sbehaviourattheboundaryofdierenttissuetypescannotbedeterminedaseasily,andthusrequiresMonteCarlosimulation.IndesigninganIMRTtreatmentplanforapatient,therecanbemorethanadozendierentstructurestissuetypeswithcomplicatedboundarygeometries. 124

PAGE 125

Figure5-3.Plotofthelateralpenumbraofanitesizedpencilbeam Knowledgeofabeamlet'sbehaviourgivencertaintissueinhomogeneitiescanbeveryusefulinaccuratelydeterminingdoseinavoxel.5.4.1ModelingtheDepth-DoseCurveInthesection,weanalyzethebehaviorofthedepth-dosecurveunderbothsingletissueandmultipletissuescenarios.ThegoaloftheanalyzationistodeterminetheminimumnumberofMonteCarlohistoriesrequiredtoobtainareasonablyaccurateapproximatingfunctionofthedosedepositedateachdepthinthetissue.Forboththeinstancesofonlyasinglemediumandmultiplemediums,thisisdonebyttingthedepth-dosecurvefromMonteCarloexperimentswithvaryingnumbersofhistoriestohigh-degreepolynomialfunctions.Thepolynomialtsarethencomparedtothepolynomialtofaveryaccuratemeasureofthedepth-dosecurveobtainedfromannumberofMonteCarlohistoriesacceptedtobesatisfactorilyaccurate.ThenumberofhistoriesrecordedintheMonteCarlosimulationcanhaveadrasticeectontheaccuracyofthedatacollected.Forexample,Figure 5-4 demonstratesthevast 125

PAGE 126

Figure5-4.Observeddepth-dosecurveinwaterforseveralhistories. variationobservedinthedepth-dosecurveofabeamletinwaterforhistoriesrangingfromonemilliontoonebillion.Itishopedthatafteracertainnumberofhistories,thefunctionapproximationofthedatawillcloselyfollowthefunctionapproximationofveryaccuratedataobtainedfromalargenumberofhistories.Forabeamletinbothhomogeneousandheterogeneoustissue,thedepth-dosecurvecanbemodeledusingapolynomialfunctionoforderk.Althoughthedepth-dosecurvemayexhibitchangesinconcavityinthepresenceoftissueinhomogeneity,ahighdegreepolynomialwillcapturethecurve'sbehavior.Thevariationofak-degreepolynomialttedton-historyMonteCarlodataismeasuredbyvk;n;n0=dn0)]TJ/F39 11.955 Tf 11.955 0 Td[(pk;n2;wheredn0istheactualobserveddepth-dosecurvefromn0MonteCarlohistoriesandpk;nisthevectorofapproximateddepth-dosevaluesobtainedfromapolynomialtof 126

PAGE 127

Figure5-5.Polynomialtsofseveralhistoriescomparedtotheobserved1B-historydepth-dosecurveinwater. degreektodataobtainedfromnMonteCarlohistories.Itisdesirabletohavethatn0>ntoassessthequalityofthepolynomialtcomparedtomoreaccuratedata.Inthisstudy,theaccuracyofthepolynomialobtainedisjudgedbyitsvariationfromtheobserveddatafromaverylargenumberofMonteCarlohistories,thatis,n0>>ninthecalculationofvk;n.Figure 5-5 showsthatfortheillustratednumberofhistories,thepolynomialtfrom100millionhistoriescloselyresemblesnotonlythepolynomialtfromonebillionhistories,butalsotheactualdatacollectedfromonebillionhistories.Thepolynomialttoonemillionhistoriesisclearlyanunsatisfactoryapproximationtothedatacollectedfromonebillionhistories.ForseveralnumbersofMonteCarlohistories,thebestapproximatingpolynomialfunctionwithdegreeintherange[k ;k]isfound,thatis,k=argmink2[k ;k]fvk;ng.Severaldegreesaretestedbecausethedegreeofthepolynomialcansignicantlyaectthequalityofthet,evenforpolynomialsthatareonlyonedegreeapart.Figure 5-6 illustratestheamountofvariationobservedinthepolynomialapproximationasafunctionofthedegree 127

PAGE 128

Figure5-6.Variationofpolynomialtasfunctionofdegree. ofthepolynomialforpolynomialsttedtothedepth-dosecurveofabeamletinwaterobtainedfrom1billionhistories.5.4.2ModelingtheLateralPenumbraInthesection,weanalyzethebehaviorofthelateralpenumbraunderbothsingletissueandmultipletissuescenarios.Thelateralpenumbraofabeamisabell-shapedcurvethatcanbeapproximatedasthesumoferrorfunctionpairs.Theerrorfunction,erfx,istwicetheintegraloftheGaussiandistributionwithmean0andvarianceof1/2:erfx=2 p Zx0e)]TJ/F24 7.97 Tf 6.586 0 Td[(t2dt:Figure 5-7 Ademonstratesasampleerrorfunction.Whileasinglesideofthelateralpenumbraofabeamletresemblesanerrorfunction,acloserapproximationtoasinglesideofthelateralpenumbraisrepresentedastheaverageoftwoerrorfunctionsgivenbya 2erfx+x0 )]TJ/F15 11.955 Tf 11.955 0 Td[(erfx)]TJ/F23 11.955 Tf 11.955 0 Td[(x0 ; 128

PAGE 129

ABFigure5-7.Anerrorfunctionandanerrorfunctionpair.AErrorfunction.BErrorfunctionpair. whereaistheamplitude,x0istheosetandisthevariationofthetwoerrorfunctions.Theexpressionisdividedby2totaketheaverageoftheerrorfunctionpair.AnexampleofanerrorfunctionpairisgiveninFigure??B.Becausethelateralpenumbraofabeamletresemblesanerrorfunctiononboththeleft-andright-handsidesofthebeamcenter,thelateralpenumbraLxisrepresentedasthesumoftheaverageofNerrorfunctionpairs,givenbyLx=NXi=1ai 2erfx+x0i i)]TJ/F15 11.955 Tf 11.955 0 Td[(erfx)]TJ/F23 11.955 Tf 11.955 0 Td[(x0i i;whereaiistheamplitude,x0iistheosetandiisthevariationoferrorfunctionpairi,i=1;:::;N.Todeterminetheparametersai,x0iandiforeachoftheNerrorfunctionpairs,aLevenberg-Marquardtquasi-Newtonminimizationmethodisemployed.ThismethodtakesasinputNandaninitialguessoftheparametersandreturnsalocallyoptimalsolutiontotheproblemofminimizingthevariationbetweentherealdataandthesumoftheerrorfunctionpairs.Atagivendepthinthetissue,theamplitudeoftheerrorfunctionisdeterminedbythevalueofthedepth-dosecurveatthatdepth.Thus,foreachtissuetype,itisonly 129

PAGE 130

Figure5-8.LateralpenumbraforseveralnumbersofMonteCarlohistories. necessarytomodelasinglelateralpenumbra,andthenthatmodelcanbeextendedtoalldepthssimplybymanipulatingtheamplitudeaccordingtothedepth-dosecurve.Figure 5-3 showsthelateralpenumbraofa5-cmnitesizedbeamletataxeddepthinwaterforanumberofMonteCarlohistoriesdeemedtoyieldasatisfactorilyaccuraterepresentationofthedosedepositedinthetissue.Usingthemethoddescribedabove,thelateralpenumbrawasmodeledtoyieldtheapproximationtotheobserveddatacollectedforthevariousMonteCarlohistoriesshowninFigure 5-8 .Inasimilarfashiontothemethodformodelingthedepth-dosecurve,themethodformodelingthelateralpenumbraconsistsofttingthesumoferrorfunctionpairstothelateralpenumbradata.ThequalityofthesetsisjudgedbytheirvariationfromtheobserveddataforasucientlylargenumberofMonteCarlohistoriestoobtainaccuratedoseinformation. 130

PAGE 131

Figure5-9.Errorfunctiontsofseveralhistoriescomparedtotheobserved1B-historylateralpenumbraofabeamletinwater. Justinasthemethodfordeterminingthequalityofthedepth-dosecurveapproximation,thevariationoftheerrorfunctiontfromtheactuallateralpenumbraiscalculatedasn;n0=Ln0)]TJ/F15 11.955 Tf 13.069 3.055 Td[(^Ln;N2;whereLn0istheobservedlateralpenumbradatafromasimulationofn0histories,and^Ln;NistheapproximatedlateralpenumbraobtainedfromtheparametersttedtotheexpressionLNx.Itisdesirabletohavethatn0>n.Figure 5-9 displaystheerrorfunctionpairtsobtainedfromtheLevenburg-Marquardtmethod,aswellasthevariationofthetsfromtheobserveddatafromonebillionhistories.ThevariationismeasuredinthesamemannerasdescribedinSection 5.4.1 .Itisanticipatedthatalthoughthelateralpenumbraexhibitsdierentdosedistributionsinmaterialsofdierentdensities,thedistributionwillonlyshowafundamentalchangeinshapeifthebeamsimultaneouslyhitsmultipletissuesofvaryingdensities.Insuchasituation,thepenumbra,whichistakentobesymmetricaboutthecenterofthebeamin 131

PAGE 132

Table5-1.ComputationtimesinminutesofMonteCarlosimulations nWaterMuscleLungMuscle-Lung-Muscle 1e9222.184211.887111.318186.894100e620.54321.25611.23918.70110e62.2102.2341.2691.9861e60.2440.3390.2330.309 homogeneoustissue,willnolongerbesymmetric.Tomodelthelateralpenumbraunderinhomogeneousmaterial,asumoferrorfunctionpairscanstillbeemployed,thoughitmaybenecessarytoincreasethenumberoferrorfunctionpairsrequired.Thedicultywilllieincorrectlydeterminingwhentheadditionofadditionalerrorfunctionpairwillbeneeded.Apossiblemeasurecouldbethevariationbetweenthelateralpenumbraapproximationandtheobserveddata.5.5ResultsThehomogeneoustissuestestedarewater,muscleandlung,andtheheterogeneousmaterialtestedconsistsofmuscleandlung.Eachscenarioisconsideredtohaveadepthof30cm.Thevoxelsizesare5mm5mm5mm,anda5-cmnitesizedpencilbeamisconsidered.Foreachsimulation,testswererunwith1billion,100million,10millionand1millionMonteCarlohistoriesinDPMonaMacOSX10.4.6machinewithdual2.3GHzPowerPCG5processorsand8GBofRAM.Duetotimeconstraints,themuscletestsareruntoamaximumof100millioniterations,andallcomparisonstothetqualityaremadetothis100-million-historydatainsteadofthe1-billion-historydatausedfortheothersimulations.AscanbeseenfromthecomputationtimesinTable 5-1 ,theruntimeofDPMisapproximatelylinearinthenumberofhistories.AltoughalargernumberofMonteCarlohistoriesyieldsimprovedaccuracy,themaximumnumberofhistoriesconsideredisonebillionbecauseoftimelimitationsandthesatisfactoryaccuracyofthe1-billion-historyruns. 132

PAGE 133

Foreachofthetestedtissuetypes,thedepth-dosecurvesandlateralpenumbrasweremodeledusingthemethodsdescribedinSection 5.4 .Forthepolynomialtsofthedepth-dosecurve,thevaluesk andkarechosenas10and45,respectively.Bychoosingthepolynomialapproximationoversuchalargerangeofdegreevalues,anacceptablyaccuratetislikelytobefound.Forthelateralpenumbra,Nwaschosenas4becauseinadditiontotheobviousneedfortwoerrorfunctionstomodelthesidesofthelateralpenumbra,anadditionalerrorfunctionisneededtomodeleachtailwithreasonableaccuracy.Forexample,thefourerrorfunctionsusedtomodelthelateralpenumbraofabeamletinwaterFigure 5-9 areshownseparatelyinFigure 5-10 .ThecomputationtimesrequiredtoobtaineachofthefunctionapproximationsaredisplayedinTable 5-2 .Theinitialparametersai,x0iandiforeacherrorfunctionpairi,i=1;:::;N,usedtoapproximatethelateralpenumbraareobtainedbythefollowingmethod.Ofthefourerrorfunctionpairsconsidered,twooftheerrorfunctions|I=f1;2g|areusedtomodelthesteepsidesofthelateralpenumbra,andtheothertwoerrorfunctions|I=f3;4g|areusedtomodelthetailsofthedosedistribution.Atagivendepthz,theamplitudeaiisai=8><>:dzi2Idz=50i2I;wheredzrepresentsthevalueofthedepth-dosecurveapproximationatadepthz.Theexpressionfortheamplitudewheni2Iwasobtainedbyexperimentingwithseveraldierentfractionsofdz.Thevalueoftheerrorfunctionsdeterminestheshapeoftheerrorfunctioncurve.Asincreases,thecurvebecomesincreasinglyspreadout.Thus,itisdesirabletohaveasmallivaluefori2IsincetheerrorfunctioninIonlyneedtomodelthesidesofthelateralpenumbra,andalargerivaluefori2IsincetheerrorfunctioninIneedtomodeltheelongatedtailsofthelateralpenumbra.Forthetissuestested,theivalues 133

PAGE 134

Table5-2.Computationtimesinsecondsofapproximatingfunctiontstothedosedistribution.ThepolynomialtstothedepthdosecurvearerepresentedbyD.D.,andtheerrorfunctiontstothelateralpenumbraarerepresentedbyLat.Pen. WaterMuscleLungMuscle-Lung-MusclenD.D.Lat.Pen.D.D.Lat.Pen.D.D.Lat.Pen..D.D.Lat.Pen. 1e90.0782.6400.0782.4220.0941.0620.078n/a100e60.0781.1720.0782.6250.8280.9060.109n/a10e60.1103.4540.1091.3902.6092.5940.093n/a1e60.0941.4070.0941.1721.0630.9530.078n/a usedarei=8><>:0:4i2I0:8i2I;Thesevalueswereobtainedthroughexperimentation.Forthe5-cmnitesizedpencilbeamsusedinthisexperiment,theosetsx0iwereempiricallysetatvaluesof8.5,-3.5,11and-1fori=1;:::;N,respectively.AmethodofidentifyingthelocationsoftheseosetsbasedontheMonteCarlodatacanbedevelopedbybasingtheosetsontheslopeoftheobserveddata,andisplannedforfutureresearch.Theresultsforthetsofboththedepth-dosecurveandthelateralpenumbraofabeamletinwaterareshownintheexamplesinSection 5.4 .Figures 5-11 5-12 showtheresultsofthetsforthemuscleandlungtissues.Fromthecomputationalresults,itisclearthatthetimetoobtaintstotheMonteCarlodataisinsignicantcomparedwiththeamountoftimerequiredtoruntheMonteCarlohistories,evenforasfewas1millionhistories.Totestthemodelinthepresenceoftissueinhomogeneity,a10cm-thicklayeroflungbetweentwo10cm-thicklayersofmuscleisconsidered.Asexpected,fortherst10cm,thedepth-dosecurveofthemuscle-lung-musclecaseisidenticaltothatofthemuscledepth-dosecurve.Oncethebeamletreachesthesignicantlylessdenselayeroflunglunghasadensityof0.30g/cm3,apredominantchangeinthedepth-dosecurveisevidentFigure 5-1 .Oncethelayeroflungisreached,therateofdecreaseintheamountofdosedepositedinthetissuedecreases,thatis,lessradiationintensityislostasthebeamlet 134

PAGE 135

Figure5-10.Errorfunctionpairssummedtoapproximateabeamletinwater. ABFigure5-11.Depth-dosecurvesinmuscletissue.AMonteCarlohistories.BPolynomialts. 135

PAGE 136

ABFigure5-12.Lateralpenumbracurvesinmuscletissue.AMonteCarlohistories.BErrorfunctionts. ABFigure5-13.Depth-dosecurvesinlungtissue.AMonteCarlohistories.BPolynomialts. 136

PAGE 137

ABFigure5-14.Lateralpenumbracurvesinlungtissue.AMonteCarlohistories.BErrorfunctionts. passesthroughthelungtissue.Whenthebeamletreachesthesecondlayerofmuscle,thisrateincreasesagain.Thesameapproachusedtomodelthedepth-dosecurveinasingletissuecontinuestoworkwellinmultipletissue.Figures 5-15 Aand 5-15 Billustratetheabilityofapolynomialtoapproximatethedepth-dosecurveininhomogeneoustissue.Becausetestingthebeamletinascenariowhereitcouldhitmultipletissuessimultaneouslyisreservedforfutureresearch,resultsformodelingthelateralpenumbrainthemultiple-tissuescenariotestedareidenticaltothoseforthesingle-tissuescenario.Thelateralpenumbraatagivendepthinacertaintissuecanbemodeledbyusingthedosefromthedepth-dosecurveatthegivendepthastheamplitudeofthelateralpenumbra.Thedosedistributioninthelateralpenumbracanthenbemodeledaccordingtothesameerrorfunctionpairsusedinmodelingthelateralpenumbrainasingle-tissuescenarioofthesamemedium.Figure 5-16 illustratesthevariationsofthetsusedtoapproximatethedepth-doseandlateralpenumbradistributionsofabeamletinwaterasafunctionofthenumberofhistories.Fromthisdata,itisveryclearthattheaccuracyofthebeamletmodelisdirectlycorrelatedwiththenumberofMonteCarlohistories.Itisinterestingthatthereisnotasignicantimprovementinthebeamletmodelaccuracyfrom100millionto1 137

PAGE 138

ABFigure5-15.Depth-dosecurvesinheterogeneousmuscleandlungtissue.AMonteCarlohistories.BPolynomialts. Table5-3.Variationoftstoseveralnumbersofhistorieswithn0=1billion. Water Muscle Lung Muscle-Lung-Musclenvk;n;n0n;n0vk;n;n0n;n0vk;n;n0n;n0vk;n;n0n;n0 1e90.0500.0710.0460.1050.0570.0970.052n/a100e60.0800.0750.0750.1180.1100.1030.101n/a10e60.2190.1450.2030.1240.3300.1250.213n/a1e60.5411.1830.6980.2091.0550.1290.881n/a billionhistories,andcomputing100millionhistoriesrequiresapproximatelyonetenthoftheamountoftimeascomputing1billionhistories.Dependingonthecompositionofthetissue,itmaybereasonablyaccuratetoonlyrequire10millionhistories,particularlyinthedepth-dosecurveapproximation.5.6ConclusionsandFutureDirectionsInconclusion,theMonteCarloapproachpresentedisemployedtomodelthedosedistributionofabeamletusingalimitednumberofhistories.Usingthepolynomialanderrorfunctionpairttingtechniquesdescribed,dosedistributionswithsatisfactoryaccuracycanbeobtainedusingatleastafactorof10fewerMonteCarlohistoriesthanwouldotherwiseberequired.ThiscangreatlydecreasetheamountoftimerequiredtoobtaindosedataforbeamletsintheFMOproblemofIMRTtreatmentplanningwithoutsacricingaccuracy. 138

PAGE 139

Figure5-16.Variationsofthetsusedtoapproximatethedepth-doseandlateralpenumbradistributionsasfunctionofthenumberofhistories. Forfuturetesting,moretestsonthenumberofMonteCarlohistoriesneededwillberunaswell,particularlywithhistoriesintherangeof10-100million.Moretestsofvaryingtissues,bothhomogeneousandheterogeneous,willberuntodetermineasmallerrangeofdegreestobeevaluatedforthepolynomialttothedepth-dosecurve.Anautomatedmethodofdeterminingaqualitysetofinitialparameterstomodelthelateralpenumbrawillalsobedeveloped.Lastly,thescenariowhereabeamlethitsmultipletissuessimultaneouslywillbetestedusingourmodelforapproximatingthelateralpenumbra.JelenandAlber[ 89 ]andJelenetal.[ 90 ]havedemonstratedthatabeamletcanbemodeledveryeectivelyusinganapproachbasedontheonedescribedhere.ThisapproachwasimproveduponbyscalingthemodelingparametersaccordingtotissuedensityinJelenandAlber[ 89 ].Despitethesophisticationofthedensityscalingmethodemployed,themodellosesaccuracyinthepenumbraregionsandattheedgeoftissueheterogeneities.ThisstudyalsousedaLevenberg-Marquardtalgorithmtodeterminethe 139

PAGE 140

modelingparameters,andalthoughthedetailsoftheimplementationarenotprovided,itispossiblethatwithanimprovedinitialguessordampingparameter,thealgorithmcouldconvergetobettermodelingparameters,thusprovidingimprovedpredictionofbeamletbehavioratthepenumbra.Tofurtherimproveupontheirwork,theeectsoftheMLCmustbeconsidered.Onemethodofaccountingfortheseeectscouldbetomodelthedosedepositionofanentireapertureratherthanjustthedosedepositionofasinglebeamlet.AsthenumberandshapeofaperturesrequiredtodeliveranFMO-basedIMRToptimizationareunknown,thismethodwouldbemostpracticalifanaperturemodulationapproach|whereapertureuencesfromapre-denedsetofaperturesarechosen,insteadofuencesfromindividualbeamlets|isemployedinsteadofanFMOapproach,asthenumberandshapeoftheaperturesinconsiderationarepredetermined. 140

PAGE 141

REFERENCES [1] AmericanCancerSociety.CancerFactsandFiguresReport.2006. [2] MurphyGP,LawrenceWL,LenlardRE,eds.AmericanCancerSocietyTextbookonClinicalOncology.TheAmericanCancerSociety,1995. [3] PerezCA,BradyLW.PrinciplesandPracticeofRadiotherapy.Lippincott-Raven,3edn.,1998. [4] SteelGG.BasicClinicalRadiobiologyforRadiationOncologists.EdwardArnoldPublishers,1994. [5] MaCM,MokE,KapurA,FindleyD,BrainS,BoyerAL.Clinicalimplementationofamontecarlotreatmentplanningsystem.MedicalPhysics1999;26:2133{43. [6] BortfeldT.Optimizedplanningusingphysicalobjectivesandconstraints.SeminRadiatOncol1999;9:20{34. [7] AlberM,NusslinF.Anobjectivefunctionforradiationtreatmentoptimizationbasedonlocalbiologicalmeasures.PhysMedBiol1999;44:479{493. [8] JonesLC,HobanPW.Treatmentplancomparisonusingequivalentuniformbiologicallyeectivedoseeubed.PhysMedBiol2000;45:159{170. [9] KallmanP,LindBK,BrahmeA.Analgorithmformaximizingtheprobabilityofcomplication{freetumor{controlinradiation-therapy.PhysMedBiol1992;37:871{890. [10] MavroidisP,LindBK,BrahmeA.Biologicallyeectiveuniformdoseforspecication,reportandcomparisonofdoseresponserelationsandtreatmentplans.PhysMedBiol2001;46:2607{2630. [11] NiemierkoA.Reportingandanalyzingdosedistributions:aconceptofequivalentuniformdose.MedicalPhysics1997;24:103{110. [12] NiemierkoA,UrieM,GoiteinM.Optimizationof3dradiation-therapywithbothphysicalandbiologicalend-pointsandconstraints.IntJRadiatOncolBiolPhys1992;23:99{108. [13] WuQW,DjajaputraD,WuY,ZhouJN,LiuHH,MohanR.Intensity-modulatedradiotherapyoptimizationwithgeud-guideddose-volumeobjectives.PhysMedBiol2003;48:279{291. [14] WuQW,MohanR,NiemierkoA,Schmidt-UllrichR.Optimizationofintensity-modulatedradiotherapyplansbasedontheequivalentuniformdose.IntJRadiatOncolBiolPhys2002;52:224{235. 141

PAGE 142

[15] HamacherHW,KuferKH.Inverseradiationtherapyplanningamultipleobjectiveoptimizationapproach.DiscreteAppliedMathematics2002;118:145{161. [16] BednarzG,MichalskiD,HouserC,HuqMS,XiaoY,AnnePR,GalvinJM.Theuseofmixed-integerprogrammingforinversetreatmentplanningwithpre-denedeldsegments.PhysMedBiol2002;47:2235{2245. [17] FerrisMC,MeyerRR,D'SouzaW.Radiationtreatmentplanning:Mixedintegerprogrammingformulationsandapproaches.InGAppa,LPitsoulis,HPWilliams,eds.,HandbookonModellingforDiscreteOptimization.Springer-Verlag,NewYork,NY,2006;317{340. [18] LangerM,BrownR,UrieM,LeongJ,StracherM,ShapiroJ.Large-scaleoptimizationofbeamweightsunderdose-volumerestrictions.IntJRadiatOncolBiolPhys1990;18:887{893. [19] LangerM,MorrillS,BrownR,,LeeO,LaneR.Acomparisonofmixedintegerprogrammingandfastsimulatedannealingforoptimizingbeamweightsinradiationtherapy.MedicalPhysics1996;23:957{964. [20] LeeEK,FoxT,CrockerI.Simultaneousbeamgeometryandintensitymapoptimizationinintensity-modulatedradiationtherapytreatmentplanning.An-nalsofOperationsResearch2003;119:165{181. [21] LeeEK,FoxT,CrockerI.Integerprogrammingappliedtointensity-modulatedradiationtherapytreatmentplanning.IntJRadiatOncolBiolPhys2006;64:301{320. [22] ShepardDM,FerrisMC,OliveraGH,MackieTR.Optimizingthedeliveryofradiationtherapytocancerpatients.SIAMReview1999;41:721{744. [23] RomeijnHE,AhujaRK,DempseyJF,KumarA,LiJG.Anovellinearprogrammingapproachtouencemapoptimizationforintensitymodulatedradiationtherapytreatmentplanning.PhysMedBiol2003;38:3521{3542. [24] RomeijnHE,AhujaRK,DempseyJF,KumarA,LiJG.Acolumngenerationapproachtoradiationtherapytreatmentplanningusingaperaturemodulation.SIAMJournalofOptimization2005;15:838{862. [25] RomeijnHE,DempseyJF,LiJG.Aunifyingframeworkformulti-criteriauencemapoptimizationmodels.PhysMedBiol2004;49:1991{2013. [26] RomeijnHE,AhujaRK,DempseyJF,KumarA.Anewlinearprogrammingapproachtoradiationtherapytreatmentplanningproblems.OperationsResearch2006;54:201{216. [27] DasSK,MarksLB.Selectionofcoplanarornoncoplanarbeamsusingthree-dimensionaloptimizationbasedonmaximumbeamseparationandminimized 142

PAGE 143

nontargetirradiation.IntJRadiatOncolBiolPhys1997;38:643{655. [28] HaasOC,BurnhamKJ,MillsJ.Optimizationofbeamorientationinradiotherapyusingplanargeometry.PhysMedBiol1998;43:2179{2193. [29] SchreibmannE,LahanasM,XingL,BaltasD.Multiobjectiveevolutionaryoptimizationofthenumberofbeams,theirorientationsandweightsforintensity-modulatedradiationtherapy.PhysMedBiol2004;49:747{770. [30] ChaoKSC,BlancoAI,DempseyJF.AconceptualmodelintegratingspatialinformationtoassesstargetvolumecoverageforIMRTtreatmentplanning.IntJRadiatOncolBiolPhys2003;56:1438{1449. [31] NocedalJ,WrightSJ.NumericalOptimization.Springer-Verlag,1999. [32] EzzellGA.Geneticandgeometricoptimizationofthree-dimensionalradiationtherapytreatmentplanning.MedicalPhysics1996;23:293{305. [33] LiY,YaoJ,YaoD.AutomaticbeamangleselectioninIMRTplanningusinggeneticalgorithm.PhysMedBiol2004;49:1915{1932. [34] LiY,YaoJ,YaoD,ChenW.Aparticleswarmoptimizationalgorithmforbeamangleselectioninintensity-modulatedradiotherapyplanning.PhysMedBiol2005;50:3491{3514. [35] BortfeldT,SchlegelW.Optimizationofbeamorientationsinradiationtherapy:sometheoreticalconsiderations.PhysMedBiol1993;38:291{304. [36] DjajaputraD,WuQ,WuY,MohanR.AlgorithmandperformanceofaclinicalIMRTbeam-angleoptimizationsystem.PhysMedBiol2003;48:3191{3212. [37] LuHM,KooyHM,LeberZH,LedouxRJ.Optimizedbeamplanningforlinearaccelerator-basedstereotacticradiosurgery.IntJRadiatOncolBiolPhys1997;39:1183{1189. [38] PugachevA,XingL.IncorporatingpriorknowledgeintobeamorientationoptimizationinIMRT.IntJRadiatOncolBiolPhys2002;54:1565{1574. [39] RowbottomCG,OldhamM,WebbS.Constrainedcustomizationofnon-coplanarbeamorientationsinradiotherapyofbraintumours.PhysMedBiol1999a;44:383{399. [40] SteinJ,MohanR,WangXH,BortfeldT,WuQ,PreiserK,LingCC,SchlegelW.Numberandorientationsofbeamsinintensity-modulatedradiationtreatments.MedicalPhysics1997;24:149{160. [41] SoderstromS,BrahmeA.Selectionofsuitablebeamorientationsinradiationtherapyusingentropyandfouriertransformmeasures.PhysMedBiol1992;37:911{924. 143

PAGE 144

[42] SoderstromS,BrahmeA.Whichisthemostsuitablenumberofphotonbeamportalsincoplanarradiationtherapy?IntJRadiatOncolBiolPhys1995;33:151{59. [43] RowbottomCG,WebbS,OldhamM.Beam-orientationcustomizationusinganarticialneuralnetwork.PhysMedBiol1999b;44:2251{2262. [44] GokhaleP,HusseinEM,KulkarniN.Theuseofbeamseyeviewvolumetricsintheselectionofnon-coplanarradiationportals.MedicalPhysics1994;23:153{163. [45] MeedtG,AlberM,NusslinF.Non-coplanarbeamdirectionoptimizationforintensity-modulatedradiotherapy.PhysMedBiol2003;48:2999{3019. [46] ChenGT,SpelbringDR,PelizzariCA,BalterJM,MyrianthopoulosLC,VijayakumarS,HalpernH.Theuseofbeamseyeviewvolumetricsintheselectionofnon-coplanarradiationportals.IntJRadiatOncolBiolPhys1992;23:153{163. [47] ChoBCJ,RoaHW,RobinsonD,MurrayB.Thedevelopmentoftarget-eye-viewmapsforselectionofcoplanarornoncoplanarbeamsinconformalradiotherapytreatmentplanning.MedicalPhysics1999;26:2367{2372. [48] GoiteinM,AbramsM,RowellD,PollariH,WilesJ.Multi-dimensionaltreatmentplanning:Ii.beamseye-view,backprojection,andprojectionthroughCTsections.IntJRadiatOncolBiolPhys1983;9:789{97. [49] PugachevA,XingL.Computer-assistedselectionofcoplanarbeamorientationsinintensity-modulatedradiationtherapy.PhysMedBiol2001;46:2467{2476. [50] PugachevA,XingL.Pseudobeam's-eye-viewasappliedtobeamorientationselectioninintensity-modulatedradiationtherapy.IntJRadiatOncolBiolPhys2001;51:1361{1370. [51] HolderA,SalterB.Atutorialonradiationoncologyandoptimization.InHGreenberg,ed.,TutorialsonEmergingMethodologiesandApplicationsinOp-erationsResearch.KluwerAcademicPress,Boston,MA,2004. [52] MorrillSM,LaneRG,JacobsonG,RosenII.Treatmentplanningoptimizationusingconstrainedsimulatedannealing.PhysMedBiol1991;36:1341{61. [53] OldhamM,KhooV,RowbottomCG,BedfordJ,WebbS.Acasestudycomparingtherelativebenetofoptimisingbeam-weights,wedge-angles,beamorientationsandtomotherapyinstereotacticradiotherapyofthebrain.PhysMedBiol1998;43:2123{46. [54] RowbottomCG,WebbS,OldhamM.Improvementsinprostateradiotherapyfromthecustomizationofbeamdirections.MedicalPhysics1998;25:1171{1179. 144

PAGE 145

[55] WangX,ZhangX,DongL,LieH,WuQ,MohanR.DevelopmentofmethodsforbeamangleoptimizationforIMRTusinganacceleratedexhaustivesearchstrategy.IntJRadiatOncolBiolPhys2004;60:1325{37. [56] WangX,ZhangX,DongL,LiuH,GillinM,AhamadA,AngK,MohanR.EectivenessofnoncoplanarIMRTplanningusingaparallelizedmultiresolutionbeamangleoptimizationmethodforparanasalsinuscarcinoma.IntJRadiatOncolBiolPhys2005;63:594{601. [57] WoudstraE,HeijmanBJM.Automatedbeamangleandweightselectioninradiotherapytreatmentplanningappliedtopancreastumors.IntJRadiatOncolBiolPhys2004;56:878{88. [58] D'SouzaWD,MeyerRR,ShiL.Selectionofbeamorientationsinintensity-modulatedradiationtherapyusingsingle-beamindicesandintegerprogramming.PhysMedBiol2004;49:3465{3481. [59] EhrgottM,JohnstonR.Optimisationofbeamdirectionsinintensitymodulatedradiationtherapyplanning.ORSpectrum2003;25:251{264. [60] LimJ,FerrisM,ShepardD,WrightS,EarlM.Anoptimizationframeworkforconformalradiationtreatmentplanning.INFORMSJournalOnComputing2006. [61] WangC,DaiJ,HuY.Optimizationofbeamorientationsandbeamweightsforconformalradiotherapyusingmixedintegerprogramming.PhysMedBiol2003;48:4065{4076. [62] FoxC,RomeijnHE,DempseyJF.Fastvoxelandpolygonray-tracingalgorithmsforIMRTtreatmentplanning,2005.SubmittedtoMedicalPhysics. [63] SiddonRL.Prismrepresentation:a3dray-tracingalgorithmforradiotherapyapplications.PhysMedBiol1985;8:817{824. [64] SiddonRL.Fastcalculationoftheexactradiologicalpathforathree-dimensionalCTarray.MedicalPhysics1985;12:252{255. [65] JacobsF,SundermannE,SutterBD,ChristiaensM,LemahieuI.Afastalgorithmtocalculatetheexactradiologicalpaththroughapixelorvoxelspace.JournalofComputingandInformationTechnologyCIT1998;6:89{94. [66] AlemanDM,RomeijnHE,DempseyJF.Beamorientationoptimizationmethodsinintensitymodulatedradiationtherapytreatmentplanning.IIEConferenceProceedings2006. [67] AlemanDM,RomeijnHE,DempseyJF.Aresponsesurfaceapproachtobeamorientationoptimizationinintensitymodulatedradiationtherapytreatmentplanning.Inreview2006. 145

PAGE 146

[68] JonesDR.Ataxonomyofglobaloptimizationmethodsbasedonresponsesurfaces.JournalofGlobalOptimization2001;21:345{383. [69] JonesDR,SchonlauM,WelchWJ.Ecientglobaloptimizationofexpensiveblack-boxfunctions.JournalofGlobalOptimization1998;13:455{492. [70] CsallnerAE,CsendesT,MarkotMC.Multisectioninintervalbranch-and-boundmethodsforglobaloptimizationi.theoreticalresults.JournalofGlobalOptimization2000;16:371{392. [71] LagouanelleJ,SoubryG.Optimalmultisectionsinintervalbranch-and-boundmethodsofglobaloptimization.JournalofGlobalOptimization2004;30:23{38. [72] EpperlyTGW,PistikopoulosEN.Areducedspacebranchandboundalgorithmforglobaloptimization.JournalofGlobalOptimization1997;11:287{311. [73] BarrientosO,CorreaR.Analgorithmforglobalminimizationoflinearlyconstrainedquadraticfunctions.JournalofGlobalOptimization2000;16:77{93. [74] ThoaiNV.Convergenceofdualityboundmethodinpartlyconvexprogramming.JournalofGlobalOptimization2002;22:263{270. [75] TuyH.Onsolvingnonconvexoptimizationproblemsbyreducingthedualitygap.JournalofGlobalOptimization2005;32:349{365. [76] PhongTQ,AnLTH,TaoPD.Decompositionbranchandboundmethodforgloballysolvinglinearlyconstrainedindenitequadraticminimizationproblems.OperationsResearchLetters1995;17:215{220. [77] BomzeI.Branch-and-boundapproachestostandardquadraticoptimizationproblems.JournalofGlobalOptimization2002;2:17{37. [78] CambiniR,SodiniC.Decompositionmethodsforsolvingnonconvexquadraticprogramsviabranchandbound.JournalofGlobalOptimization2005;33:313{336. [79] AlemanDM,KumarA,AhujaRK,RomeijnHE,DempseyJF.Neighborhoodsearchapproachestobeamorientationoptimizationinintensitymodulatedradiationtherapytreatmentplanning.inreview2007. [80] KumarA.Novelmethodsforintensity-modulatedradiationtherapytreatmentplanning.Ph.D.thesis,UniversityofFlorida,2005. [81] GemanS,GemanD.Stochasticrelaxation,gibbsdistributions,andthebayesianrestorationofimages.IEEETransactionsonPatternAnalysisandMachineIntelli-gence1984;6:721{741. 146

PAGE 147

[82] GelfandAE,SmithAFM.Samplingbasedapproachestocalculatingmarginaldensities.JournaloftheAmericanStatisticalAssociation1990;85:398{409. [83] SmithRL.Amontecarloprocedurefortherandomgenerationoffeasiblesolutionstomathematicalprogrammingproblems.BulletinoftheTIMS/ORSAJointNationalMeeting1980;:101. [84] BelisleCJP,RomeijnHE,SmithRL.Hit-and-runalgorithmsforgeneratingmultivariatedistributions.MathematicsofOperationsResearch1993;18:255{266. [85] KirkpatrickS,GelattCD.Optimizationbysimulatedannealing.Science1983;220:671{680. [86] BomzeI.Fastsimulatedannealing.PhysicsLetters1987;122A:157{162. [87] BelisleCJP.ConvergencetheoremsforaclassofsimulatedannealingalgorithmsonRd.JournalofAppliedProbability1992;29:885{895. [88] AlemanDM,GlaserD,RomeijnHE,DempseyJF.Aprimal-dualinteriorpointalgorithmforuencemapoptimizationinintensitymodulatedradiationtherapytreatmentplanning.workinprogress2007. [89] JelenU,AlberM.AnitesizepencilbeamalgorithmforIMRTdoseoptimization:densitycorrections.PhysicsinMedicineandBiology2007;52:617{633. [90] JelenU,SohnM,AlberM.AnitesizepencilbeamforIMRTdoseoptimization.PhysicsinMedicineandBiology2005;50:1747{1766. [91] SempauJ,WildermanSJ,BielajewAF.Dpm,afast,accuratemontecarlocodeoptimizedforphotonandelectronradiotherapytreatmentplanningdosecalculations.PhysMedBiol2000;45:2263{91. [92] BaroJ,SempauJ,Fernandez-VareaJM,SalvatF.Penelope:Analgorithmformontecarlosimulationofthepenetrationandenergylossofelectronsandpositronsinmatter.NuclearInstrumentsandMethods1995;B100:31{46. [93] SempauJ,BaroJ,Fernandez-VareaJM,SalvatF.Analgorithmformontecarlosimulationofcoupledelectron-photonshowers.NuclearInstrumentsandMethods1997;B132:377{90. [94] KhanFM.ThePhysicsofRadiationTherapy.LippincottWilliamandWilkins,1994. 147

PAGE 148

BIOGRAPHICALSKETCHDionneM.Alemancompletedherbachelor'sdegreeinindustrialandsystemsengineeringattheUniversityofFlorida.ShewentontostudyintensitymodulatedradiationtherapyIMRTtreatmentplanningoptimizationinthegraduateprogramoftheDepartmentofIndustrialandSystemsEngineeringattheUniversityofFlorida.ShewillreceiveherDoctorofPhilosophyinIndustrialandSystemsEngineeringinDecemberof2007,afterwhichshewillpursueacareerintheDepartmentofMechanicalandIndustrialEngineeringattheUniversityofToronto.Dionneplanstocontinueherresearchincancertreatments,aswellasotherapplicationsofoperationsresearchtechniquestothemedicalandhealthcareindustries. 148


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101211_AAAABM INGEST_TIME 2010-12-11T21:37:32Z PACKAGE UFE0021522_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 16409 DFID F20101211_AABLSH ORIGIN DEPOSITOR PATH aleman_d_Page_112.QC.jpg GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
5861173c1cfc3f2d4dfccdc8408e89e1
SHA-1
1cefb607b8fafefff6acda4fc4db168e49d0b4ea
22552 F20101211_AABLRS aleman_d_Page_090.QC.jpg
e74b0c775709c901a1a62c6367805c2c
5bdb9a1a3103624e8dcce8342977a45e6d6ca762
37978 F20101211_AABKPF aleman_d_Page_045.pro
0c6f4ef6e70dff18b002577ea51944da
997c40d76e916532c3273bdb4edc6c34fb24837d
27020 F20101211_AABKOQ aleman_d_Page_055.QC.jpg
d4da5c0ccf717a5e7cfe436f0800758f
bb0cfe272704e43bb0c3d558198fde24489bde68
5397 F20101211_AABLSI aleman_d_Page_112thm.jpg
ca53c7df56075319799dfa5a261fa395
19cfe1fd39fdeed77ad1277925ce8adbaa1646c5
27002 F20101211_AABLRT aleman_d_Page_092.QC.jpg
0a5b05141fb59c67f58321fec89ab685
305f712c4a5b595a5c3e80c2bcea3417b53a588b
56905 F20101211_AABKPG aleman_d_Page_108.jpg
f71d3aae039f86c21ca14b86d6ccaab0
67b179412db62493925212a592f560c929fdd56d
25271604 F20101211_AABKOR aleman_d_Page_118.tif
8cfa68a3399a76aa673929bd2b266db3
b8b4b0195a2e3b65295782cbf54b548148d99215
17299 F20101211_AABLSJ aleman_d_Page_113.QC.jpg
2c2381be28248b8489c6e2ffae1d4805
4a3560edfa09147983e4503fc9d3617bf50add7e
24371 F20101211_AABLRU aleman_d_Page_094.QC.jpg
6949fa55d38a28cdf459842ba0276a14
47dd5afd64655ec922ba6ec8ed8b2ac53535eb26
39914 F20101211_AABKPH aleman_d_Page_099.pro
e05353a7887145865e53fa9f2e7fd072
161b81d890cfaa441bdac8d7ba57a6f7494e3fa9
17425 F20101211_AABKOS aleman_d_Page_051.QC.jpg
3e9232904f83a11867976bdfc15c01b6
79033ef8fda920b5cf31d167914473fc4be89042
5413 F20101211_AABLSK aleman_d_Page_116thm.jpg
0599a7cafa859977493318bdaa227a38
c6621895b1b2968f6cab6eae0c5f4cf9671992df
6035 F20101211_AABLRV aleman_d_Page_094thm.jpg
f75a45b7c7d878f05b11cea3335e89d0
30619ebbabe316660ec482e31a07ff5d0520c54f
3026 F20101211_AABKPI aleman_d_Page_148thm.jpg
3209b2db5e83fd4fae2f7e43084aff91
fb402e2679625bd659f39c98d555fcc2b9d38c9c
6802 F20101211_AABKOT aleman_d_Page_065thm.jpg
d6d52736a9ff2b2aaf48cbc669283a6d
6a738375cec73cccf9e0d0242c8920a8d14f8fc0
5331 F20101211_AABLSL aleman_d_Page_117thm.jpg
eea3ca75bbfb6af14407e07fd0cb677b
0d5b855ddd16c2c8e27a727dea35e97f41907f13
3208 F20101211_AABLRW aleman_d_Page_095thm.jpg
988533b1e02d986cf53238604c0a21fe
a6e9d043fc773033370083cf015f7b7f897633ad
F20101211_AABKPJ aleman_d_Page_110.tif
e1364174af1cd4ae056bca2f066bd03b
cc41afb4a0913af02520a0fa2157986f174ace0b
19634 F20101211_AABKOU aleman_d_Page_148.pro
286d1d79e0ed5653c2f0e5151d1cd23c
84577fbe8e60f677262e743eee22a13b8c56200c
5734 F20101211_AABLTA aleman_d_Page_147thm.jpg
820f3e083389b2c205c12c06d1da1717
6ac51d291c6997fc74b91ff5334eeb25ed327fec
5623 F20101211_AABLSM aleman_d_Page_122thm.jpg
ab411ace71aa0cf3447d0ec86746c619
d206c946dc42e322fe513a7f5040fed5197a58f1
23672 F20101211_AABLRX aleman_d_Page_097.QC.jpg
979f0ca2d3fb6fb7788abd7de480fdee
3271e5693131504fcea047fe3f4f0106aaf55d0f
6071 F20101211_AABKPK aleman_d_Page_132thm.jpg
1821ceb9e0d7be885f5440961d45cb65
9a11dfdf7b6e51d6a06d92c2f3f35096af7e280b
2162 F20101211_AABKOV aleman_d_Page_066.txt
4c166778bd83923f3178ea9710d389f5
eb43d8e148b8acd00a50af68e0a81e34c12ecbe7
171008 F20101211_AABLTB UFE0021522_00001.mets FULL
6536e4bed89fe3487388d9fba57c3fbc
4119d2652ad2ec17a3ab25559af6155d7eb03552
6153 F20101211_AABLRY aleman_d_Page_097thm.jpg
37fdf82b44bd30d22ea24a3f44c9d086
baead1a81a054eb31b703d02e37ec3e844b7d21d
54863 F20101211_AABKOW aleman_d_Page_118.jpg
32e600ecac84359852604d98e70014ee
83a2a7956b6509949e299e892a9affc829724e58
5128 F20101211_AABLSN aleman_d_Page_126thm.jpg
84d1383df25a00a3ca5dbe2764aabaec
3a3477ef3b2114cf07edec7f21e56b5c7bbb88a2
17009 F20101211_AABLRZ aleman_d_Page_099.QC.jpg
39872e144c74efc9a60061a69732991d
4e711fc5d124b5975ddabe48d409474acbc019ed
2187 F20101211_AABKPL aleman_d_Page_102.txt
97532fd349ed2013985c43f764586482
9880d3d7b3c2d6afd721d2ece556a2fb27624a53
2145 F20101211_AABKOX aleman_d_Page_139.txt
5d4d2d7e83ae362ed22d25b3e6566929
49f5dfd48298981821e5d9b8c97bc2ae87a66139
842813 F20101211_AABKQA aleman_d_Page_124.jp2
ee50f079dfa2ad0afc05127c19ec8cee
5bdef93b7a6e3dad24a52180a6e8a7fb0052cd4a
16686 F20101211_AABLSO aleman_d_Page_128.QC.jpg
fac85136f88e6661f6b784d5b0eb9e9a
f731adae0ed70eaa5c50e243f868e2777a5fe9f7
54703 F20101211_AABKPM aleman_d_Page_083.pro
d67637d3a058113f157a88246b9a94ab
44254634d0c4accd9c749b76815179ae3f920953
27727 F20101211_AABKOY aleman_d_Page_004.jp2
51ea80b5eba28ca6766340d5df46e0b0
87c2f0b5e23da4a042a571c54121866a9ba3fdcd
84477 F20101211_AABKQB aleman_d_Page_083.jpg
b961d1eae5c9f5d80c5e798a4de8ca87
48c67bc2f232711f87bcfc98d266d2e9e002c462
5223 F20101211_AABLSP aleman_d_Page_129thm.jpg
9fbe3fcb90f46e757a5544854b450af7
6f6d57262169cf8b6b148e60606d81a1566238f0
1051 F20101211_AABKPN aleman_d_Page_106.txt
4c6307e5fa59782e159147ea317ceddd
651c818d3178dcbba12ad4944c249d8cdaf613e8
117805 F20101211_AABKOZ aleman_d_Page_147.jp2
f8696aae92e7b9080a12177756f79a85
908c47c55471e8efe8da6790f0a1de2fbcabaa36
54684 F20101211_AABKQC aleman_d_Page_053.pro
85a549da8aabe4a873cebf5c4648e33b
b2deb63f61ec81c56a9f562c2f77ea939c2334f9
4964 F20101211_AABLSQ aleman_d_Page_130thm.jpg
3b89727cc5e37d27bfa2b2ad2b64bc37
8bafc2a2a2d912db820b4c0795da083200fedc69
1360 F20101211_AABKPO aleman_d_Page_002thm.jpg
d304bb4043bf02ade9f136d5898b0348
8aa579f198f1dbfc6f00462b21b1de5cb78629af
1101 F20101211_AABKQD aleman_d_Page_078.txt
564fec906db7ea0537ff1823bc9c72fc
34062cdf650067ea92b76c91c73b6fa36db026b6
18548 F20101211_AABLSR aleman_d_Page_131.QC.jpg
7bbf277c86825b1b972b1c53fc052348
37e2240f9dc5162dcf693f9d539ce8307115b69a
F20101211_AABKPP aleman_d_Page_087.tif
2931dedd2953a627582e79779abba382
7873ec39bf9d2de37fffaf93b768130e796bc719
55839 F20101211_AABKQE aleman_d_Page_022.pro
49afa34951eae1845cdab12b4cea1a11
29ac938aaebf45af3bcbdb2edb1fcf696a3d8ec0
5256 F20101211_AABLSS aleman_d_Page_131thm.jpg
e468f034baf1e7239e1019ee71130647
832d19689c56b87155193c1ec0889e4f05897371
F20101211_AABKPQ aleman_d_Page_009.tif
49c27e4322d9fe718206749ca12befbd
5eabe03168a587ae0fb810ed1c8a6c1a94776af8
6809 F20101211_AABKQF aleman_d_Page_111thm.jpg
a3be34caa9e17a330ab8801aa119db93
08523e3660e3ba95f4a60fcb16312534c5c1a855
25593 F20101211_AABLST aleman_d_Page_133.QC.jpg
11502a455c7b8333072d8324db54c333
c3259eb45ed065dd591992b49f6e257800162d88
5076 F20101211_AABKPR aleman_d_Page_061thm.jpg
f333dfe8cb20eb7dc3e3f1cd985d4a01
b5f806e417b76b6c70af2b5098665250d8664235
6129 F20101211_AABKQG aleman_d_Page_144thm.jpg
29fcbdf0a34ba4fe62f082e8a0a0b980
1c31ea75aa9210fc59febe26eb05fbb89c042b2a
6525 F20101211_AABLSU aleman_d_Page_134thm.jpg
e8acdbd18b7d70a00edff075e2bdc860
afea72e78a56c30d327a8e3948a8211071ab099d
760642 F20101211_AABKPS aleman_d_Page_136.jp2
010cc535228e7c605a56262d0c398a25
04acecba888675d7efbbfe324528c19f49f3676b
1633 F20101211_AABKQH aleman_d_Page_126.txt
0ce47ac826727ebdbc40f8a462387656
c9322be534b0db925cb741f35f9a1605a95ed14a
3622 F20101211_AABLSV aleman_d_Page_135thm.jpg
258411a4b63d740278e52863c4c779f2
1fb3f8b9a4f5b8517b3d1f08a2e8a2a7f1d5e916
26839 F20101211_AABKPT aleman_d_Page_054.QC.jpg
7216d59526bcac20b4c09cfe6ac684f5
8a4104261b41153036788ae5f97cb540f2a17c6a
6910 F20101211_AABKQI aleman_d_Page_047thm.jpg
f612fd85932dacab3e141218be71a149
2ec8af512938d1e1465814c981037a22f8dc24a4
4241 F20101211_AABLSW aleman_d_Page_136thm.jpg
2ef3fb50851aa2f741d6121a1f7cfe30
fb2ab937dd0e2c07251600cdccdaa87baeb047e9
F20101211_AABKPU aleman_d_Page_056.tif
328c45030ff1a8121c1a5cc8c1d3d748
b3356711bbd34d9ac812d03b2246791d784b4a84
F20101211_AABKQJ aleman_d_Page_058.tif
f5a86c08d456961915751e926358776c
4bbe4b977412d7c6188ccd7a72331217b3e74f0b
21625 F20101211_AABLSX aleman_d_Page_138.QC.jpg
ce4493828bda5341b0ba1bc873e9a61a
c7289a405f572b5d3e6637a06615b155f6b64b5b
4722 F20101211_AABKQK aleman_d_Page_078thm.jpg
551fa39ddd96e2418abee8d8dc43986e
3ee74b40d7eee18b5e0c0bdb3d360e323c05c904
912295 F20101211_AABKPV aleman_d_Page_060.jp2
22aaa2367432806f311d2980f96c0ce3
d56d1423051f44ce0c537db2006cd13248f64689
6619 F20101211_AABLSY aleman_d_Page_142thm.jpg
bdcb7dd63172d2fa58f0019eb8be5695
45b2b07e2b79b13ec02808431ed9270da36f5112
F20101211_AABKQL aleman_d_Page_068.tif
36e618acd2dd49103bb10200de8a509a
c843c6ed9635a3744fe5a4c07a017022fb9cc7e8
6398 F20101211_AABKPW aleman_d_Page_069thm.jpg
e7957062fedadd9519357c91716349c8
bd85b591b71517b7ae075fc2c49f12b3846d13e2
6522 F20101211_AABLSZ aleman_d_Page_145thm.jpg
8d1a2240b56f3d083b1a5a721ad8c8fb
75a099956c51c2602b6b8edeb9285df667caa4fc
57207 F20101211_AABKRA aleman_d_Page_023.pro
55224671d73015edc3a35749effcb9e2
1d4f9bd0333e5080ed188096f947c7589741399c
1051971 F20101211_AABKPX aleman_d_Page_005.jp2
721496e58b2447cb4c48c5a913ae3700
71d9ccbd03e35073f6225f00c233e6606301a432
18929 F20101211_AABKRB aleman_d_Page_124.QC.jpg
77991c220285f6fcbbbd98ffc7eec39c
6f41451a9c890407270cd7a813b72fe810789356
1051943 F20101211_AABKQM aleman_d_Page_022.jp2
993f2efc13759f07077a58ce2612d15a
19b302fc68fc03c132404d2fcd251818796ac931
F20101211_AABKPY aleman_d_Page_074.tif
23cb31641e26fce01b5f1b3b5e2d659d
40b1bdc838863f68a44aa64562ebd910928061c0
2275 F20101211_AABKRC aleman_d_Page_123.txt
c156c76759820a749a41a6b2e4a52608
1c59dbe06bce3fc180bb862040c4685556b3ec4e
F20101211_AABKQN aleman_d_Page_106.tif
fbe7b90053e648b4fee9d4c7ac57d1fa
7206ff8addc6718763d96e6d313d5d9f607c1e60
15600 F20101211_AABKPZ aleman_d_Page_116.pro
f79bcd53c6a927687552854f73b3a7e0
562c02e6d22647ce2dfef10be78da9c833e41748
65477 F20101211_AABKRD aleman_d_Page_127.jpg
15a1e0d1e01415e2761acbc5a94e87fb
a6b7bc765c9598f930818d4da9f12bf2b9e44040
10964 F20101211_AABKQO aleman_d_Page_004.pro
ec3dea6a514e39c96234278753de7cc9
527c12ae3aeee44fa5ead3ecce1347454852dfca
1588 F20101211_AABKRE aleman_d_Page_100.txt
22e6f60b6c2f85992fe311f22a7e3d95
93bd650868c8caa2e85aa35b0a4b6b22cfd5ce85
54876 F20101211_AABKQP aleman_d_Page_011.jp2
4d405c120008eed4e8eee0cfe98748f1
1209440ecc7501e3523bf43d4ac7476491962756
941146 F20101211_AABKRF aleman_d_Page_090.jp2
54d0087fa599cf71f84468bcc9d159b5
bb6a65067b4f308d57e2390942a651f1d28a4ec5
25887 F20101211_AABKQQ aleman_d_Page_049.QC.jpg
06b9fc457992765b0ed269dec7b13f0b
aef0d6063cd6631ffb9ca20c2fbbe81a46d1cc20
F20101211_AABKRG aleman_d_Page_082.tif
10d2610c3dc6183c3155d1fb8617a1b0
036c149635d51c27414636e00b244ee91932ee92
F20101211_AABKQR aleman_d_Page_125.tif
1077cd5eed975bdf1ee72384064d9c5d
01b779a8a0b4bb41294fba588c45a99528102702
477 F20101211_AABKRH aleman_d_Page_004.txt
134cefae901bfb2ebdeccfc210fe5da6
b5b11f66eaf2fdd1525cc56c78466d358c329b99
F20101211_AABKQS aleman_d_Page_026.tif
6423086feba2f8ad55f86a5d1b0f4e72
c56901bd8f8beda6e4ace1de3c1abe5f380b6add
F20101211_AABKRI aleman_d_Page_102.tif
c5aa16a95281f5ab1eb16d34c1673f8d
a1f4e4577c1b89a48c31bdebb0bd25fb4a1084e2
1198 F20101211_AABKQT aleman_d_Page_090.txt
ec6f3f69ec0dbe2a544ea0a48dee9ffa
29ef03d41442a644fe3b8b5863c06ee0b105e001
110861 F20101211_AABKRJ aleman_d_Page_079.jp2
d3882a35d3b32dc2e5f4706f97c644af
568ab8885ad39dfb13a86138f4324daf16ba7195
56053 F20101211_AABKQU aleman_d_Page_113.jpg
dbef14ab6b6c8af9f35395a01dfcfdcc
e848484632e59eb4742d7a3f7aa506056e73edfb
1634 F20101211_AABKRK aleman_d_Page_129.txt
55f0d8ee44010b2c7d9517382019e6b6
04a0a82cc8bcb864221f40aef72face2171515b8
920351 F20101211_AABKQV aleman_d_Page_077.jp2
b804bc2d200684581e99feef735e8829
e906410d625ea996ce4949b548504cb4b6847146
57590 F20101211_AABKRL aleman_d_Page_028.jpg
2861138268ae34a5b1aedc829b4fa1b4
16ded6ff0c0964455172e9ee39b396469cfee7b6
F20101211_AABKQW aleman_d_Page_067.tif
080ee04ea5c8d625695730babc3e0b44
fba39c32d8d29154c825b6a0c0ef608d7607bf80
2469 F20101211_AABKRM aleman_d_Page_143.txt
271096d10bc4847306ffdd4f8cb02c53
e958bb845f9da6f5dd860e6a80ec5b0cdd39db01
F20101211_AABKQX aleman_d_Page_062.tif
1bbcad3c5809a750f583b43df8aaecef
a782a0886e45396519e85d0013a60dc096b3f735
54579 F20101211_AABKSA aleman_d_Page_035.jpg
ead7de9321c221c8256552537c6ffe8a
1ee75fae0c9d9f5efc6243a1959ab7ad07e5130f
6785 F20101211_AABKQY aleman_d_Page_073thm.jpg
fd4f00957d30d051ae756f7e2b67bf9e
5f6a176b46ef7513dad8809ec9df87acc2c43e34
938 F20101211_AABKSB aleman_d_Page_135.txt
73c7f1d7ee3f0fb5ee5156debbd6a1a3
7118b5bad9998eed5af7ba75084e966d865de2e5
67272 F20101211_AABKRN aleman_d_Page_084.jpg
18d099a20b25111278aba592c63b2305
baca932bb1b942a79a89208c633bfb907776d1c9
6210 F20101211_AABKQZ aleman_d_Page_037thm.jpg
4f582772721457adb768869749e03917
1bf00ea1252f0d72e01212f93a1703c7894a1026
21861 F20101211_AABKSC aleman_d_Page_141.QC.jpg
fa246c050216e327c2e5e233cfab4690
0ec8d8e8cc9dbe325c4c4ea9c8ff443f4eaced25
78594 F20101211_AABKRO aleman_d_Page_037.jpg
b139430c34f0b0f5f9eca5a08ffc4210
ea5a327f7cb511851444dd72f1e855f1e6899cbb
1001 F20101211_AABKSD aleman_d_Page_140.txt
b77e1cee25d90810d1bf6778a2582997
89ca01e9973d46de3e5c1e0af5e7d3b83ee111dc
1051976 F20101211_AABKRP aleman_d_Page_116.jp2
39c2fff23668257485bd51c5fa2f9fe7
f056e5b8ed7a292a042b6a1398c50f895bee19d9
F20101211_AABKSE aleman_d_Page_060.tif
9febe19dcbfd7ed60a8b43d4ed943cf9
7f109a137f86bc91b867b552a92c0e2daafc91d1
120286 F20101211_AABKRQ aleman_d_Page_094.jp2
99750f93ac455038c1521735726c1518
4f2644a4e0d697867605ba9233be8591415e044a
14583 F20101211_AABKSF aleman_d_Page_112.pro
8727b82079614a631fdff179ad2eff07
75bcf078eab9b5df183977e07f26b11076d2c3be
119169 F20101211_AABKRR aleman_d_Page_070.jp2
148536b9896a61ba4cb54dacbc695b6e
5f489bc208a725ffb0d510a6ab4933d72f1a4b28
17657 F20101211_AABKSG aleman_d_Page_108.QC.jpg
0799e9ade0c53458c133b234b26451e1
92e40caec1dcc37595252502c536b8715dee16c9
18890 F20101211_AABKRS aleman_d_Page_061.QC.jpg
91720c90b4c289bac3340fd9355cdc3a
458c0a03478cd60b95bb4ecf145a9c17226bd93d
79140 F20101211_AABKSH aleman_d_Page_076.jpg
a9c1675daec88a0b2c5e0f3fd1ad8555
651cb17f7d23d9160dc7434bae793be55f637da5
F20101211_AABKRT aleman_d_Page_117.tif
5ae07f9c9a138a82fe1847db471636ab
e3b9fd07d1de07134a9503627e7ad5e3e5f53502
138379 F20101211_AABKSI aleman_d_Page_142.jp2
cfe389adf1b1432654b5767c46d98f31
111222f54ca6e13ca5beb8d8315f349d7cca4bdd
2169 F20101211_AABKRU aleman_d_Page_147.txt
676cfad5a7bd11520df46b667f1544ab
96f00b6be12a1858ccd971d522dbdec0b0904e01
6272 F20101211_AABKSJ aleman_d_Page_070thm.jpg
ca4f2f1b56e000d3215bb13a4fa9bdc0
b8d9ff4547d5a64f554bd5413eacfcccde719631
5112 F20101211_AABKRV aleman_d_Page_091thm.jpg
9bb2aae7e02cc5d53e2ee5ec3f360290
69d889f5bd3e96f9814cd405043bd1e72751150a
17258 F20101211_AABKSK aleman_d_Page_105.QC.jpg
00a8492b470e19013e2065aa4619ca43
573672fb2f37194e65fef8a72d0ed44ce47c635f
3556 F20101211_AABKRW aleman_d_Page_044thm.jpg
37b29f054425ca7a115a5ed32ce912cd
3a22a5c8c871eb150fa7d7186140d90621597294
F20101211_AABKSL aleman_d_Page_049.tif
7d002d1580514dad318f05f836599e3d
8e2f33742347dfc60ec220c417ab94270686bf6c
70769 F20101211_AABKRX aleman_d_Page_057.jpg
3ce2474a8fbc08a3f55eb8f0802e1298
059fb8a73842a8e5953f1b4741bbb1ced97e5400
21359 F20101211_AABKTA aleman_d_Page_147.QC.jpg
c442abc2ea97fc70cf15b9e7400d6b80
2afdbda5f3b5e6662b64569fb047b712c7961732
74361 F20101211_AABKSM aleman_d_Page_082.jpg
dbdb12d5f82a043507292a7278348bde
a13f3dfca8b260035e80673dda7e45c0c2aa8565
1042956 F20101211_AABKRY aleman_d_Page_057.jp2
bd2beb75262ffaaf3d08d873cfdd8f67
1cf23dddd799d00dfd4d3daf9dbc79781b57aca6
F20101211_AABKTB aleman_d_Page_021.tif
5808a8d9b1acd000527fdfb12d33a1cf
34180e2dc4deb08f03dee84cc29b9c0629d7b7ab
2296 F20101211_AABKSN aleman_d_Page_014.txt
0a34edd570d610783b68eaf6a17f7d4c
1422abad0d8c1e30d1bf3b076d48094268e1d1cf
1053954 F20101211_AABKRZ aleman_d_Page_144.tif
903a89d84f3acc832774b45adac30572
ca200538e09167bfd712526db8ad127a39e99353
25363 F20101211_AABKTC aleman_d_Page_075.QC.jpg
df9ab9140cf07f9d9499d6c961c1d161
28fd7fd0ea16a5cdb61f37240ee3eb710a116181
59202 F20101211_AABKTD aleman_d_Page_044.jp2
da77883a030a2bdeeb6f08c32eb3883a
0f812123497e07b464e1f38fe56261c0a00161cd
134333 F20101211_AABKSO aleman_d_Page_143.jp2
fa85693a69b4389c08576d6f7989f791
4443fc88629a7dd3b652aa958068fe060319b8ce
6782 F20101211_AABKTE aleman_d_Page_123thm.jpg
1f897ba52f1f87d0ee384b6b2c457fdc
cad164305e29a7eb7b3951705c32a2dab2953468
53505 F20101211_AABKSP aleman_d_Page_075.pro
2c6744c09bf1e00d9794ffb0daf75f77
f1abe026de39f2cd72472caba9c3161e4e06ecbe
78007 F20101211_AABKTF aleman_d_Page_017.jpg
45b8f715bb2a2c5d9e4c2da4e2502e99
60972e6e7d3138d3fe06d4cc73425c12ec2e2fc0
F20101211_AABKSQ aleman_d_Page_131.tif
76b0c60ebf51599203dc9c55bce1421f
92c9c7aaa3349441c5a5d611e8db9c312989552e
F20101211_AABKTG aleman_d_Page_036.tif
039593dcd63065488b57e1e2ccc8f3a9
2f5c4afcbc149e70e580b26a73e457bd7d30353b
5029 F20101211_AABKSR aleman_d_Page_128thm.jpg
83e0f5ca58b7b98680b021f8d973317b
0279668a261ccc9446fb15cbf5c00117538dbc5e
24858 F20101211_AABKTH aleman_d_Page_144.QC.jpg
67c68e80c6da339b8e4f5687490a011c
f40d1728c927231c1b6afe8dd69a0beace66edaa
F20101211_AABKSS aleman_d_Page_045.tif
f86513cc0492cdd6ce07dc3259845f27
dd255e8208f5840f275d9fd862d13db59f1f0968
22992 F20101211_AABKTI aleman_d_Page_056.QC.jpg
fc40dac35b0bc8fe8cc6d2a27d2cf694
76d0735ee0145fbe45e38db8471264014b0fded7
5764 F20101211_AABKST aleman_d_Page_077thm.jpg
9d44f77d6a944edeab1c94fadfb3dd79
38ac9fe1dda321340c30fdfc34af6a6a89607f36
F20101211_AABKTJ aleman_d_Page_139.tif
a125a6f775e6c4f5bcb86883016c9f68
85bc4cb3cf6c69bcf1e17ea8c44804f6bc956d7f
69512 F20101211_AABKSU aleman_d_Page_006.pro
702f00796674e3e925a2ea615c8f3ff4
0bdaef2d8b40c5074a7641653875ba8ce560eaa1
F20101211_AABKTK aleman_d_Page_008.tif
2739b4a56a036a885dd1d9d9d772cf40
15fee056269b6ab5321289917f1a8a8f2db2c800
1051954 F20101211_AABKSV aleman_d_Page_118.jp2
277d52d5db07c0bf3a89809421c448c6
f87f233b0f85f46686fa235d330e0e7ae806ac54
F20101211_AABKTL aleman_d_Page_015.tif
54b6ea912b6b06a55c906355ddb3e067
28e440c2a38423b4462f5129735a5814e4bfff00
2219 F20101211_AABKSW aleman_d_Page_021.txt
1c058c3ebed3173ae3b81fded7cba93c
4789ee7ec12109adb60e54581964f9ce3c8581c0
F20101211_AABKUA aleman_d_Page_003.tif
170223d6261dda05368d8ba34df39010
be76ce4db4278877d40174956d7ffd90e29df3fe
4790 F20101211_AABKTM aleman_d_Page_119thm.jpg
96e3530131d8f23c40690bd72353b402
d5e07e9fb11ad8da013b94c31c3b54feb985218f
15366 F20101211_AABKSX aleman_d_Page_114.pro
3cad785cc86b17080881cc54e7ab4f57
b5736cd5377e998ef6594fd37c870e58d967b162
85096 F20101211_AABKUB aleman_d_Page_055.jpg
354787efbf1b0085f993c40c80c4f512
0aff0ec0bfb4e334846feb4c8f59f3c9e5837529
56080 F20101211_AABKTN aleman_d_Page_016.pro
b339250704bf20ce35307f9864016abf
ac5e3c920cc9b91ec77ba8d1842f1cfc007459e2
F20101211_AABKSY aleman_d_Page_098.tif
4d460281481d37eba6a9988ec943831c
226473639aa65bef70fc9d83694f5f385ebc1333
75548 F20101211_AABKUC aleman_d_Page_137.jpg
44cde928033c3a3c9ac4d7280ba94281
18743a5d4e12ca2f273d685b892bde335d88589f
41081 F20101211_AABKTO aleman_d_Page_119.pro
3c0104393ea1c038b68c3f482f3ad95e
8e95858f6488a33a8cdaa084cbd588ed3fe4da47
F20101211_AABKSZ aleman_d_Page_096.tif
426aa91bcb240bf4316475675e2cc97e
9014ab785ff2df492bbe48ed76564d47fbf1436f
1051959 F20101211_AABKUD aleman_d_Page_105.jp2
c20461e208324d70a12b57611ec30a20
934d96466e9950893ab32cd4d34827c2a3f331cf
27488 F20101211_AABKUE aleman_d_Page_123.QC.jpg
82d08cbdad2b67735e429c38c2edaf37
4615fe5f040fd2f82059b5eed4d125234f11f386
F20101211_AABKTP aleman_d_Page_116.tif
b100597eeb7613b3fa4cc38c41d0688c
2293978abcf1512ca3f7d8825361f94a9bc68266
F20101211_AABKUF aleman_d_Page_022.tif
207a03901a98b2d6cf92afc22c577de8
2f88b59708f284b4368f6648a0eefd81c81f7d5c
26197 F20101211_AABKTQ aleman_d_Page_036.QC.jpg
3d375476a644bdfa27bb7dadb34334fd
85fd4c2935860bd49b5755c36d9961771f14ee51
2329 F20101211_AABKUG aleman_d_Page_081.txt
dfdf9ee5de7fddbf4b2048f9351fd2d2
cd52078d147766378175a7e55755df5218df31dc
F20101211_AABKTR aleman_d_Page_145.tif
399a584e1db34a0718cbed8cd7306bb4
ba57769ae2dec9400e3c95f2df0b79b954387fee
5451 F20101211_AABLAA aleman_d_Page_105thm.jpg
a03c9e89b66b5ccb3dc1ba25470b7bea
2cecf7c9bd74bb9ad3bf624a80b2ac3f10d55dfd
120009 F20101211_AABKUH aleman_d_Page_013.jp2
4837e5ea3079dbfcef6b78578e9e7e7a
2c2f8315488fe3d72365f55c0ba85bda662711c8
16443 F20101211_AABKTS aleman_d_Page_104.pro
ef63dba9ad1d16977ba81f944b6fff03
f245a7bac5f4bc5632a339ee7353dc7091d74772
84777 F20101211_AABLAB aleman_d_Page_053.jpg
b7edadb0daeee535a6d51c842e6a3061
738b3aae2a5db8b05d669c761a4da41b4c18e520
23251 F20101211_AABKUI aleman_d_Page_079.QC.jpg
1123439ae61b3d3647d98d99a4809742
9d018fc1a418d94579b3b86c473f3dd946359bed
2272 F20101211_AABKTT aleman_d_Page_137.txt
e778d17a528f0b41fa7a28ebbb65d63d
04dfd84d8b9176da25b7e1fb7982d01709e94eeb
2399 F20101211_AABLAC aleman_d_Page_052.txt
a600c2f78c23b2c39b21ef2eb2f3e2e7
9c131cdb6814f605378cbc477b6ecc39c709849f
50847 F20101211_AABKUJ aleman_d_Page_095.jp2
6dbef86bd47507789d1d90c7a9813417
f134021c1ae44762eb795ace8824189fee1ce4e7
F20101211_AABKTU aleman_d_Page_130.tif
c7b337fd923f5ccbd1de33a35132b089
9be95326ab6a5d4451bb02838b9c673a28b17e56
86560 F20101211_AABLAD aleman_d_Page_092.jpg
9336e8639c75fe12487641617ec5a5ec
f3c662a2e3f47a486d7307407bb9f8479f1206dd
25200 F20101211_AABKUK aleman_d_Page_142.QC.jpg
d3b434b777752945d1d29e8de054c5f0
4fcb3a7d9202f833ddb82daa3cc1dfc501ef0a34
F20101211_AABKTV aleman_d_Page_034.tif
b6cc7a736a80c30e2b7e0d811a0beb6a
4b32508f3e34b8084b81e16ff7f0b0455ae4c641
55155 F20101211_AABLAE aleman_d_Page_140.jp2
6ef1dacfcbef8b9f0672ea368669766b
064406badc2f5ba661030be8202f9b13dfe69bed
F20101211_AABKUL aleman_d_Page_092.tif
a6920588c7bdb5a77afc49c12db91cf4
b89aa273730e8627130c9c6438848e91dbf1e5ca
F20101211_AABKTW aleman_d_Page_090.tif
ebafea216a2c8ad4da2157585ababd77
e4b0229a7400a1488cb7ff8732b8a5d5b4f0dbaa
24561 F20101211_AABLAF aleman_d_Page_145.QC.jpg
88947e386ea94265a4a1129a583530cb
f80cca7d8202209bf2a5ea96dba59b236b092e8b
75906 F20101211_AABKUM aleman_d_Page_097.jpg
13702df6ca0afe23d6c282a43771569e
cc8dc64877dc864f1d0f6521361dba4657912c10
6345 F20101211_AABKTX aleman_d_Page_059thm.jpg
0ccd5c2e2ed796ec723b3950c4e536ba
13d9e0fbd2b09e85fbdb4760707c0c4ac729d037
4792 F20101211_AABLAG aleman_d_Page_042.QC.jpg
e5782ae978982d86408ed7c9e2ca93a9
77ccac794fe82750753b1ccc2d56930c5af9c954
58170 F20101211_AABKVA aleman_d_Page_014.pro
d8210bb921a0484b55067d9e33f29d9f
7f43a8414a819411b03703bffc6b763cf42d98eb
1051935 F20101211_AABKUN aleman_d_Page_120.jp2
0bb0e71f0d9488831faf7e03b8f4dcd7
fc3db61c909adf65fce051b9ff47087a310d7987
6267 F20101211_AABKTY aleman_d_Page_137thm.jpg
f2ecc360a60f9dd662d5907df25b8882
6786d22bf74372da2f6265491310dc00fcf71d3e
67601 F20101211_AABLAH aleman_d_Page_122.jpg
fd56e414a64d84b48d9aeb7c9e800fd3
68e4cd1ab47f7d56c7c53f0ae0b5374ef098d865
F20101211_AABKVB aleman_d_Page_121.QC.jpg
ff367b53c56ba08ad95339d2d3136150
7253fb6afaa353dc194a0f0af3fb48bfff7c4cae
22394 F20101211_AABKUO aleman_d_Page_095.pro
c65a0f41c7a7c793b0cb8ef91a03ca98
d5d15fc5981f228f032e2201e0c0a75c8147a1f9
938901 F20101211_AABKTZ aleman_d_Page_027.jp2
06bb25bc7894b84214b66776cec9d336
da4c1ea07c87e90e3b30202f440e7089e22a0c1c
50897 F20101211_AABLAI aleman_d_Page_086.pro
ed4b61dd9ea9b8d64d213e8c7dc7d978
a1d8d5c62e3a5e2f4e6659cdb2ba0ff44401f1b5
1056 F20101211_AABKVC aleman_d_Page_117.txt
6c57c02623eb6f2fc198a593c750cd2c
7a534d8545e1c7aca2233b86488452956276364f
2060 F20101211_AABKUP aleman_d_Page_032.txt
d085a0364ee044704cb8d1a967b3d9b6
92c26ec977bbb3bfd4f308f54e014bd7c234af2c
20715 F20101211_AABLAJ aleman_d_Page_068.QC.jpg
742b61588cd3e1ed51c86dbff71a8115
6fd8311b5c1870f9b57687d2ec4370903ca93706
F20101211_AABKVD aleman_d_Page_014.tif
e49812ec7052fafc2eef9cfb8ac06b26
4d408de733070d94bef8c8a6c87ffa67d56d1789
F20101211_AABLAK aleman_d_Page_119.tif
a1a2f543c36cdff9211f6d46f0077daf
aff9e261563419878dcc0971d44d4cc37c6969a8
50493 F20101211_AABKVE aleman_d_Page_101.pro
c300b2e15042501936c27d9b41ef3fe7
7bacb934d45336c4b32769cb0f30d10466c71198
730286 F20101211_AABKUQ aleman_d_Page_078.jp2
be4d3e7c06594bae77cb74130085392d
8c93eb4c499c74f00c039a583a94fcf66e16b4c7
37278 F20101211_AABLAL aleman_d_Page_030.pro
aa693e26a82017359acd3ba07ac58bac
569bbba8b3b8a2282087afe07027fc09d1d9bd4e
58778 F20101211_AABKVF aleman_d_Page_039.jpg
088f09e1a8adc4f65cdcb398a46f23fa
926aafc57b2bd6c541181859a9e699ee9494ccd0
2281 F20101211_AABKUR aleman_d_Page_063.txt
e9e4051d611b68af364c734403d0fa7c
b9ed24fd151d53e31e6f76cf23d1b658a58dfea1
12708 F20101211_AABLBA aleman_d_Page_003.jpg
7e91083d4fbee0a4a0b5d5c0770f861c
4c48c75dbd84f9eb721c51f77fea921c29049d8d
55973 F20101211_AABLAM aleman_d_Page_119.jpg
d2e6aea7dcc9c0f502e9460d19acdb41
236ab0555bcc5f5d8b90488138e4a8befb7dc03e
57655 F20101211_AABKVG aleman_d_Page_093.jpg
28061209daf3b6a655d3a57c499bc8f2
4978f9ff3765b676d3f9c526d5ae6d0c0340d4e6
55682 F20101211_AABLBB aleman_d_Page_073.pro
ccf3494b828e27ab49f554e31fc03273
a3628ff3177cf254d35a7e69bff8d9a8cab48ecf
86484 F20101211_AABLAN aleman_d_Page_111.jpg
6db5fa1055724b46c224070e93fcab04
9d99ada00054355f3d11dfba77cb986b851d7832
23928 F20101211_AABKVH aleman_d_Page_032.QC.jpg
ae6f79f223d2fe7a6683246f253b907a
c6d296c8ab03b742fa5ec12560c384072e4c7287
1659 F20101211_AABKUS aleman_d_Page_093.txt
84f849e9cceb2a6f903ac35afc0d7216
189385051793470dc5de6d731e73e8840c3bcf8f
17583 F20101211_AABLBC aleman_d_Page_107.QC.jpg
393bbdef415dd58f68a60c6fc1353920
cfa681e5a38d1c13fb79fd57403c987a61198a77
23617 F20101211_AABLAO aleman_d_Page_137.QC.jpg
cc3dc88ffffc614f1380b4b372859a55
7a9f7d82f36f940cbb7897d828a368382a0d2796
F20101211_AABKVI aleman_d_Page_042.tif
571b54e0cde0b962a88a3c65c3c81d32
b2acabe7623bd53e333ce6b2b0a2103ed36ebf71
22358 F20101211_AABKUT aleman_d_Page_024.QC.jpg
b6b7ea8f21ceb12cc9d3d16e6a2ab14c
f633a0521356e65d908fc31b4fa7155c627b2058
1051960 F20101211_AABLBD aleman_d_Page_049.jp2
eae10b7ce595302ddc7126cee736e476
12fd23b5d51df42c48be79e28b3c0e529c083011
F20101211_AABLAP aleman_d_Page_146.tif
99a23b8fcb029c7feac72696db89c440
8bf67a90626bd425a1ba94e7c5bae50d4815bf79
12536 F20101211_AABKVJ aleman_d_Page_136.QC.jpg
c664703dd51a803218acb61be9054b10
7ab0dd3fbf06fb7fe68309cc3ee4a4b2a83c5ff9
5689 F20101211_AABKUU aleman_d_Page_127thm.jpg
0c6e97359fd00a955e32c24bf7344352
f09c9a4383c5a08dd05b31d47698921149be78c4
F20101211_AABLBE aleman_d_Page_008.jp2
93dc223670fb29a97ea45cd991f58641
34a21458d048a1048ab8764ac8e7a297bedf8c51
6526 F20101211_AABLAQ aleman_d_Page_053thm.jpg
80acf16da1ee4485f7a4eb7ea0a1e1c0
089dd2965dea80bc65cdec8967dbdc3f8ecdb7d5
57349 F20101211_AABKVK aleman_d_Page_130.jpg
b29ef3b933d82c365d8ea5ef37b3a2b7
44e0cee3fe3b87c2484664af6e11f706b55dac07
18108 F20101211_AABKUV aleman_d_Page_091.QC.jpg
2e4c4ed75341189ea62cb7dd21d48702
945128c4dc525645483c55bd883ea1bddebbe44a
111155 F20101211_AABLBF aleman_d_Page_096.jp2
8ed0cfd86081e906d0e46d6413637afa
6a333c58e1c5798e992c8cc2c62669d6093aeae2
56460 F20101211_AABLAR aleman_d_Page_115.jpg
385a43759cc5e686888598bca99b771e
7b3b6e2277b22934f868212d4d03c4c3613cb4e1
1051972 F20101211_AABKVL aleman_d_Page_137.jp2
55a69070d8b2e79ec54a05565cc962cb
7292b5ff25a8738b43a565ec8946eb9bffa7a3fd
89240 F20101211_AABKUW aleman_d_Page_071.jp2
f54700156c45f8e9419d25cccb4b888d
88438d29dcdc521df9be81783c69a274ad22976c
38526 F20101211_AABLBG aleman_d_Page_068.pro
5472a5280ea5546b9214f2af45b2ba60
23f742f22b936038ae01ea0a4b3da1136f6844da
12405 F20101211_AABKWA aleman_d_Page_140.QC.jpg
bb178deb3079a2e60fe29c1afa62bb9f
f419cef640b2943c1829721e12ea6d6dc13e16bf
1561 F20101211_AABLAS aleman_d_Page_061.txt
5cc6d1b7c838722c4f11e1236acaebf7
c04d7bbc60c6e2be13efca7360b48ffaf6b8b10c
6322 F20101211_AABKVM aleman_d_Page_143thm.jpg
47762e962c4a293819662245c9922416
6baff4bdb9f66991e17fdde6ebed4c0a673ec293
37085 F20101211_AABKUX aleman_d_Page_091.pro
45ea045ff4dff6f6cce34eb6c40ba581
9df123db064d037a9d3eebc888689813de4b9a6d
45443 F20101211_AABLBH aleman_d_Page_020.jp2
5cd7bb0229a8235f14c5079d990df077
8b9818e2068d952e0523ba5dbc5a193d33174ff9
6357 F20101211_AABKWB aleman_d_Page_121thm.jpg
a518d682ea0f116a2ddd5cea85ba0eb8
f70ee01bf7458dd86c3e8ca4761d0c244e7cfc87
5640 F20101211_AABLAT aleman_d_Page_115thm.jpg
bdcb4a3339abc573f9175b29486a3f38
20e70772c971d8ccc1726275a912263f3fc409ca
27983 F20101211_AABKVN aleman_d_Page_052.QC.jpg
4cdc567d5cf67124843b1176253f0883
1d9931e5d29c90858ec26a7ae1c5afe6ab37b422
11860 F20101211_AABKUY aleman_d_Page_095.QC.jpg
0c1badf6e7efa6ef6f16563b7551ff1c
1bb76aa4776fa331c1a7f8f2798aeb57ccd9a5fe
38556 F20101211_AABLBI aleman_d_Page_131.pro
aa0fe5765abdeb0a5398bdcbee3cda7e
6ead12fcc3c15944e8033e288a1b2372866ed737
85236 F20101211_AABKWC aleman_d_Page_143.jpg
17933e3384a12b07cdf009f8d327c97c
decd75d4d846a4d7dc490e9dde402c3e7a3b9335
83688 F20101211_AABLAU aleman_d_Page_062.jpg
a89f00624a0ef86f0dd979f1f04fbd86
7c51938983115707ec386c82d290be295eda88e5
2278 F20101211_AABKVO aleman_d_Page_023.txt
fe7831e97ca71cd7a22c4ed98a474e6b
b4e1103e671752f14c3bc0f7eaefb3914e8c2268
5676 F20101211_AABKUZ aleman_d_Page_108thm.jpg
2fa8dee876c459b52ba92805a7fa7e49
8276d9633ef4416513699e664f2c97e5bbc8a4db
3220 F20101211_AABLBJ aleman_d_Page_002.QC.jpg
bedbcd1c438381eb491f45b9faa761f1
69e5221ae558d14da24fdee6502d8f0b9d843d3f
5911 F20101211_AABKWD aleman_d_Page_141thm.jpg
58201e89a266f89a648b598213936ca6
af87dd5e51c486a6f4eb728b24622a7a3346d0cc
16935 F20101211_AABLAV aleman_d_Page_135.pro
2ea827c1fc793e122a5a2f7404358773
2597b58932a433c76c737acbcf3d43338cddc058
49976 F20101211_AABKVP aleman_d_Page_138.pro
a374a98a927fd328bf348ea8df778fa9
965c5e498c66adfb1caa4331527eaaafcb978dd5
6804 F20101211_AABLBK aleman_d_Page_038thm.jpg
20099e3d4b243243f2851901ad28cc65
b4a985090214ac378f27f9e9010d9d23832db630
1051979 F20101211_AABKWE aleman_d_Page_047.jp2
fe38a8e7539287639d212615773f7549
f42958b5b377e79868a5c365357969a070583cd5
53437 F20101211_AABLAW aleman_d_Page_040.jpg
3ed74978f37e4181b3e43a416b8ab079
1ae11a51cd37153071e2dea163a6460d4daa4774
1051967 F20101211_AABKVQ aleman_d_Page_108.jp2
f5bfe0972b0984d93d6093961ff72433
6cc22c13b96ed28b0c6dee47d0176abb4ec5e5ec
5825 F20101211_AABLBL aleman_d_Page_082thm.jpg
f4cfa47857a2a885bcc5fc5da301e5cf
79a8ddccdf84ceba8479cc5b101becd9b69b63b3
5984 F20101211_AABKWF aleman_d_Page_024thm.jpg
9b019ad91506833d6a264e8ae3a1d38c
27ff78450e103adbcbf4fbbee8f917111a3c2411
48061 F20101211_AABLAX aleman_d_Page_057.pro
85a366a7b0627ab1205a768e850927c4
2ce0ad1c5535852e316c9d98e0b22e79c7abb0fc
4955 F20101211_AABLBM aleman_d_Page_030thm.jpg
b90fa9a75926fd834744cd5b4fa4b494
58c4be3f8be72a1d1e0fd6dafa01bdc13af30be3
807297 F20101211_AABKWG aleman_d_Page_091.jp2
fb4459599374d8268965e515f92ca875
a4304b973b531f18ae7939b2345d8fafcfc50f7e
5980 F20101211_AABLAY aleman_d_Page_096thm.jpg
daabc7fa5fd46ed9548359393e441f0a
e62656ab0c30d16ed5804a2ee2301ca9aaad4648
993756 F20101211_AABKVR aleman_d_Page_007.jp2
c218d2b57b974bb9282b59123311e0f3
4441e1a549af6e32e2b0a6580e9dac7c663c027b
F20101211_AABLCA aleman_d_Page_073.jp2
ef4d75b06e86dafb5ddc8ee6bb81e01c
2cc59410a366409da40ee920904390a03ded4de4
25020 F20101211_AABLBN aleman_d_Page_140.pro
68866cb4dbf7d296b47b4a124b059a35
2ad3c66392e34357582e0dae3bbed685ac3bd2c9
6063 F20101211_AABKWH aleman_d_Page_090thm.jpg
91ebab78d559deccd5db390d27adfc6e
966adbd226406a190d9d8682c5fd456402e3c455
1051984 F20101211_AABLAZ aleman_d_Page_092.jp2
85835bd195e2d68198f5b14b1eb1d488
d2854e97cbb6e885eed7c2962d78a85e4d25ce44
F20101211_AABKVS aleman_d_Page_086.txt
ec1b663010a6d2c331cbc3468d64712f
723cd065253911126320e655253294ebcdeb4643
33526 F20101211_AABLCB aleman_d_Page_064.jpg
ba7f6e3bba1fd79e48fca539d0febe19
ec7ddedbd9c3741463900a200f6091637f033718
53489 F20101211_AABLBO aleman_d_Page_045.jpg
8d4b4db8c7e718b72ed54bd151ff257e
d1582a1cebc30398032685c5add2d601dd669bdd
F20101211_AABKWI aleman_d_Page_135.tif
7077a8ccf81ac69d0a71ce1c1dc24350
5e67ee813c0514e91a4e88d7b19f1036177b5a83
6557 F20101211_AABKVT aleman_d_Page_026thm.jpg
401272ea0d7457a4b90eddb9a99eafc1
5e692f8b53783703eb49863c20a7d821a22ee5ac
7017 F20101211_AABLCC aleman_d_Page_041thm.jpg
45ea343f48609dcf019955b845760650
d7a81c51e4a67f15231b8a507e715a643ac2dc4c
21176 F20101211_AABLBP aleman_d_Page_084.QC.jpg
c7c269fa76ca7dd0995d9926ced0422e
48a152961c0ad04bf70c2f94595fa63d8cca21ee
6701 F20101211_AABKWJ aleman_d_Page_083thm.jpg
dbebc34acd53940d4f101f63de5212ae
d4e164c4c38daf89676fe8cd2d5dee9c58a8900f
53931 F20101211_AABKVU aleman_d_Page_110.jpg
8953479e63bf66f7b952f7ca931ea7cf
0f38c8c57d0487b65dad1821978e5b9210ff2105
1051980 F20101211_AABLCD aleman_d_Page_075.jp2
48c1f016588032ad39dbd1cd756c2fee
2c387da625c9fae0bdf7692a4ece163410a257fd
1051915 F20101211_AABKWK aleman_d_Page_083.jp2
ee866c14e21d72e5834049cdc12da37d
7336c18894143b2e8b2094e158205e91a9e86a66
F20101211_AABKVV aleman_d_Page_062.jp2
b1a0ea433e4d3fb1728c1b58b3f945dd
ba7b0f93e695fb8b283eec441cfc5f55079c7ca7
2199 F20101211_AABLCE aleman_d_Page_073.txt
a66851933cb66b5c82df821b1c9bcdf5
91a37801bd99025716023d2ee1a607798dd1deaa
61846 F20101211_AABLBQ aleman_d_Page_041.pro
bf9c5652103798201b0de7944d3746ca
238f1805aff0685f5cefe5114d677f182f5e7617
5827 F20101211_AABKWL aleman_d_Page_138thm.jpg
d4f50602d445c5ee976b0706080011e1
014182f680b724a3ae38672dd59d7d115c2afb9e
22411 F20101211_AABKVW aleman_d_Page_098.QC.jpg
1c1be0d3cc83b1e85f803d51291060a2
e1da49e3a09edb0bb63433cd89f24b093902f593
5594 F20101211_AABLCF aleman_d_Page_124thm.jpg
18551326c3e57b214300bc649a3d6a89
ff2088deb513f1dee6c5bc9b67fa9334229cb3f9
F20101211_AABLBR aleman_d_Page_127.tif
d951e260e0d2310f02a3f5ba8573f50f
67921d4749489f6cafeb4d4078e2af09951e6db2
1458 F20101211_AABKWM aleman_d_Page_125.txt
dd69194f472b83b29ce967168fedc974
a955548e4ba10c07cdcef97d6eb51ab128e3ec6b
1000888 F20101211_AABKVX aleman_d_Page_024.jp2
d695ce67f642144f2d6afa0a4776f08f
f0a55cde61bc7894dd3f7d368dec671dcfb2a78d
F20101211_AABLCG aleman_d_Page_052.tif
2474fc5e70d0d3f8658d34eeee3552fe
f4550ae897cacbeb30f37bdfa7c856b941c9a3e9
F20101211_AABKXA aleman_d_Page_004.tif
5401721c6271dde470c5dd05cb81f011
45dd3bb56dd43e04a77492d4c841aa84f20036ca
6524 F20101211_AABLBS aleman_d_Page_049thm.jpg
3def1385bfff7bfc2917ec447a744808
71eaa2bc60fb3ef0c42327ec294e2adbb0309d54
F20101211_AABKWN aleman_d_Page_007.tif
0b0765ceafef3e5133fd8042a1b7ed0c
fa9ba8082583a8f7818c6239ce28dc0346c8bac0
323761 F20101211_AABKVY aleman_d_Page_067.jp2
6fdf7ccb7aca4567f32831ba7ada2959
b39ec2f023c75a12245692cbc5304b6061b2091e
2003 F20101211_AABLCH aleman_d_Page_132.txt
84bc2f7132b1f06c04188a79d89a3cfc
9babf1ae944cb3cc08ef17adadc7d5150edaad4f
1610 F20101211_AABKXB aleman_d_Page_028.txt
747b601528677975d95b9eda60dd066b
c91eb3792a641d553cc2ebf1d5d611050164da4f
F20101211_AABLBT aleman_d_Page_069.tif
7228cbd58a4ceaddc897c9305df27011
d26f57a7f7a0e5cd24403f867d665a4bc1b305f4
25315 F20101211_AABKWO aleman_d_Page_134.QC.jpg
508baadea79added13f7ce7a404af273
e2dfa9cc7d891cb450fdecc23fa324209570915b
57117 F20101211_AABKVZ aleman_d_Page_125.jpg
c706ed913c453cf2bb52404d422410dc
3fb072f9c15fa2931c8729e5f7af302157f2359e
F20101211_AABLCI aleman_d_Page_086.tif
c098f4161a5ff3fcdabdc160376e459f
1ab241ebb2f03f729b5b64f332637cc0e968283b
23153 F20101211_AABKXC aleman_d_Page_006.QC.jpg
d8fd3540152554a43c525eb820621d3f
b7c06011616b03473dbe3b6998485f2ccf61f472
83017 F20101211_AABLBU aleman_d_Page_134.jpg
0decdd6a93752990d79d7cd4e8f02f0c
4b26572aa9398a30f3e59d7f256783b1c8db7133
F20101211_AABKWP aleman_d_Page_044.tif
da7f0d70c69fb017addcabafbeb100f9
4808f844e927e73aab1b69e7d31f5f15c6980b11
2253 F20101211_AABLCJ aleman_d_Page_054.txt
37ee430c9f341f6b3630e096add7003f
22a681aeca20ae8e86a46e470ddaa9bd20524852
17613 F20101211_AABKXD aleman_d_Page_046.QC.jpg
52d0e4433ee5bc686fb6f7e0129c7c64
db649c56f5e774c97878a146d68d991db4d78721
83698 F20101211_AABLBV aleman_d_Page_089.jpg
19906cad25eb9f0efaf55fad14d1453d
0496848bf2fbe35670bddc70dcd2d6c67797fa88
6328 F20101211_AABKWQ aleman_d_Page_056thm.jpg
553b2678ced5dae88132c6bc7765997c
498551dcb6fd155f1dd27d12f6bb53da163dd315
79776 F20101211_AABLCK aleman_d_Page_026.jpg
ec2b0a6d4733c5bf42d295df69676ad1
14a23c5ca9635501a25773c1a37949ea2e164ca5
F20101211_AABKXE aleman_d_Page_021.jp2
2f0f15328aecf485dce2c8cae0a37860
068d0027cb2b062d7a83580bfbf22c95f4a88d15
22597 F20101211_AABLBW aleman_d_Page_132.QC.jpg
044dd14ceabcd5043de3a2f1a2c3efe9
46851bb338a222a148312d63b0f4445d85f20464
70807 F20101211_AABKWR aleman_d_Page_074.jpg
fa559af21b1a798f54e3702897acce48
242f3adc706182dcff5057ac401e7ede20c22a90
1051973 F20101211_AABLCL aleman_d_Page_088.jp2
60eebd3fe806b0ea369c7e8a56013564
f9bcecd2e507583d870b5476aa3b21979eba33ef
71783 F20101211_AABKXF aleman_d_Page_147.jpg
ffac352b9d2ec76f1790665ae629bbc3
4b9b0db18edf66622b2d4cc0c29b0aa149dae4c6
1051937 F20101211_AABLBX aleman_d_Page_031.jp2
2fbe1b16ca3da77f7185709859f4ae58
b09632dab4b28c62b541d81c89e16c77dfe743db
72565 F20101211_AABLDA aleman_d_Page_138.jpg
44aa355f97b3b21c339253e4460145ff
8686ac3b239f38674c5265e6643d29f94fc35627
F20101211_AABLCM aleman_d_Page_017.tif
f9ebdd02a179bd791ecf9121e529c2e1
6dbf7d5a522a636ff05e10902914790c6c45a908
F20101211_AABKXG aleman_d_Page_097.tif
f7ae3598855a2313b8a06ff43410b0fe
9d6c1a71db5a90da3e241c62348e4e993750fef8
49391 F20101211_AABLBY aleman_d_Page_032.pro
84cd0b90b700c8ef76e68879ed2d49e9
981c3e9491822a3fbcf93ee1dfc6a46bc72f97ac
27731 F20101211_AABKWS aleman_d_Page_067.jpg
9f02cde652cbacdc5267851370cd1625
fa7ec4ad7297c8b24e0de6b1accd39546352eed9
6767 F20101211_AABLDB aleman_d_Page_120thm.jpg
8e35d75d688e738eb00e05f717a826da
8767603e1814d247cbfaa580e10e0830f319c511
F20101211_AABLCN aleman_d_Page_101.txt
ff057848e36db97af0c05f3d7780de4b
d3ddc90c405b3f0d1fc7cb54294eef99b4fca21a
5963 F20101211_AABKXH aleman_d_Page_015thm.jpg
602b4d14c1d100f2514c24c16abf4bda
e4023b9ba3858abba65781a1e0a78b3b0919c6a1
F20101211_AABLBZ aleman_d_Page_088.tif
03d5486f479de6db1c2e509de700bf84
7888707a83f5fececc24d84802a54ce1f5914b9e
23924 F20101211_AABKWT aleman_d_Page_011.pro
393bbc924ef83bb60accd060260d3e2d
41c4b1a9503f0b760a0f02581c87dd984c5e2844
2420 F20101211_AABLDC aleman_d_Page_047.txt
693c1ed13eb8b5900b8be6baec4923f4
826909bc356c02395d5454b3ceb35f147def6c15
87012 F20101211_AABLCO aleman_d_Page_048.jpg
d59480eee048ab49b0addca367300405
9a6bb044c25733f3be1cd00ab82ef02125424f02
24056 F20101211_AABKXI aleman_d_Page_010.QC.jpg
e3975300c2cdd38b60916a3c5a9855f2
f50cf53f1a2fcb8308ce03947087b99c332a468c
2364 F20101211_AABKWU aleman_d_Page_134.txt
1260db0738a61c42799072cbd9f0133b
7039be07ff9860094575f33cd1e21a8b39e565fe
17373 F20101211_AABLDD aleman_d_Page_126.QC.jpg
8c97a3e52ad8476092dbd068618678bd
04f24ed66a7dda347e5c58dc5ad03739d949d3d1
2128 F20101211_AABLCP aleman_d_Page_085.txt
ea245aea2e22217d4664bb9005ce0cf1
4eb9e6160b3065e11b2283d15c9dc3ed49e9a89e
5361 F20101211_AABKXJ aleman_d_Page_109thm.jpg
44abc2f78e18152f28f4d95f3e964712
cfa3999cc64787aef04a01ad63f7e5cbe8a403f7
5966 F20101211_AABKWV aleman_d_Page_029thm.jpg
6dfde3538978cbbc7098b6d9cc09bd02
f99675f72566711f1f0ef87105576aa98129887d
59116 F20101211_AABLDE aleman_d_Page_071.jpg
a47d6708d783535fe700289ee25748d1
b1dd0bfc3689a27387d932f8461a2077f71a52ae
81970 F20101211_AABLCQ aleman_d_Page_010.jpg
65aeaf113027f3ad5d193fe8502fa9b0
44b24176dde3e53553bfa2840ddccfd35158941e
2230 F20101211_AABKXK aleman_d_Page_141.txt
677d94f226d6d004c15feda5a3765f47
7f9d93295e4b7364d4bfb35d2e289833fc59bb74
40597 F20101211_AABKWW aleman_d_Page_077.pro
968b0085a13ec17a1e8c4dd1c98d6f78
074b6dcfec8ad963c5ce9eee3c523de5a46d4f9c
55411 F20101211_AABLDF aleman_d_Page_107.jpg
3139b90066936787f52fd82a581323ae
1b47306977bb6e3935225cd573b22ae68342a02f
1051953 F20101211_AABLCR aleman_d_Page_053.jp2
f39e06abffbdca8af056190b945c3163
09ed8c7f6adde402b76c6413bfa5584f65b76cc7
17916 F20101211_AABKXL aleman_d_Page_125.QC.jpg
0a56f82cbd701e4182ade45dde1732ec
957c7b764b5362b23faa023d23846e0611324ae1
75196 F20101211_AABKWX aleman_d_Page_056.jpg
b672f9cb238a9281d6ecc49223de9e11
3b1874bb1ce9ed01a28643fcd2c35043e211afa0
46076 F20101211_AABLDG aleman_d_Page_074.pro
e9f81cdab2cf203956017b053756a287
a7693b448a298188d12245e0503de36d2f613dd0
6194 F20101211_AABKYA aleman_d_Page_025thm.jpg
db68e1c7e24dac3a685262055d68a660
773b4ce1943c264c03a107dae134b15958e5c5b6
71111 F20101211_AABLCS aleman_d_Page_098.jpg
4e778ae07f95d0271322d25c259aade3
6718db520ac7a2f74b8ce23346f571a8811cca37
23201 F20101211_AABKXM aleman_d_Page_009.QC.jpg
955663823a0b694a5b5057d18a0ce547
a5c1ac2b63eb02a152d671ce3696aa73bc300062
F20101211_AABKWY aleman_d_Page_041.tif
bc66c36cee358895091b31fc94102abe
291f51f9dd028dfc9d72602b68f0ee6f499ce297
1051906 F20101211_AABLDH aleman_d_Page_050.jp2
c497632ed607dffe864f94867545b6cd
7f57d0500af070013da878f66176f84f00d9b52d
23043 F20101211_AABKYB aleman_d_Page_082.QC.jpg
adae41e6f7c86a041fb8fd363c8b721d
23f770bb42d9f6358551b19891df2ee19b980ddb
81137 F20101211_AABLCT aleman_d_Page_133.jpg
54472a718ba2263ddf23f7b680116df2
da0d79df9b1c5040aedb8deb25213a6e664be49a
2121 F20101211_AABKXN aleman_d_Page_059.txt
dcfc655e13948c8cb7f7cca27153a2a3
79dc6e9d4ef562c99fbeb64ee634639b26b0711a
F20101211_AABKWZ aleman_d_Page_065.tif
45fa55cace200ffdeeab45aa550fa74c
2ed1dfd6179818da4c8db527a00e8469bad240c2
53325 F20101211_AABLDI aleman_d_Page_079.pro
87fb9ea0ac369b9cedfbb7d70c921a93
7816ebf39653f1e0a5594cba2c072e3a3443a64c
F20101211_AABKYC aleman_d_Page_137.tif
0a672523af87645dd8ae4ad8b6fe0c7e
765b5c5040333a855b69c6ed4bfc184f9f1a8b68
2197 F20101211_AABLCU aleman_d_Page_019.txt
eaf4a0f13e8a0e2b22190a2d1dc59e20
8150bb9bf1e73107f07695eb78c8da7d926c515b
84570 F20101211_AABKXO aleman_d_Page_012.jpg
953f8de092fc785730650d2c7be3e88b
7761ffdd9cbcc9aded3f418ee5579aa202bb0ddd
49626 F20101211_AABLDJ aleman_d_Page_080.pro
60b30ddbe31b5aee9f407732f670fb66
90fe5b8e920e3ebda0995128b04a9091f9ec1521
23317 F20101211_AABKYD aleman_d_Page_103.QC.jpg
bd9cf33a4d6fd18f79b495da040505ef
52f07686d6e4040251d16d582d7bcd8ea5697b22
18359 F20101211_AABLCV aleman_d_Page_119.QC.jpg
4f1b51794ce72919f1cc456e330ca529
3af29cdc245e924fc59976fcaa9853aa1ed1bb02
F20101211_AABKXP aleman_d_Page_114.tif
8eb40675e5d6516ffaa8a190d799a11b
2ea49bca73c737e01d6af9d8ab91ec7b708defd6
3851 F20101211_AABLDK aleman_d_Page_007thm.jpg
4326efae9914a1afa075a0978294a502
7eecae0a8e1e82e791cdad8c633138e2cc38f5ad
F20101211_AABKYE aleman_d_Page_077.tif
8c5376b1928fe5fd6b18c3674552d952
3fb831111b2e3b098ab0638e7c6db9fe366fd71e
6933 F20101211_AABLCW aleman_d_Page_092thm.jpg
2fedf27fc88e8947ea56626f48b438af
604a4e1e01b836bc44acd8b0bfc946ddcb50a494
5474 F20101211_AABKXQ aleman_d_Page_110thm.jpg
02d09dc4734129c3e4eb6bcf4f2437b6
4cebfe33243f281f66494c701114794171c7e594
2119 F20101211_AABLDL aleman_d_Page_133.txt
a579aa7cf570ad586864fd74e308c805
4fa67861855c8178eb2cc6bc51a71dff483ee25f
F20101211_AABKYF aleman_d_Page_032.tif
597a5a2c3c5f752bf8402d1fd7a1c011
2ae472009fe5d66b76e3fb0348531e467d378742
908 F20101211_AABLCX aleman_d_Page_115.txt
b5d6e6acd3bc9e88d1395fa1bf45115f
198e86e3a24182424d11ffc35a4cfd8ba1dd9311
48515 F20101211_AABKXR aleman_d_Page_043.jpg
a431427ac2307b7d22b88f8a9a03ac9d
3f1b86c361d782e1f714b9dcd29343b2c9076eed
6036 F20101211_AABLDM aleman_d_Page_098thm.jpg
d865828dd3b00da4bb589d8dbcb0a6c1
09878267eedbee755e8ab610371f94d99b221d16
874248 F20101211_AABKYG aleman_d_Page_046.jp2
75926815f3d9db62765fcf348816b1ad
2f1f76c21a370d256f39da7a831ae9d3f9107061
112697 F20101211_AABLCY aleman_d_Page_015.jp2
577e72643b5fe11b81e38cbccff545a6
65a265f43822061e46fa95e1a19600dd687ec457
27531 F20101211_AABKXS aleman_d_Page_008.pro
a859bcf82c36dbd3042fb50b512f195e
b2dc5e146da2f044f6ba3edb99129180b8e7f278
42100 F20101211_AABLEA aleman_d_Page_060.pro
b7526613a36fe08987cda08cd2d3b932
6fac9db19dde711a3c44740e08ec615cb9aac019
6182 F20101211_AABLDN aleman_d_Page_019thm.jpg
19805ed4e1ccf7fc66adda8c85b348a5
ac7d78eb66e9bde6e61b957f57a7eabb4acf4a2c
37469 F20101211_AABKYH aleman_d_Page_095.jpg
5f71d2b5a71eff0d8c9354af9faf00bf
2e6d0b93f092d670cdeee30f3ea715c056f0f324
2235 F20101211_AABLCZ aleman_d_Page_031.txt
d7771c494fb8bcea02bd3e74e347bf41
94637b773f72282c742c3745a93738034b419637
57661 F20101211_AABLEB aleman_d_Page_018.pro
f5f413f13cc6edf563b73149b3ade4fa
33fc91b9d34356db750035bd35ce2d29f94933ba
6720 F20101211_AABLDO aleman_d_Page_031thm.jpg
2d8068a5ebcae578668a00ba1507b33d
5d03bee326af5473925a066e250c65ffc9854752
55702 F20101211_AABKYI aleman_d_Page_019.pro
19298c925c3d84aaae4070552503ee65
aa38bfb57e8890c8e5e39ea905a35ec21dd65d14
2184 F20101211_AABKXT aleman_d_Page_049.txt
58a44ba347d2c591021159a98877e7d2
e6e143a93ef3b004c2f4816da00ace5d44973c87
F20101211_AABLEC aleman_d_Page_039.tif
3896c71bc6e80387191bba4bd562484c
b734c14c71dbaa336a204bb6774e5598a1d307a6
1015211 F20101211_AABLDP aleman_d_Page_127.jp2
1628108396eaf7bc8105ac8f360e3338
ed1b4ac225460d8263fc355c052a60fe047bb0ae
115855 F20101211_AABKYJ aleman_d_Page_019.jp2
7cadddf2b5b00a3d5d4f8774b56d8fc7
b0649d5c087ea2e0a28d12259b01263f862fb0b9
1651 F20101211_AABKXU aleman_d_Page_003thm.jpg
bdaaffd8cfd49877564c255cf404a736
1df6942d7f56e3bf34a4f87ea57517bd11ecf58d
16373 F20101211_AABLED aleman_d_Page_035.QC.jpg
2438125ed1eb344fee998c96a7cc9274
fd239de7175d35eb82d8c78a23050e56f24803e1
F20101211_AABLDQ aleman_d_Page_072.tif
07703a2a621fd9d5f721d2e9b15372da
eec81b478351c7557c45612e3c8fdf2f58907631
77856 F20101211_AABKYK aleman_d_Page_080.jpg
0def0f8dd03547ae2f7b3e2e08303614
a47ee5b44ce3cb08a63d1c8f7317fcb3f521dcab
23513 F20101211_AABKXV aleman_d_Page_080.QC.jpg
af7cbd3b7c8ed1156afeaddf62417aca
ea8325f3fbaf7115cb4d5d0890928be0aa3316d3
3343 F20101211_AABLEE aleman_d_Page_064thm.jpg
9322bec0e22b468078a53f31897c3ac9
f6d90fc64356a2c3a726d82d2f5d8658bb7836d4
2239 F20101211_AABLDR aleman_d_Page_016.txt
9e3bd34815deefb08725fa90c8be5f0f
87ec426ae943deb5b57fb02cafb0b0d815f3d7f0
30249 F20101211_AABKYL aleman_d_Page_126.pro
40d327106a0eacad22c8c1020c50ffae
056d8e5c5f2f439cf9458b566307703e93174dc3
52680 F20101211_AABKXW aleman_d_Page_099.jpg
9ecf34b438124c64acc805ee1237b263
b57caebf57326749566bf2fe0452a10c4e0bd3d1
F20101211_AABLEF aleman_d_Page_059.tif
f926d3c8d0b35ea015bdb750f7a11199
7960f7c74acd363824f1bc97225b7f59d1264576
1646 F20101211_AABKZA aleman_d_Page_030.txt
d6a0569851fcbeeda5f98f25b402193a
66ca88488a32d624143dbcafa74d4c371644b282
965 F20101211_AABLDS aleman_d_Page_110.txt
48f8f42083d0acb003d02ea3201495f6
9f2423fa4e7be42295ce8aeef890c8eaaee8f3f4
F20101211_AABKYM aleman_d_Page_133.jp2
60896349caee3fa248ed903cbda6707c
02b54a0a98580ece8309633444bafc46e4908c7e
6411 F20101211_AABKXX aleman_d_Page_013thm.jpg
02ffefc7b7b1f0ae6c4853f3311903ad
517b5364313034d239ed1eb2f80e8077df27e37c
F20101211_AABLEG aleman_d_Page_134.tif
883ba035c5c37a181db46ceca17b7a08
38f0a2d10dfe129c37ac63262fd32ef39abdfff0
34380 F20101211_AABKZB aleman_d_Page_087.pro
4d0b4c6dd84e1d96bdd0c1e5ab625799
249f6c6783ff63cd6d2572404702d01ab0217c32
137052 F20101211_AABLDT aleman_d_Page_144.jp2
de8a5b2c61107a952e35192a11951ea1
a18b6eb8155d42b9d767578aeab029f1826cc3f9
43945 F20101211_AABKYN aleman_d_Page_139.pro
29e3c5633e0aff0926ab40a729ec9f62
7f82139a52087f2661f55e98609f73fc11add5e4
71807 F20101211_AABKXY aleman_d_Page_029.jpg
c3e4dc374b8babbdd5b5351cccbbea77
3d4db74a39c94cb513ce05b1157ff6bb73b2466a
22410 F20101211_AABLEH aleman_d_Page_037.QC.jpg
a269abf79f16cbbee946848dd1c82596
994bbea5828741c04e7f10802d01a6be87f83d50
75888 F20101211_AABKZC aleman_d_Page_103.jpg
c3fd9d61bf5d3fc02d0b2428c7c11cfb
b9a739a402c411f50e0b246963e6024e9d1f419f
1051966 F20101211_AABLDU aleman_d_Page_054.jp2
08b52929969714a1c440ea571b984ad4
2c85bd74f77b3110511063d69ae8a43db9fd8bab
1914 F20101211_AABKYO aleman_d_Page_056.txt
1ae6ca1afd2fc801380da042eaedc409
e36908d6ab126bd61dfe547a9930efbc2ca8f41c
6723 F20101211_AABKXZ aleman_d_Page_054thm.jpg
658e54701cb9c9e3a59314ddd87faa09
43ed7a488283b3833f9e399d8d08bbf992a10410
F20101211_AABLEI aleman_d_Page_024.tif
9e4cc7f9e64dba2bbefa8ce5a0bc19c4
d4016483604c8af735d330dee46a47dc5d5c734b
56725 F20101211_AABKZD aleman_d_Page_033.pro
0263f7850d71aa8cb9f29764d99173ab
9361887ff428caaf85f4417330bb025333714803
799 F20101211_AABLDV aleman_d_Page_020.txt
61405762aa7ae5cffc6392102742e26e
329092329f35fcf3434da88cd03d5c34c00ca01a
F20101211_AABKYP aleman_d_Page_138.tif
461bb31d75dcf5fbb406a8cfc0d379d1
aa98327e352c0e910033fee87c4b651cd331bfad
17260 F20101211_AABLEJ aleman_d_Page_115.QC.jpg
d716decab1942f6917b0fec48f5d9148
8b44aab864afebfd4c6fcfcb721913c59c9bf9af
1519 F20101211_AABKZE aleman_d_Page_007.txt
7b797a997e31386948de8f55076592eb
110b1aff1d82e9d21f99db9a7ce9487f43d1a71f
80639 F20101211_AABLDW aleman_d_Page_101.jpg
17ce82e698dcf2381a7c991a0b64454a
75ac1dcf45793881d51b0c3b360b3f6b5255e091
1425 F20101211_AABKYQ aleman_d_Page_128.txt
124d10e5bc0bfb48bf18f530b6023716
af6342b3e6286cd55db38749a245e41a547a6797
18798 F20101211_AABLEK aleman_d_Page_058.QC.jpg
34126694c24a7c82bc3bff0f23c0818c
02a8a11f8d6929bd368c04dbbe3673f108b62ff6
60182 F20101211_AABKZF aleman_d_Page_052.pro
de9ebdfe4419dfd1fb0bc6444ca51921
5f6c2fd4d497bfb6fea6d2fc3daa82b8751b7097
2035 F20101211_AABLDX aleman_d_Page_037.txt
6f32d0244986f85b7a9155ca7d8e7f47
8806d13f3bff499cca63b5c894cdb2a5eb5ec135
2262 F20101211_AABKYR aleman_d_Page_010.txt
d74ca728e4a8f91a2ce4b7c3216865ba
b76e7b113ba3d00aceda331790825fe4ab3e3f78
6196 F20101211_AABLEL aleman_d_Page_017thm.jpg
58ba187ad49c8261d3b6d37ea3e40fcb
978b5519dbb01218b293a4a5db2ebd7c9c585469
863 F20101211_AABKZG aleman_d_Page_064.txt
d801bd088d330fdfb212c40b9925fb31
c380199a09cf0e59bcbe4c5f9a141e79537e15c4
5909 F20101211_AABLDY aleman_d_Page_034thm.jpg
902967bbdd2381eec220d5f13493e311
5ef9c391505b1382d71fa1ed4bdac1c15b55381d
2549 F20101211_AABKYS aleman_d_Page_144.txt
7b00aba6a8736d22aff2c2a8d26fc4ee
517bebcdc3e154c9e952dbb076c0d747ff21af65
122366 F20101211_AABLFA aleman_d_Page_017.jp2
3c08f6cc27acdf94951190f14230d500
0d9dd9d3ceb2b475f2f5a9de79f2bc3b1a311e5e
17099 F20101211_AABLEM aleman_d_Page_124.pro
d9cbb91efa2e792c78b2605805fa451c
770e97194447878b7e4ac0d1a257561e43a2586b
41580 F20101211_AABKZH aleman_d_Page_071.pro
7d988859dcdbc3aebfddc6c9eba1369e
da20871042b443272d32dded21e9386ace1843f8
15595 F20101211_AABLDZ aleman_d_Page_106.pro
334496d39d5f6a01306fc467d3d8546f
32bff00295f56e827a05ec78f94c5b17ca14c8e0
50248 F20101211_AABKYT aleman_d_Page_121.pro
3d2cfdcbc075b19d08b938b2119d6608
c7145485689728f2d64361f47b79a1c91afb37d3
F20101211_AABLFB aleman_d_Page_064.tif
af566652dc4b1b798775f451aa9f78c0
e78c0a45b9fa24aaab0d4ab44791cd96edc8eb0f
F20101211_AABLEN aleman_d_Page_073.tif
3aa3ab9d5a2b1aa3574b7cc3a7368ffb
50c84e68fb52d6d0ca2ed71bec6a05abde4b2542
10529 F20101211_AABKZI aleman_d_Page_148.QC.jpg
5e27d915224be5bacdae145050281a46
351e17c12382ab3c91b369ffd53e8d6ba3794e75
5778 F20101211_AABLFC aleman_d_Page_088thm.jpg
49c2b7b9aaa06fbed15d34ff46e3c07e
3e9520b5d08baba1c9000ce87a4dabbe1407b793
F20101211_AABLEO aleman_d_Page_083.tif
bf9d42d3d0654c6028a785c7639a0dbc
ee619ad43907f71a6ed6f38dd51d314af9e0cdc9
5807 F20101211_AABKZJ aleman_d_Page_139thm.jpg
5704b7244a4178ff4fa2094ac3fff54f
64809af00032fade1ea90f7f0e43490449d928bb
86116 F20101211_AABKYU aleman_d_Page_050.jpg
ae946bb0db544ebea21cacae4ce42efc
f71979a035e00c00ea6ad1995348034b5e92521f
54762 F20101211_AABLFD aleman_d_Page_012.pro
2f1a6df75e43f382bd5c9cbd4e951905
53679d63c3f9b01d79fa3a479248cb28fb536f0b
18958 F20101211_AABLEP aleman_d_Page_129.QC.jpg
37154fb8f0228c886e93df453a2b7bc2
0afc7614257627f976eaff6c3dd0f4acb798d959
2041 F20101211_AABKZK aleman_d_Page_076.txt
75618afb784bab2f499b58adf7b9f168
fa10d9c03a3da965402759401614caa340583159
3392 F20101211_AABKYV aleman_d_Page_140thm.jpg
d5a082c722ee85e36f6afad20a7dda4d
fae1492c329fd1d2e679bb3cd888750b059177c8
90247 F20101211_AABLFE aleman_d_Page_081.jpg
f02dce023181a7c168522715dfbde924
665e95f32338abd3ebd6ec3c61e8d8033ef4bc37
5045 F20101211_AABLEQ aleman_d_Page_046thm.jpg
7e8bc78608e3e21b68cdc354ef8152d6
28c9161db68d59f3a6784666d55f5c52f852a726
21600 F20101211_AABKZL aleman_d_Page_005.QC.jpg
41cb47b3d616c2eac50844f9bd350fdf
89c6930ccc7eb5454ae4605aba0bbe8605603f6c
18252 F20101211_AABKYW aleman_d_Page_116.QC.jpg
749cbcd239598a9478931aa1b5f75a36
2bde80a32eb5760029be1470c242a73f4b99b968
6369 F20101211_AABLFF aleman_d_Page_133thm.jpg
4215e24dd70a2af440472099e9bc6c29
3faeb2723d915d2fc0b18e5e1eaf4fe9e65354a8
1723 F20101211_AABLER aleman_d_Page_077.txt
f8aba0a308ae347f7573e2be923105e3
3eb0851f6fe4fc0a830cd5f0e461765d1c19a79f
56150 F20101211_AABKZM aleman_d_Page_062.pro
c4848b37702fb8a7a8f20328eeed05ce
c386b381e54168814f5c0b5c721cace7a1fbcc15
25744 F20101211_AABKYX aleman_d_Page_031.QC.jpg
c57880bf77e7c3dc5c816eac39cc0ae9
ff7e4bd3cb70b0366cb680ef3d1d74079354ca29
83772 F20101211_AABLFG aleman_d_Page_066.jpg
54bfa4bb87f72cbfe807c092277fc905
fc80072622e2d5bf3da3ea91fa99fc4963f4839e
56000 F20101211_AABLES aleman_d_Page_092.pro
aecc311675427b23647995c7cb4ef17c
d2730ce05d28d3eceb92346380b47c798ac5d27d
93 F20101211_AABKZN aleman_d_Page_002.txt
ff5ff0bb2c87353e235fb43193abfcc5
18d67a7c9b2629d5866dc9c4ec76cda06edf57d5
6719 F20101211_AABKYY aleman_d_Page_072thm.jpg
afe73580b84b11c358444634947f6976
8a45a2468db2573c9163f088fec87c4c22b8dfa1
1051369 F20101211_AABLFH aleman_d_Page_132.jp2
c81bf693f7d558d2ebdeb0f685f41c79
862193adc4dad25413ecd2300a556885750981ea
F20101211_AABLET aleman_d_Page_140.tif
111721239c52dce6d867809b6bfc00d3
a59a268b7be65373d529c0de581062316429a1aa
21378 F20101211_AABKZO aleman_d_Page_027.QC.jpg
c8b9b313fcc363b7c27a6fd1e897d7a9
9920c4182e445e3564e442e13b9f942bc9463b10
1474 F20101211_AABKYZ aleman_d_Page_005.txt
dc542980be24563f34a5c7f4c339e115
ee542aa27ad59581bf6d219d2a4eed4a7ad0672c
17148 F20101211_AABLFI aleman_d_Page_130.QC.jpg
eea508c2f395cc24048c43c3bcf907a9
8e14f9ea4ce880f0fe5160cf6a9c6c58953e318d
F20101211_AABLEU aleman_d_Page_087thm.jpg
43cf351a7d489a442b538098419fdd13
e0630d2d8b1fa8b9adb5a0f029fe108d8860a2dc
23625 F20101211_AABKZP aleman_d_Page_023.QC.jpg
b5782766445e05d2aca001918895c5cc
7169b5f045e9d53a2dbecb13a29cddaf94212f66
80860 F20101211_AABLFJ aleman_d_Page_045.jp2
7e36465af0adcdbc180fd00231d2e37e
ebae3b70d9af6d15687f12997772957fbb30eadc
6472 F20101211_AABLEV aleman_d_Page_021thm.jpg
b308a458783c90a23eb780bae8c3fb1c
a14d713861699eed2dbd8f6d95cab239b22fc567
1051970 F20101211_AABLFK aleman_d_Page_037.jp2
d98777d6efd89463c4545d098941a581
ed7b4682b3816a6718f7973dbdafcd60f16d8dfe
32615 F20101211_AABLEW aleman_d_Page_100.pro
d3f1f70d358f2449fbb38e3e6e12716c
62773d2fe9b2496954bf0c217003a6483521995c
4199 F20101211_AABKZQ aleman_d_Page_003.pro
f63a56d8ec3dcd659c27d703ead4e202
ff176c6b26372ad5f3d913f7b6137ffa45f72f05
1643 F20101211_AABLGA aleman_d_Page_038.txt
da04d777ce4e71d74bc2421d67b814f7
cff6c74afa0bbeb0ee11122fbc2ab3fe35ac0557
1051905 F20101211_AABLFL aleman_d_Page_117.jp2
8571c12929170e598a5b6c135a5d368c
7d141c0012b2ed864c15efab924fa5d3f409a674
119064 F20101211_AABLEX aleman_d_Page_014.jp2
5a8b765dde02231f11d05cabe4f04b5d
89a2a7dbed9ea535e609129d130db9fd0162ea66
77053 F20101211_AABKZR aleman_d_Page_094.jpg
1eb6521fb480a1e15e276b2ec6e7b2d8
62977337efee8c29d18d202b6297a674675c33f7
35221 F20101211_AABLFM aleman_d_Page_058.pro
cf1548bf4b31365049d51f78130a905b
30c72e8a03c976c91fc0e5a14de4bb93038bb153
21937 F20101211_AABLEY aleman_d_Page_057.QC.jpg
a1ce51618e12af4b74095b00ec4f68c9
ba6f77cb6559307a7c94280df67718c2cb8779cc
4147 F20101211_AABKZS aleman_d_Page_008thm.jpg
c29d838e87ae694216908b352acf3d52
2bb3a05eca4681a7f71d8a67b8d7d864d1caa8ee
82627 F20101211_AABLGB aleman_d_Page_030.jp2
e43af19c27168cd930d161474cac844a
f884ef33c0da971ecad3510170bc73606a267600
27095 F20101211_AABLFN aleman_d_Page_111.QC.jpg
7f10a61619ded31a835304a724abb28a
68a630459d66515da778b0c90965784e4c3ce3da
F20101211_AABLEZ aleman_d_Page_111.tif
09dfcd954b3d7512a9e3f2e7811e1a41
391e8c6373e160424e6b92e54151a4a3ea2368f3
20579 F20101211_AABKZT aleman_d_Page_139.QC.jpg
d043abd7ad58c50b29cf61174fe8413e
0b5d60c4424bcae32b8872636c3667f8923bd256
F20101211_AABLGC aleman_d_Page_132.tif
55b3efe81ef5da709a1048f62553196a
5f596ddfe3233abac5e027bbfd168c8824e1b625
1884 F20101211_AABLFO aleman_d_Page_029.txt
8f0ca78e7f006e39099afb072716b704
1a3c2f36e4f5c80c2152d22b7c88a044d2707191
27113 F20101211_AABKZU aleman_d_Page_016.QC.jpg
0b5378b3e8f679277db595ebf7a9b22a
d183b6bfb30aee389153608bdcfa69982d82f244
4985 F20101211_AABLGD aleman_d_Page_039thm.jpg
65db86903c91437ddb90370b7b846f9e
1292dee5924b8745c3f596122a50c384e09f6b6f
F20101211_AABLFP aleman_d_Page_012.jp2
f10671929aee361c9d80d44d1e13561d
f4ed8c76aabca4f5ba258fcf32ed2f8b0575cf4f
1479 F20101211_AABLGE aleman_d_Page_035.txt
cf3b653078b90c7feff66bc98d1a069f
27a88d045d20e48fc2620021411f539ba543b919
39712 F20101211_AABLFQ aleman_d_Page_136.jpg
da2f53678640e3d6e77a0275c09f7344
a7d588f3bc4ab3ae4eb28e074e95bce3557536ed
1051924 F20101211_AABKZV aleman_d_Page_089.jp2
9963fa08f0da2adb621fcf1a801f1777
4ba155a5b7fa1d0f6b5ec07d72339a492fdd5bd5
42582 F20101211_AABLGF aleman_d_Page_127.pro
2d1a618bec8dd83fc881c3c5ae2a74cd
5243d21ecbf1b51b0d9ddfd6750b289e35400790
F20101211_AABLFR aleman_d_Page_061.tif
600f939f43c3cdfcd99b2b9a2560db9f
c87e3d774adc17cbe45417d95500f4fd898d5c10
F20101211_AABKZW aleman_d_Page_094.tif
9ce2a2534beed08d0794d658682595fa
4044b00a4255e63da5209dffde01ffe939030acc
52984 F20101211_AABLGG aleman_d_Page_021.pro
67bbfb393f9ad433e2c32cf2813de88c
02f0b38ebfa56a87bd743eabe1de2ba8927eb7d5
14481 F20101211_AABLFS aleman_d_Page_042.jpg
fba1f745bf455c6344f44d38c534cc53
9c576ebd5c0267ac0887dd55bd2ff3eb5e7597ae
24835 F20101211_AABKZX aleman_d_Page_025.QC.jpg
e99c88f65a73705db13ed800d36a47cb
ef1671b67cc621e78b55aef7fd9c49dc74f9e1b7
904 F20101211_AABLGH aleman_d_Page_095.txt
067cb05a5ff27fccaaaffbad9cccda6f
d0cfe645692dc857dac94e6c1d1698d4f22d2b6e
6120 F20101211_AABLFT aleman_d_Page_146thm.jpg
7116fa5cc386e05591384411689d12e6
9319b4f7e18c2bb7b2fb60da0d722cc60091a87a
20172 F20101211_AABKZY aleman_d_Page_060.QC.jpg
6653175eea5c4540853f52eac8064ec1
03b8332857383f46f5c29ec6bd79798c97d00adf
58240 F20101211_AABLGI aleman_d_Page_013.pro
42a7167aec18b8d00d848a23c00c744b
3ae3c83b87e7a3fb6d97a4a55f8f26a476230e76
F20101211_AABLFU aleman_d_Page_071.tif
a20530c192d4d120b158411a64012cb4
a8cd766354d40068e6afbe7c6a473fe5a7662bd8
27903 F20101211_AABKZZ aleman_d_Page_128.pro
986fdb3a04145b76dc11a9b331235878
56d4b96f756b0070fcef494aac0f1e4244d9782f
1051982 F20101211_AABLGJ aleman_d_Page_101.jp2
1126b577314603a371c37d5853d8a3aa
22ddf3cd33e0044047932666645134b3c21c18ff
65901 F20101211_AABLFV aleman_d_Page_027.jpg
ab63e0b5fb227f8080fbc92f260e9945
235ff3c74a165382f893eef774affea4cc7cec42
18811 F20101211_AABLGK aleman_d_Page_071.QC.jpg
28ee2464522c63be76ff6f3e5eb8438e
8dbb67409f5b0cd0c493086dadc56e9b7d632642
16124 F20101211_AABLFW aleman_d_Page_115.pro
5f71ea05a69e94d5c16e4028c58aa035
2322144835ba1aefceaef718e3d71de56de00902
24354 F20101211_AABLGL aleman_d_Page_014.QC.jpg
decd9062609b5df6a9b00f179b40d20e
c16801f178307f2e18bef696e2ac991f5e9b6f8d
F20101211_AABLFX aleman_d_Page_053.tif
1228e062221589596ddae08c58f7087d
aee5d0ce2c39c57562c7af9199bf9546e4722f11
56699 F20101211_AABLHA aleman_d_Page_054.pro
cd2346a2a6966762ca17331decd83249
b8d2f22189baa0438941ab32c442e09d49274c9d
50557 F20101211_AABLGM aleman_d_Page_007.jpg
8369b6265dc6552520c75536ed7a50e2
2d9edf7c40a6dbbd7333d52beaaaf0151f508bda
84092 F20101211_AABLFY aleman_d_Page_072.jpg
7a8dd2d529ce228a46ad47632005d7ff
930d68ddb176e971df0cd72f34b361525f7cb4b0
1023345 F20101211_AABLHB aleman_d_Page_074.jp2
3129c4f0dd971e74b4c6af7ab4f8b3c0
6b2b7af8666695e1abb808ddac3babbe7e1f95cc
51486 F20101211_AABLFZ aleman_d_Page_025.pro
6e626c478a38e8dfea31449885c48598
3d85c5f639aa11b21935230840f2f16aa8f67688
131025 F20101211_AABLGN aleman_d_Page_146.jp2
8e7a737f4ef7c6b35ff87c1431bd5a2a
23312fafc91969c8d87a42cc42e3a8aae6307e08
829092 F20101211_AABLHC aleman_d_Page_061.jp2
3359439eb90faff8b8423b31613d1da6
b77d28f0e84317d47ca9283ca22acd5afe5dcc9a
29943 F20101211_AABLGO aleman_d_Page_046.pro
0f003a0e46fb7fff33d4bab817998532
90a31b8e9809e5c7d3bf7efe694e552619fdf326
810 F20101211_AABLHD aleman_d_Page_107.txt
6574528d38bae6079c57e8949068ffef
e5a357e2b269ee114737330dd9c27e917a8c326d
2192 F20101211_AABLGP aleman_d_Page_072.txt
de4f4661f15e570acf9d049b94c1982c
5d3afe76a9093d2568c8a503a47bf7a3b98142b1
2016 F20101211_AABLHE aleman_d_Page_098.txt
0ba1a491655096f6882b62cf93e0f406
f89381bf5cfc41a8a468440f71ac0a227c1c8a71
5929 F20101211_AABLGQ aleman_d_Page_079thm.jpg
defa9c3e6d3196622f00db2b419c94f3
d9ebcbd80ddcfa8e8c65e61c28ac526c2b4e510a
221411 F20101211_AABLHF UFE0021522_00001.xml
4ea9a8582b3708cd695f69b3c98c6d75
38f9098ea62d38b6d4595cf9747e1402b6f178fb
F20101211_AABLGR aleman_d_Page_029.tif
46b01542a3f4dcda9169e80c5bbec544
51aab894d11bd1456d3183e2b1058fc8cf7d69d5
24079 F20101211_AABLGS aleman_d_Page_019.QC.jpg
7a0cf567721b27cb4b4a951933808561
4ddd96302d148355118ea4723cbc9d19738ac91f
45828 F20101211_AABLGT aleman_d_Page_082.pro
b315edb125f1dde6d4a37efd9d63a88e
55c4563154e1cf542831a6a7ec1b24db23a057a1
9935 F20101211_AABLHI aleman_d_Page_002.jpg
bc79eb6adf1fb5d4c987a5fcca76c2cd
2cfafb87e4ad18758c6020044bfa734649832d6a
6243 F20101211_AABLGU aleman_d_Page_076thm.jpg
d57649380da4666af8f2eb62672221ce
641035f295eaf4adf99a1142752b07341a83132b
22210 F20101211_AABLHJ aleman_d_Page_004.jpg
f73a9aac561a34d0ac5b677fb4fcc4ca
b849e1ef5cb588832122581dfcc7462df9fc1bcb
5662 F20101211_AABLGV aleman_d_Page_107thm.jpg
2ddc144302361e1f72d70ef76969dec2
b71f1ed7559aff65c456b8edc8125b0e10cfffed
78650 F20101211_AABLHK aleman_d_Page_005.jpg
f88a9701f59f1f201a7163ac8d8e4574
8111e4a391b52dc6b66e759043d63ea070cdd3a5
84776 F20101211_AABLGW aleman_d_Page_099.jp2
5e824b7907b784ee75bb8fc09c7bf9a9
f094d7d66726a0b5b8875db1ad70ae1957286dcb
59137 F20101211_AABLIA aleman_d_Page_058.jpg
bfb09edb2404144a4f6a17228c5a5984
8cd75db5a089d17966b7c040d52bda466b1af410
89976 F20101211_AABLHL aleman_d_Page_006.jpg
c2392d38867c507f2bcde0e74f0b748f
7c3bf08e22244ebb8570cdade0c0f09ad61748c2
26944 F20101211_AABLGX aleman_d_Page_063.QC.jpg
aee5bdabb825a5c71af6f1a03b6d84af
af4057d26124e04e7f31467893ba306a3b0b224b
80774 F20101211_AABLIB aleman_d_Page_059.jpg
70814fcd82379a4226dea7c260a5bf0e
c147c9ce586275ae9c59f1bd0da8273492e54070
37393 F20101211_AABLHM aleman_d_Page_011.jpg
986e54e5c44ed65f08dd6ca9ffa3c6f2
b0dd2c4265784c226b8c7af2d1f32b212e479675
26207 F20101211_AABLGY aleman_d_Page_053.QC.jpg
9fc39543c820a36b7ab6a6af3c0702d1
9c39b1500ca51e2c54b41b720565398b6cf7fb42
58640 F20101211_AABLIC aleman_d_Page_061.jpg
f9d6888f93b6472974c0a748e2ac1be4
3413c5555d471dc46fc482e89b93296a776e7bf0
76521 F20101211_AABLHN aleman_d_Page_013.jpg
62a003a7c321480c6f898c9b4abe63d6
d93f043f82d826525081a526d343ede7c4672938
27261 F20101211_AABLGZ aleman_d_Page_120.QC.jpg
aeec035ef85a56d049b2c4742abb389c
a13ac5ff62523272687cf68af59c2c7d26b57306
75026 F20101211_AABLHO aleman_d_Page_014.jpg
2b8b59452ee13248b76faee2b0df859b
700fb606d13b128f8792f8aa114836629b9da158
571336 F20101211_AABKFB aleman_d_Page_135.jp2
05fbcd1f3dcc4ec00a6b582aad8eea6b
64c7413235c5bf68f1cff0b3ba591e206fea1cf2
84566 F20101211_AABLID aleman_d_Page_073.jpg
4c41db93e9b180af7a00c3f7027c5b1f
e7466dfaa6b74e0d08f068f84c142b1bcfbdc1c1
85477 F20101211_AABLHP aleman_d_Page_016.jpg
60039d3f08c788687c54df19959a23ff
a363c8fb3ffce36cced1fc78bbc6fc7d2cbfdf0c
70637 F20101211_AABKFC aleman_d_Page_079.jpg
75224c94e94422d236fe8d38db81f15d
b7f8d087e9720589e37b207c2c28f39030d36821
63463 F20101211_AABLIE aleman_d_Page_077.jpg
80d1efa9cb4d53edf731480ee3d22bd4
fcf9fe4c5d7a1aee9d4fa66da953865cf7be6fbe
73237 F20101211_AABLHQ aleman_d_Page_019.jpg
b44153820ff9e0e92d015c8366187b33
c0af6eb79cb79c6f56714840de11d3d2c53b0069
F20101211_AABKFD aleman_d_Page_109.tif
825f840d3c71902c2116e0e700e0c4bf
baec49b67c92f01cb9e0ea8fa14e26214ebddee9
50168 F20101211_AABLIF aleman_d_Page_078.jpg
b35b5e2cde00c61097ec098b28c8e3f1
867e148b451cf840fd6fa95cf9fad2536939153e
84213 F20101211_AABLHR aleman_d_Page_021.jpg
6b2d5f9c995820c347ff76917b6808b9
896294fe7500573b2f60c3a6d6051c5d95904eac
789780 F20101211_AABKFE aleman_d_Page_028.jp2
e7bb118fecfbb91148b30ff2e16af55d
8fba32fa638122fbe1d0f3d94b93230eb28c1ad8
55996 F20101211_AABLIG aleman_d_Page_087.jpg
60e448fac4e543453667f5407bf5663c
c9d46b8bb46c49a56607129bd6316a4061c92dd7
72045 F20101211_AABLHS aleman_d_Page_024.jpg
dca74f12775d7f538a5a0706e01ff311
2da102f4071044158af2e2bdc97a79ba0456912c
86671 F20101211_AABKFF aleman_d_Page_142.jpg
89ef81074620aad6f0aa4755c62e4a6f
c4cb64358cc0b3222aa73e9ee951b1efb3dff511
74618 F20101211_AABLIH aleman_d_Page_088.jpg
5ba72d445ef9a8f4527335a6716e9ee2
0d87d848ac07eb9dc942bcebb681edb2eacf95d4
54687 F20101211_AABLHT aleman_d_Page_030.jpg
2cf268d5af699b2d561ed45c09a78a93
bda42ef7eaff3e3a6e68b8f57cc2961475ecb77e
F20101211_AABKFG aleman_d_Page_120.txt
acf3621903746fdfbe40104e05a2d161
e299932428e4377a4b738aae3aa49a955c379426
59949 F20101211_AABLII aleman_d_Page_091.jpg
3a7bfc3eda6d06d2879b34a2c0b0fb02
4d121500e279a110e88abf25633279d3d084bdd2
79490 F20101211_AABLHU aleman_d_Page_031.jpg
1eda73fa6c9a17a74a231e3de627f350
01644e5ac7de694404b5dc960014d5f65865edf7
2159 F20101211_AABKFH aleman_d_Page_089.txt
005fa7c70dd51755947b7c96e01a3731
7f731d271b880d1469d9f14ded64452cb2a4722e
72813 F20101211_AABLIJ aleman_d_Page_096.jpg
4c3f452689cacbee9f30273cc7608dcb
a407bc7c138c3d4452a9e4fbeb3bc3f647dc16ee
76844 F20101211_AABLHV aleman_d_Page_032.jpg
b1144d4cb2dc760387eca069206da0cf
255ee47584d747cf51940cf9a81914e131bc4345
1730 F20101211_AABKFI aleman_d_Page_071.txt
38ab070fc867cc50de806110baa57fe0
4c84b67e8cb5b812c633c9082d56790a3720250f
55173 F20101211_AABLIK aleman_d_Page_100.jpg
66359e4271bc31c0433bf05c8a428e4d
944cd0715e8ae34002a5667edcdd0dab6d3ddd37
70496 F20101211_AABLHW aleman_d_Page_034.jpg
3377e8dc2c635e7a809b89f9562b388a
fae7462063614bcf7790b48e1e331992278e8b30
1051925 F20101211_AABKFJ aleman_d_Page_056.jp2
6344fab2cfb9fbb9c6b0ae2403fc0e80
280b63fa949915bc2a5e3315ffe6d227b17769af
81055 F20101211_AABLIL aleman_d_Page_102.jpg
96b7a0367cf16d3467288209bf279738
a23a8b4037eb2f423cfb181f2ef1529073657e01
75492 F20101211_AABLHX aleman_d_Page_038.jpg
a83893ffede9f84501f914e0de956b3e
89126aff7a23a3acd6b0fa83351d1a1737611795
118786 F20101211_AABLJA aleman_d_Page_023.jp2
564e2a77d678a4d26acdd2b7ab7cd213
315aa1a1991872064641b859272d9858b3b2687b
5149 F20101211_AABKFK aleman_d_Page_125thm.jpg
c3b2c27f9b5bacc4a84a71ef43d1600d
aa437d1c1277c9c117e5b553aedcbfeb92acd6ba
54223 F20101211_AABLIM aleman_d_Page_104.jpg
a74a948a7e57bfb79f2526311bd7db6a
e5c1bb1c08e4e0d76ec9381d63e0ad11d22f8c5a
81787 F20101211_AABLHY aleman_d_Page_049.jpg
602178858ae29793a434c9826c0f7c6d
a15c6c61ed1546956403aefb7189350cc5ac7801
F20101211_AABLJB aleman_d_Page_026.jp2
f63bf1e6ecde8eca2e621e36caa26ffc
08b899f4b1c470815a7c367e8f9a41665b99df1a
22536 F20101211_AABKFL aleman_d_Page_039.pro
f307e32ce40cac0c3205148811c18613
af1aca6245ea4acdfb8fccd22ab39e2de6f150b1
54590 F20101211_AABLIN aleman_d_Page_112.jpg
32e78a8ec6f637fd7a2e9634218a344c
0aefc82394d8464c70dcbf9698641f5b2264dbbe
87293 F20101211_AABLHZ aleman_d_Page_054.jpg
0105783b9cc023824403bf6782ee468b
b7a104f08125235d3131022d4eb6f86a5342c9ac
1083 F20101211_AABKGA aleman_d_Page_040.txt
318b8dda314edb17fa7a3b70f424f928
42681925fee53754be6b75913596b4b5c3587967
1051965 F20101211_AABLJC aleman_d_Page_033.jp2
d80ee1a948c2cb4fa286893accbbca23
6a14e4f7ea95781c24cbbae9d4551eee4cee4a92
133555 F20101211_AABKFM aleman_d_Page_145.jp2
f4ce2577ce521c76fbdae63381119dc3
768d4772bcd0a824ac6f07370d270723b3ff2c23
53894 F20101211_AABLIO aleman_d_Page_117.jpg
5424e4bc01df8822f759e5e9a99cabec
ca245262cb0f5221858b2618f21f18db9ce4d196
5383 F20101211_AABKGB aleman_d_Page_042.pro
55f1bd143089f5f9707416d449a2185f
62dbb984e8d2c5ab43d556629a963626345ff68c
984095 F20101211_AABLJD aleman_d_Page_034.jp2
fcd1cee35804157ca058bac0e9e1fd89
07571d58ceb846721aeca53fb083c7b7d28769a1
21780 F20101211_AABKFN aleman_d_Page_074.QC.jpg
dbc5ffbdcc57956038c5d8c049798f4d
0205ef912dd006fa94b0ccc5c19d91f0e76d61b8
78914 F20101211_AABLIP aleman_d_Page_121.jpg
f16ecb63656d5eaa446821eb0b82e122
45ce9613cf044ce44e78305e602f774af1f4c15a
23904 F20101211_AABKFO aleman_d_Page_001.jpg
6faf57c6ba3a32cb94e5f914cf8858ca
157816d548ad33d72ec218dee24a716e9a601307
57133 F20101211_AABLIQ aleman_d_Page_126.jpg
045f77389ccf386088e8180cfafcfc94
8b62e30bcd3b207b270458bb868cd1378cf466fe
59282 F20101211_AABKGC aleman_d_Page_114.jpg
d178678fbcda7a2894dfbe275ca4e8c6
b2c21756cf3340308789fa86c8d952f3cf31ac08
1051949 F20101211_AABLJE aleman_d_Page_039.jp2
83600694574416a4f7bd32db036b816c
31257245ad9c81cea45373609d76408a5af9c907
54475 F20101211_AABKFP aleman_d_Page_031.pro
da15a5d84285ca76b8d86a8998454191
efb03fe5e1a2bda2a8f1465af5c1b5dab94e1c24
58699 F20101211_AABLIR aleman_d_Page_131.jpg
b6df2eed68248294e171b8c0f3d9add5
829137d193f411d1a594ca8f534000cab44c9e07
14842 F20101211_AABKGD aleman_d_Page_110.pro
3b6023a68e0bde89db2d00012dca9ffe
36b1990764139a3505c3e19ade3c0317ec02ca01
1051896 F20101211_AABLJF aleman_d_Page_040.jp2
0af1b49d99c04312f92b2f636ab5e7c5
1e0db6c55baa6fab1f9f607472d89b2d06f32a3d
1067 F20101211_AABKFQ aleman_d_Page_109.txt
f6e3478507fe22164b3a50a9bf6a0827
74ebafbf212e6fd21d8b5e2ee0b7e10d6229bf7f
75327 F20101211_AABLIS aleman_d_Page_132.jpg
a7ee8cad7118cd48a5750076b93f0a63
cce4ca7ad3ab6d180e03e2326e11143feded3f3d
31612 F20101211_AABKGE aleman_d_Page_093.pro
69af819bb7b525fc5dc5878a865f9734
a067dca969f28b957e1334ffc98d72bcdd8575e8
128155 F20101211_AABLJG aleman_d_Page_042.jp2
a5f343782b53323a64f9bc5db8f583b3
b4cee3c500c08b352135907bc26ccb65d94dedda
75282 F20101211_AABLJH aleman_d_Page_043.jp2
d3ce96033d977e66cb2cd10d1c7a2b7e
07258c76728ed2a3e43a91f8212c43c303f1e9ab
5232 F20101211_AABKFR aleman_d_Page_113thm.jpg
0f8fa65ea8ac4399412d8997babc5d4b
c907def5c1c548c5c93dfbba1cc09d8ded7107b2
67472 F20101211_AABLIT aleman_d_Page_139.jpg
5e7e2b886e1fff930bf2291b2e60caaa
fe368bdbf71e83d7f0bc1550612dc600a80d079b
51431 F20101211_AABKGF aleman_d_Page_008.jpg
5692a54c986992904e8d43f5c764bd1f
f5a43eff4e5782ac01d54c1a0b03d9a94fcbfd1e
1051985 F20101211_AABLJI aleman_d_Page_048.jp2
6d62f8e0b1cb48eeaadfca7400e4fdc7
83ea5ca1e14d38422ed5aad51064fa81edd30e3a
817088 F20101211_AABKFS aleman_d_Page_087.jp2
727bf56cc6ba2b87d5882f7119311f06
4414521576d6d95d006177d786fce5bdacd59b8a
84666 F20101211_AABLIU aleman_d_Page_144.jpg
6709cc7e7d2cadb199150ba9a94d18eb
338351455b5e4fcb5c91cee8ca9591bcd85cd443
61018 F20101211_AABKGG aleman_d_Page_047.pro
70d5e2fa528ec48cf0156f83e626f93c
75ae083d53496de7f2828d6025de2fd5024c05fa
F20101211_AABLJJ aleman_d_Page_055.jp2
db6a747941d5a8183637363ffa8ea0c3
c1b03060aba19a100e77c3bb2b54284f55d85f63
74664 F20101211_AABKFT aleman_d_Page_141.jpg
6205b2154942ed3444f464b350f3c04f
73671a313a86bd7fff9d78748afe034b9af32b9f
82810 F20101211_AABLIV aleman_d_Page_145.jpg
5df6a9365646cdaa994e47b7d4419add
5c668c63a03cfaf5b7f742d6495d83f2eec9d071
9900 F20101211_AABKGH aleman_d_Page_043.pro
4acb8fa2b4c61cd0f336c434059a5809
fab419aa84e5ae144e5eed99bf32261e210c40d0
793257 F20101211_AABLJK aleman_d_Page_058.jp2
4e7e2df985b76179de24e8926de77e9e
f4381d1ffcb64614031d36cea44f93b279a3157d
24969 F20101211_AABKFU aleman_d_Page_101.QC.jpg
eaee224cb464188d06c77f19d76c1f9b
110cc598ba58367d0c8e609a55f2ea0149a8a248
80545 F20101211_AABLIW aleman_d_Page_146.jpg
642fb0878351db1b1f5ac5fb40dd52b1
a14b62924fc8a76b9bfdbfeea26269ef59f7440d
62031 F20101211_AABKGI aleman_d_Page_143.pro
edeea6ec7fee36cf90e828b3d41dba38
c193dd6c414c27a1bf1a21ece8b17d928e2750e9
690558 F20101211_AABLKA aleman_d_Page_128.jp2
d01c08a0a2f2007648615520254c1574
33ba2ec4e50140cb0806ffd25c8dcf449b4b4642
F20101211_AABLJL aleman_d_Page_059.jp2
d1e51f9b71c2d03020180ec594dc5bca
4419058015ea190f87ba1c4ec258bfb58cab4201
F20101211_AABKFV aleman_d_Page_018.tif
aaa1230c46346f1a7072545c19748fcf
f030b3df1fe8ccc7669ee68c19338a1d23d27f09
1051978 F20101211_AABLIX aleman_d_Page_009.jp2
42192413db35c1502d47295110c8ac65
ddc1b5181aae70fcf229ca11caeae9c6bc095063
F20101211_AABKGJ aleman_d_Page_104.jp2
9ac8441de952245d28c5794c9867bdd9
ff97ca028658c90a630648265d2bd72dac027be4
837135 F20101211_AABLKB aleman_d_Page_129.jp2
c3f4593c547afdb83a1fd111436817e3
1d84f73650138a0a302f39ef601d8d7bb0a1e9bb
1051974 F20101211_AABLJM aleman_d_Page_066.jp2
62e7c58c12fd11cac9064b446114e303
727511f9ab923d0fb7218346aea619d891155296
2346 F20101211_AABKFW aleman_d_Page_065.txt
4c03969a6657333e625cb41a9ab705c3
637710220cbee1f98799713b68d952893c63c5ee
F20101211_AABLIY aleman_d_Page_016.jp2
44e7094df208cab7d73468798b090bec
dccefc1df27372d17ff6802c31a0ef1dd4921226
1051962 F20101211_AABKGK aleman_d_Page_086.jp2
bd638b941065317231f97bb5de6eb6f0
f0db2bdf0e5fab5b225d89d7dd40a6b8ea3747c8
896172 F20101211_AABLKC aleman_d_Page_131.jp2
2ccfdc34f0f6f80b7ee48c6a2ac213a2
d4f493ac305ed4d57123b9685af2a2d31304fe6d
951477 F20101211_AABLJN aleman_d_Page_068.jp2
719e2f2e3328f04eef25f81dbbfa418c
86c41370ee3604f3015bc03d129f948e5f4a3b38
70198 F20101211_AABKFX aleman_d_Page_090.jpg
27e8c4281405e36dbe12116514d8f256
f267b2648ac665309ea2182d7cec06db84e10682
120242 F20101211_AABLIZ aleman_d_Page_018.jp2
e4824586fc74b5629ed6f2d68fd2487c
da44f3eb35f698ccdff71f23501eb941662f3703
83883 F20101211_AABKHA aleman_d_Page_022.jpg
708b782eff4ff898330b570e28f67c32
bd5a66d9594ba4015e4b05fa7fa07335db19777c
F20101211_AABKGL aleman_d_Page_101.tif
1b74cfb2608abffa6af1337c3e083bc3
b8b825b27b465ce9839b5efe7c084837e73bafa3
45086 F20101211_AABLKD aleman_d_Page_148.jp2
19b3c515142229e0c4ae058ee352c9f2
a3ab2d8a3512d254f6f3a07a5b8562164acb51ec
1051952 F20101211_AABLJO aleman_d_Page_069.jp2
3bf5a6f2160bdca7794ffc131d750949
b510418f57ef8e2a28f9ef6bbd1d4be6f3deeb23
82473 F20101211_AABKFY aleman_d_Page_085.jpg
29f2733dff157baa4fe3d593914ccf05
19429f0b7a80d5c806b3ff7444732beaa2ce0b5b
50908 F20101211_AABKHB aleman_d_Page_076.pro
ffee5c62b4046c7529ebf8bffd3d82c5
b1c75b488a44bc2449f5a41d912997a77f4b6362
809164 F20101211_AABKGM aleman_d_Page_125.jp2
973d895d5232ce52110f20327cc88976
9f3acd8d8f98d1a9cf82ea4c15355e3c4625a94e
F20101211_AABLKE aleman_d_Page_001.tif
63a38138db6d39e1c37be6899b809d04
bd9dbd68da6f92ef7d7891d535e5a355060be2b6
1051938 F20101211_AABLJP aleman_d_Page_072.jp2
0b60627fc9c33e80bad304b0ed867ca9
0d6a665546b49af7fde5b147698ec08d876be5d3
F20101211_AABKFZ aleman_d_Page_081.jp2
e65cbdec87dda4eb2dd24b926b55c192
adb1bd47e90ee32b692c492693d978411ba282ce
F20101211_AABKHC aleman_d_Page_142.tif
53a0a602d338350094bbc3444a7dbf22
065d26f078ea2d83bb1bb1055e929ece5ec6b3de
67352 F20101211_AABKGN aleman_d_Page_068.jpg
9119be2117034e6f3b7325869ce0b3e7
571cbeb9ff0098297b5601f42c320b1664b06af8
F20101211_AABLJQ aleman_d_Page_080.jp2
d85b83664fcb0eb2eb4fc3ae2c785313
73c52d11a202f8f8b4e809e4e1c41220aa3d5826
F20101211_AABKGO aleman_d_Page_104.QC.jpg
1c013fe7bec021da008652232e614c92
8e10541d37111e07b7d5c6d659017aeed03f9350
F20101211_AABLKF aleman_d_Page_002.tif
293a578ff26504effb98ed4a17f47bc9
8688aa36e1241fe3fa8086920bc30acba4ae1189
F20101211_AABLJR aleman_d_Page_085.jp2
79b13b3e7ff435469d46326e3c6ba352
643b5a16b34100b4127522ff6ffb8e53a39d0c57
28442 F20101211_AABKHD aleman_d_Page_047.QC.jpg
23a17ead212763fffa81947889b6047f
d28346181f0f1224b9d5a3f869980d38ce58d731
6157 F20101211_AABKGP aleman_d_Page_014thm.jpg
264e2d540f37178053fb82e0fa5a01b1
b210fac1b7c9b8bab206ae52d35d6a94fcc05d0c
F20101211_AABLKG aleman_d_Page_006.tif
88c5f509a3c9a860fbca42ffadf51334
6e77d54a843728e9001f7bd19edb1a1f8cdb44a3
1043336 F20101211_AABLJS aleman_d_Page_097.jp2
b9b4e07cfe7c8cacc2af53a2a2abd23f
caa35a7b33949bce60327d90cd19c797c498a168
54892 F20101211_AABKHE aleman_d_Page_046.jpg
68e66ab37c306fffa5f39218b6ff0f20
e9edd262ee8e6f30283a01726a5615ffb31b67c0
747 F20101211_AABKGQ aleman_d_Page_108.txt
eb5c7499c39001439d5ac9cb2a7efb57
34f136afd4552bab629b5e6850176b423a9f8e14
F20101211_AABLKH aleman_d_Page_010.tif
6930fd0fcae3e3d0cdba4ba4b5fb5200
effed0a79b31c3b68da6e685ba1c3efbda2906c0
1007894 F20101211_AABLJT aleman_d_Page_098.jp2
397d968c6a906f044fab71b7ef44183a
898417330da7f17f5f335e3e52e07f764291dc54
18335 F20101211_AABKHF aleman_d_Page_114.QC.jpg
35c7ca5d23a376791846c0f0790f2556
a6e8a7c3c3ff85a200b91dbb8c9efcb1572d6bd7
26408 F20101211_AABKGR aleman_d_Page_066.QC.jpg
2babc36ec09cfe98ae1e6564dfd7bc8d
a02f7f36c0741e6c7ddf4257a2b5928b4727548f
F20101211_AABLKI aleman_d_Page_012.tif
c780ac15d543503c69f79b449ac4725f
e9dc4cd082461ec2ee3731d2666d912b22463532
757751 F20101211_AABLJU aleman_d_Page_100.jp2
3e77fd4d0c43a892af3f7a3644e012e1
cf6c38e18ec06ca8cfb00d09925cbbdf6e74ccfc
47769 F20101211_AABKHG aleman_d_Page_009.pro
102800c4958c0bb53334b0d85e34a2ee
66cfca978abae1aeca9b48601945176557c5002c
F20101211_AABKGS aleman_d_Page_115.jp2
a995d41beadca9cc1259b23b2cca6ca1
a7d9d78ad412b7f857bf9c2497da1d34e39020e8
F20101211_AABLKJ aleman_d_Page_016.tif
56fee58a1ff472372bd41a3cec275db4
1f4e806cc9466ddc3619042755406050bc9aeb41
1051908 F20101211_AABLJV aleman_d_Page_110.jp2
85b3090dea8dc6122590a55a7a393375
7207480f44f3910152bd135971ccb36b9110bcc3
6687 F20101211_AABKHH aleman_d_Page_089thm.jpg
3db6830af6181a47a02c85543edfc4c5
042a4e566a1b7dce7a75c993272e1b3ddd3d5ca3
11565 F20101211_AABKGT aleman_d_Page_011.QC.jpg
5d8053d1963c402c8febb9af575f74ef
6690e2ebf40de8bcaa1c0d86315581e7ba9017bd
F20101211_AABLKK aleman_d_Page_019.tif
0fddc76d5ab719e120a9837765f00178
6f1f34c0b3ebf6ea57d8a88ff1b935fccc5f3910
1051963 F20101211_AABLJW aleman_d_Page_112.jp2
0d36a1feae793787a61307d430bcddd8
7599a8d2861f4d88dfca52bb742f1de515558c2b
21394 F20101211_AABKHI aleman_d_Page_122.QC.jpg
dcaec73eafdbd25685b5ea8473357d9a
ecd1bf2157bb35cea777b6afe6f0b4077bb56d0c
2323 F20101211_AABKGU aleman_d_Page_004thm.jpg
1322863a697edce709c4c6990df7d630
fffa2a0c23a612972664def87761fc4b333b2093
F20101211_AABLLA aleman_d_Page_063.tif
7dd084e4dcfeac38f648e378f812e1c5
435014aa8f3d7d8403c021e3aa48cfd2a7d27f55
F20101211_AABLKL aleman_d_Page_020.tif
dab849b11a7ca47e68677bee029d04a6
3be935f56094e93ebf28413533eb99fd6ec03ce7
1051983 F20101211_AABLJX aleman_d_Page_114.jp2
d51a1fc1e060adba35c829ddb0a658b0
b243329fb9fc60cc70b2d35748978b67ff3ce743
1606 F20101211_AABKHJ aleman_d_Page_084.txt
0bd4d07f520335e8aba8f564dc6e8f22
3e81abca9b39ae0534feb6f599ee653739e50f62
1531 F20101211_AABKGV aleman_d_Page_136.txt
81e8659e43931075ecdbd629e9c80ecf
5ffc18a576fe48840b9720cb1c880816c91a0063
F20101211_AABLLB aleman_d_Page_066.tif
d9eb728eedcd80a332d6fd9726b0ee83
f1926a7a91e9260fda7b292882dd9cc288c2ba4c
F20101211_AABLKM aleman_d_Page_023.tif
faa3e2a1ce817fe4f0a4341616895599
3fdd6e4206a933a66053145183f77ac237b9c05e
1051930 F20101211_AABLJY aleman_d_Page_121.jp2
09450f86c88e00585b478529d7a37d57
6f798eed99a8cdbe2730744a6601d4387c9936bd
4360 F20101211_AABKHK aleman_d_Page_093thm.jpg
ff60020c535d6e5072d351649dafb2db
b54b690d769ab6af8f1d00633fd5f87b52278679
481 F20101211_AABKGW aleman_d_Page_001.txt
4b287eda1fb1b9e045cf6a9e2126bdc2
ab8e273e885a9fe8fca123c396b0a5e120004e11
F20101211_AABLLC aleman_d_Page_076.tif
7232e0a9b4907a7cd2a50201d1548c4b
c513d0ed9d4c9d8933c3065170f5c6e255a941b0
F20101211_AABLKN aleman_d_Page_027.tif
f90c73f5d097a5dc26afa6096bf68231
df29fbc966a4604d67d10ba313918d5f116bc69f
1045550 F20101211_AABLJZ aleman_d_Page_122.jp2
d6ce1e2d184833f00fb887811b302d66
9d73fe5296f7174ec70bb50563fb258f3f4a76aa
72882 F20101211_AABKHL aleman_d_Page_023.jpg
4ea0e512fdd2335552e0860256a35525
d9578ec098782e8e2f4d0d08f1a6c194ceeeebf9
11678 F20101211_AABKGX aleman_d_Page_044.QC.jpg
db430a24a96c3c21a63bd5a5209a7256
c87f241664e54860167ec1967efc9b4c10a67724
83134 F20101211_AABKIA aleman_d_Page_086.jpg
c2242d13efb26b627d82bfce403e2084
d0771477c3651d624294445d80bf09fab8dd72b4
F20101211_AABLLD aleman_d_Page_078.tif
ac12eb046d12b216aab8436de83bfd41
b83920141900ce6e68c45304e00d0ad0264e4b0f
F20101211_AABLKO aleman_d_Page_028.tif
faf61bb605d746105d61e88c5c817e5f
d62b6fe9b01e484b2b9c4ae3ff8a562553775476
F20101211_AABKHM aleman_d_Page_036.jp2
e4f8535d5908ee05d83645667ddfc550
2b71c7365adb02bba606b77fc4f12815e8520f2e
79297 F20101211_AABKGY aleman_d_Page_069.jpg
d759816235a1af37f0e1e9661d4d475f
11c12eea39aa07d9e3db1e8d9d1ba9225efbbb2d
F20101211_AABKIB aleman_d_Page_051.jp2
640c13981acd9c619a7c19ac059c1816
c33142b826e7e2bd9fe07ca62afdf012d40cf6e5
F20101211_AABLLE aleman_d_Page_079.tif
e3e936a614dbf990976e404477ca957e
61e1be8927f2dc6089f93f119018cbfee5416adf
F20101211_AABLKP aleman_d_Page_033.tif
3db8427c2aa5cc53875cc11d9c8b9ab6
0a911dd04e0facae52e23e1050fac826940c5b67
39057 F20101211_AABKHN aleman_d_Page_140.jpg
8dc90d52b47ae8c1c99343c3eda6438b
e76a70c199104d00bf495e3768bd69cda415ed1a
F20101211_AABKGZ aleman_d_Page_136.tif
98dc78185500e4615077384ad80e774e
3e5880ded35a89650f7a889ee8138be4f95f4d79
F20101211_AABKIC aleman_d_Page_046.tif
9fde8e308cd0d521d974fd7498c2faf7
3e600f5e8a79b10057894edc9c49080039a86c87
F20101211_AABLLF aleman_d_Page_080.tif
3420a91ad599d8d12fdaa1eb798cffdf
ef53a4a652b7f3a87d7005e2439c6c35635350df
F20101211_AABLKQ aleman_d_Page_035.tif
8b09cda2ac1cbb58feb7a22bc9c69a38
edd575ed3fbd21fd33bc08833d8d62f9619def3b
54169 F20101211_AABKHO aleman_d_Page_096.pro
f984ad63e865d04ad344e1baae4c9d85
4c01edc93ecda8c338c172f3bcf6e04521014c18
3278 F20101211_AABKID aleman_d_Page_006.txt
cfdb0f74c7b61509b410ef4f77054cfc
617247967f6509d0a8ea0685256c838333ffa125
F20101211_AABLKR aleman_d_Page_037.tif
894fddf15fd364caf0c80d0ac3c3debb
434af66286a831f59c2010e6784fe3337debc236
F20101211_AABKHP aleman_d_Page_124.tif
ab9dc00b847843ebdf600158e5b8aad1
32f175ce0f78bc3b9ce2714f642dc04ded8e1ae1
F20101211_AABLLG aleman_d_Page_084.tif
f04170a924a4c538e241f28aba13d7c9
6ebc7e59db066f204606828b4bbdd378665f03ad
F20101211_AABLKS aleman_d_Page_038.tif
59c53635043a79894728af8c2f721684
548938491ff072783adbf86ed0cb74a331c779a5
23606 F20101211_AABKHQ aleman_d_Page_146.QC.jpg
a2843159676f67bb7bbd52918d3136ca
c3e6c5632b2f539f8f100efc2a2a0773eeeadcd2
19146 F20101211_AABKIE aleman_d_Page_077.QC.jpg
9fa47d35230cccb8db7b55bd136a3093
f9967012e6b07beff5d74b1d4734de9882a18996
F20101211_AABLLH aleman_d_Page_089.tif
b7849ba2c093d9c8bc550f9e5fc69b12
0c65cb8c953d475d7c34eb839521e350240f179d
F20101211_AABLKT aleman_d_Page_043.tif
fbb4d97f03ebfaf8da1acb488727fdd7
87d0f195599f3350a53a72b3a57cff0a8f47c0b3
62747 F20101211_AABKHR aleman_d_Page_145.pro
22614ea92698a754db363b838d10f6c5
902893876a5bf2520fba7843d4fe05bd228741aa
6783 F20101211_AABKIF aleman_d_Page_022thm.jpg
0965cb18fa066f2576c6ee7cf8055d6e
af62592ad07329fcba0fee7ae523f65cbd6be26a
8423998 F20101211_AABLLI aleman_d_Page_093.tif
e7efa6b562bc09cb450d1875f61f51b4
7dee290275bd6e5990a090de0107dc984352f336
F20101211_AABLKU aleman_d_Page_047.tif
b70366393c05a44fd9001830b58b13a0
4e875add42f430beff71ff007c467060ec1b29ac
F20101211_AABKHS aleman_d_Page_102.jp2
8a14751db2e21880c660ec165fd3bc26
e74f6898fd4198ffd90a03801e95aa372c91ac53
F20101211_AABKIG aleman_d_Page_013.tif
1b14b3206bb3a7648157774606e3f01b
2d9213edd4080f3d101f7318e638678682fc7a95
F20101211_AABLLJ aleman_d_Page_095.tif
59580a672a266d91fb1c3bef71171c47
d9f3a6a1b1a4b07584b9e426cd322f2ed14b9463
F20101211_AABLKV aleman_d_Page_048.tif
2d3ad87a3a0c11c514ad7fe660c2989f
c173825d16494d265997f38b2022d4fa91f0c7e9
15983 F20101211_AABKHT aleman_d_Page_093.QC.jpg
9870c40b9d025ec274946e1e65e17820
077ab597cc67dfae2d396fc9c868faa7a018d303
14319 F20101211_AABKIH aleman_d_Page_007.QC.jpg
61291a0337e6221c96aa1133b6ba280f
3107cb22927ad70262a236f5b1f1e66b22676916
F20101211_AABLLK aleman_d_Page_099.tif
32921023a77a6dce9cbe36616e8aade5
bbe2d420247a0cf333079b35031206d5e270503c
F20101211_AABLKW aleman_d_Page_051.tif
ad4c371c1ed17e03869c6425a3e4de4d
bc95d3c935b5a586e561faaa3bae88099c4075e1
1051928 F20101211_AABKHU aleman_d_Page_010.jp2
7c0de5ac51a2475fd0f827ea13416dbd
bb9500f0ac03ae519abcd4767221c16f418b9ea3
65575 F20101211_AABKII aleman_d_Page_142.pro
2183f87ee853fdd2f4a40e20afcec7e5
03b248b275ada5e0dff11cc4f93228aaeef19522
F20101211_AABLMA aleman_d_Page_143.tif
c44eb0a09514efef21b717ad0d4d116e
8c07f52709d52d335b866f2ad26f8b2df1bfcc52
F20101211_AABLLL aleman_d_Page_103.tif
10a6949bf2fe74073fe7dcad1e539dad
902a75450ac319d81cce1499f4e8d5c7c3cb6c4e
F20101211_AABLKX aleman_d_Page_054.tif
48b87738385797297e58b39d507d4513
f21cf9ff26fd1c066eb5960b4aad516a207b9d8a
F20101211_AABKHV aleman_d_Page_106.jp2
08a81759fb44b22183e2b1b07f27f3f3
3f95b8e4c890acd55bd36376e581da3838e14c5d
2167 F20101211_AABKIJ aleman_d_Page_083.txt
b2feebdf0464ab07821d6dbd4ec2f2f0
8739c04cd475db259f7d6c284d44ea9d1b33b6bd
31519 F20101211_AABLMB aleman_d_Page_007.pro
0e1beebaf6a17df934dfa64cca818d37
4d146c8bdf131eaccde7bc24a3932178d9e080f9
F20101211_AABLLM aleman_d_Page_105.tif
0f63532fcad0108072b5d4083a1dd70a
3fa78e03072f8619a49016faefdda8a96b6041cf
F20101211_AABLKY aleman_d_Page_055.tif
913895e25d27778bf3cf917f6bf7e254
1855104d45a1302b255709676f3c7a6ad078644a
1720 F20101211_AABKHW aleman_d_Page_068.txt
82a2f8c0ea241fb438c20b38fb562e51
c31605c8c90b7f53b616f8df41670e08c1141a07
54962 F20101211_AABKIK aleman_d_Page_055.pro
e28ab8b368fa64f8d1849526824a678e
0f6e990171e988d5cc7c02d7e563233bcda941ab
54683 F20101211_AABLMC aleman_d_Page_015.pro
cf86d71c5b99c4c5018569e429ab6de1
347c35b2fdc721be6624efb352231830d3e13d69
F20101211_AABLLN aleman_d_Page_107.tif
5ddd4cbeb983e8e872132f54a2b25acf
f0b9f1475ea78fef5db4881545818472293b41a2
F20101211_AABLKZ aleman_d_Page_057.tif
e1eef36cb2ca751f7c5af682683786ae
5ffa2aea4f441a4711367ddd89d503002d7a79e4
59775 F20101211_AABKHX aleman_d_Page_124.jpg
5c2c47faae2cf7a0643a6858e339fc98
519cf6c58d16258e20502e4c882b2c23a30e226f
33594 F20101211_AABKJA aleman_d_Page_125.pro
4e0f747f263e5c941518dcb84ffa371d
b934301eb09f36aab13facdfb05682188214926a
594 F20101211_AABKIL aleman_d_Page_067.txt
66c3baa225c31e95853120e4ccf89482
15d61f5c3b740102bddd6ef580f910801e59d302
59563 F20101211_AABLMD aleman_d_Page_017.pro
847688e8ca9477252272d666a67f4c68
877341e826c8f96a6e66f44a8540dfabc86a0490
F20101211_AABLLO aleman_d_Page_108.tif
8f590ee605d8223474df7b734ac810d6
df5e04e1aa9787da72778766d007981f3957135d
822 F20101211_AABKHY aleman_d_Page_116.txt
93c20cb22f805fe59a67ca31e8f6d94e
1fbfafd54431f80be355fa0e7f76b34424c618c1
1051968 F20101211_AABKJB aleman_d_Page_006.jp2
a4f8ba1bedd983e29026df460d15cd1b
41af03c97ca8d408b7cc7cf4662fd6f9c76b40f0
4967 F20101211_AABKIM aleman_d_Page_035thm.jpg
450b549055d131aed2196cc80bf4ced1
406ca5f09d1f509c1dddd85a7a55ee0656a674a3
19956 F20101211_AABLME aleman_d_Page_020.pro
a08febd17fdb71f544f5315d72054e8a
1f31dc26100c37ad531bfc4b0d909766bfb67c87
F20101211_AABLLP aleman_d_Page_112.tif
57b33e018986ab4dccc44fdaf85e72ca
90258295ce25c6feece6ab4af6a8ecaa6e1d2d11
39302 F20101211_AABKHZ aleman_d_Page_044.jpg
a3382c7d2a9c701425fe1de686457909
689ba2a59a0f31b673bd463020c008b7381dcd57
F20101211_AABKJC aleman_d_Page_085.tif
05f13b44b13406c6dc05462ce95866b1
6e2c232c1a35695babde6016f62879f8ba6ac3c7
76975 F20101211_AABKIN aleman_d_Page_009.jpg
21c9d74110e78d44c7fe1793bc20dc41
a18d45411e1d1becd4d7ec0a4487859c8a06575a
51738 F20101211_AABLMF aleman_d_Page_026.pro
8c0ccda296d65db52347acc8683f736e
91ed73032a2a7bdde6398c80117d214c3b807885
F20101211_AABLLQ aleman_d_Page_113.tif
e7f3fef445723f76aee48d70e823b0f2
9e7edaf5717929ef2d3f0b1924f400c0a03614ec
1029693 F20101211_AABKJD aleman_d_Page_082.jp2
5cfbc80c4636510535bdd46d25a059ae
7accdab1df23ff0d7aafc7d7d48e7e2026812c20
77432 F20101211_AABKIO aleman_d_Page_025.jpg
8df896c5765e4fb7d2f53f12de37ac3d
4b81e6199f32e93e6c79542d35dd02090474c77b
43021 F20101211_AABLMG aleman_d_Page_027.pro
e5a8f6fef3ddbafcd326ca07f899705e
7c3d5fd6ee383ff3fb50e0f4d08f0c562605330b
F20101211_AABLLR aleman_d_Page_115.tif
3ae2cfdd12e6c465c09469853a93c4b0
e38311bc79f3f5d5937aa5bf970c1f552d8c3900
87542 F20101211_AABKJE aleman_d_Page_119.jp2
cba151aaa52983dbccad1ad2ffd009f8
81fb6290134643c4f6c9c5017c9e0e252642c39c
1939 F20101211_AABKIP aleman_d_Page_027.txt
ec8053c711ee49d361bdeb18957a9a8c
f18b900872a22955d258142d01d805c51d086ddc
F20101211_AABLLS aleman_d_Page_120.tif
6a5f04bd44178cdc68e2deada854c062
10c7de4c6ca837c8abdadd2b001278cea91d0d94
F20101211_AABKIQ aleman_d_Page_139.jp2
4b2229b5edaeccfddafdfd3fed1abbbc
9f10bb56dff41a357cf72c324cb5cc9134582862
36358 F20101211_AABLMH aleman_d_Page_028.pro
0dd2493395abbf59a30e329ed96d1ff3
d05f58dc3341162a7d6896c655a88af6631db408
F20101211_AABLLT aleman_d_Page_121.tif
26514097a037c721fec79ee4f5f8f8c9
e0350b21d780b7cbe2576ff6548de2439aa62d1c
32844 F20101211_AABKJF aleman_d_Page_020.jpg
f71d4d18579ba5fe8caf361e13bed584
c61d8398e8cbe31a802dde80d4f6434c85a26394
F20101211_AABKIR aleman_d_Page_030.tif
e88a3247a03614ac80b1dd40fc745efb
f0f3c6ec3cea6dbf0c28ae0156287f32227e6ecf
44472 F20101211_AABLMI aleman_d_Page_029.pro
0b7a686b6bb8334d1fe1ad2bc71b8a47
f267e3dfc7c244fb9a53d13df14c6402c72af40b
F20101211_AABLLU aleman_d_Page_122.tif
b1ec5b0891ef5484457ffbbc69afe7d2
87a8f7aebcd1c277b3ccf0067037eb76be89567e
8316 F20101211_AABKJG aleman_d_Page_001.pro
a4d35db1efda2a5ffb4ecb16e9f852d8
43fd89b955d1413cc9823cc404425924cac58e6d
F20101211_AABKIS aleman_d_Page_104.tif
ad62e5ef028ea70281a25d32ed0fd820
0162a0308b15400d1df3a4358b6c55e47b42e3ae
44738 F20101211_AABLMJ aleman_d_Page_034.pro
33cc5a356953ae71123cd81e40deac64
e831adf05adaf8acb951ff854b124bd48c257c1e
F20101211_AABLLV aleman_d_Page_123.tif
aee25df26a786f03a09a6df90d6dcfc0
7d3e82e38e7a8eff9c4a9f36ef8682c31e850ab0
1850 F20101211_AABKJH aleman_d_Page_127.txt
9c1a514d72e198a2c3f9b603cc057f7e
e42c4f50c1125f9f457dd974f280726de52e7ba7
2054 F20101211_AABKIT aleman_d_Page_103.txt
9b5f99eee7de71ed60eacb8854664251
71885eee159a4babbec40a63dd98539ec7637c93
30775 F20101211_AABLMK aleman_d_Page_035.pro
ade279de6854045a67a3a05cb2691ce2
37a90619aba10de7ce0058aa22f01c31559bfa07
F20101211_AABLLW aleman_d_Page_126.tif
1873db4ce746deac95b088c5df241ccf
39067ce86930233ee8dc03d721bf6da72f9b7a96
25958 F20101211_AABKJI aleman_d_Page_022.QC.jpg
e4b59dfb83766f9a7a497a1a48fe68f5
49ca9e100dc291732cca9a2c3e6e3a0df4d7fc94
58092 F20101211_AABKIU aleman_d_Page_094.pro
97ebd5606c0bae4712fb9f37f8f50d0e
1b500cee59c3afcf55c88715b7bfdf59b5f8ae9f
45856 F20101211_AABLNA aleman_d_Page_088.pro
2de2a27a427695f9d0d37fcb5bca6c86
bd551dfd5d15f19aca12f3f143a6ad0d37f9f047
54042 F20101211_AABLML aleman_d_Page_036.pro
afa9f0cb84a8d5ef1bf652cfa4c2c421
2a9c1bca72657302c8d4c27f0b7ad0822bcf6927
F20101211_AABLLX aleman_d_Page_128.tif
d00f61bb8998e4c5e98e827d8503377b
4eddf2e2db45489a29f28f8123b783eeadb7029a
23988 F20101211_AABKJJ aleman_d_Page_102.QC.jpg
32ececdd26a5d8da9cd0ec36bb1ce477
79abe830076933af02b54f8e14033d39e2e49f56
59960 F20101211_AABKIV aleman_d_Page_116.jpg
5e7e8320f0340438b329d115026a1420
b306ff0c04036cdcd90b115408f8f8a740e3fe62
54627 F20101211_AABLNB aleman_d_Page_089.pro
46cdf5192d211619da9d7c3bfc29c107
1eb3aebc753ddbe32e4d28ea250829e1c2887ec7
44912 F20101211_AABLMM aleman_d_Page_037.pro
d59e02205b284ab2d1cb33e5f3c7b28d
2ecf2a500ae1b9581b513a3e1cacf5f78c9ac9d5
F20101211_AABLLY aleman_d_Page_129.tif
ee25011cd339d600db9cad2d95e81a03
8f98d4e9cce2e0d9395c6b9e4e70d5a40b9eb997
F20101211_AABKJK aleman_d_Page_133.tif
31dd44ed9b5bd8acd5a47d5189c488c2
72d3905576f943cf45d0cefec129fd5f965f1418
1759 F20101211_AABKIW aleman_d_Page_131.txt
c7eda4f007914b78b096e398f8c1df66
f6eec7175ef631d51fa785b0baab0faa0be92359
22779 F20101211_AABLNC aleman_d_Page_090.pro
ac92329ce251cbbd8e15c0cf589d1fcd
d6506e9adc1f60a5a09274b350147d5204cce35a
58660 F20101211_AABLMN aleman_d_Page_048.pro
3e60d0a04f5c17993304fab2037a5ceb
8127ca27ccc8b3086411119666c3036ddddd2eca
F20101211_AABLLZ aleman_d_Page_141.tif
8221f3ac6feac11a76628afed00d0c01
05936ce897cc47e798653c6eabfe96e1cedc5db6
1363 F20101211_AABKKA aleman_d_Page_130.txt
1693077a706e81d81df052f0d366008d
9b326f2d6337484a2dfd36b592b4dd20b0c0c9e6
85652 F20101211_AABKJL aleman_d_Page_063.jpg
869de9be2db856186d6a858a55ea2bd3
49554161ae22847158111579d799b99f9961b301
86748 F20101211_AABKIX aleman_d_Page_123.jpg
139afa59ccc7c6e3b416f4c9048b664e
8c8d49af30c0d093aabbb1f6a73fb4a7f5768408
47662 F20101211_AABLND aleman_d_Page_097.pro
a0c84c2065ed1249ca3ee33272742d0c
e13de02f57468c3e9f6411c1005d4cb484c71e5d
53958 F20101211_AABLMO aleman_d_Page_049.pro
9089a933ca17563f39e53a2f03d56151
13565f10e49b2ece08feb9382e227f5384f36f68
1051909 F20101211_AABKKB aleman_d_Page_134.jp2
2dc7c4216f50fb6b2bf91cb9596f7b04
5ffded864e2ebd87336cd3ae235119057d40da40
56228 F20101211_AABKJM aleman_d_Page_105.jpg
c21eebdf62ddbd73875ebc450876c3f5
795a3bbe234ed5b52e610135d23ca6aedb8eb8ce
22965 F20101211_AABKIY aleman_d_Page_096.QC.jpg
9069a34caaef933b2c53b19aee5b5696
d5a2b2225a906adaa476b7c0aa5b78acaa3b0740
47664 F20101211_AABLNE aleman_d_Page_098.pro
480723060ab67ea6686647934274adf1
26c407e0d2e3e1582cb95f711321d6bcdc0d7247
22178 F20101211_AABLMP aleman_d_Page_051.pro
ec2417b18570a3f5fcd138d79df6d8ed
8cccb23196dddfac1a0fbd683f0a6867f8a54eac
F20101211_AABKKC aleman_d_Page_040.tif
8f2c454e48d677335e8c853cf845238f
ec48b1353372a654f5a3994a2087f6ddd33840f1
23846 F20101211_AABKJN aleman_d_Page_015.QC.jpg
2296a3de92518883bacdcd923840cf5a
852002f267a6b917def9aa3aefd9ab28717525e5
F20101211_AABKIZ aleman_d_Page_113.jp2
5c72c27ffac62c67e3bec075e80b5877
432ce5633eaf0096feeefefb29696f4c5990f213
49452 F20101211_AABLNF aleman_d_Page_102.pro
783c7c3ba9f5e92b0cce88bbf20ac25e
eb3d6a0d46787b4a97f4b4a728bc947952524dad
46113 F20101211_AABLMQ aleman_d_Page_056.pro
0edbbb76c9e017877ab8dbf852d21a1b
b4c54a7f248966afecb757593c3f2abb13b3d6a7
34851 F20101211_AABKKD aleman_d_Page_135.jpg
aac3024103f5282b63f91e1e8cbfc4d9
f71831925a2c588d3dc5780baa98bb4f27d9215d
4126901 F20101211_AABKJO aleman_d.pdf
d881bdaf6f7049265ca0cf42f9fce097
a2fdedc554ea0934d81fae5d7a7b5d1d597b64a6
15419 F20101211_AABLNG aleman_d_Page_107.pro
14480c22a44d715ad48834ff8cfc4471
5c5a9475a074f7b951b5133132d3247efeda7e3e
57713 F20101211_AABLMR aleman_d_Page_063.pro
105880b5ab5fee4daf2669687a5e6d5e
790c39ec07281b5e7ccf72153a1737481907aa3e
25154 F20101211_AABKKE aleman_d_Page_136.pro
a2d2724cbb783b79dc97d2e117cdfb9b
6229d12cea9352e31b4768a0b88c39096ecd8060
6188 F20101211_AABKJP aleman_d_Page_018thm.jpg
b40f33a137d6017660c6aa0a58a5a1f4
0d6e6f3409ea7ae766a9781289da980ca22cbe75
13775 F20101211_AABLNH aleman_d_Page_108.pro
005c25dc5e13d8aab73eb4ed2a947bd4
fc24861cd47a5f4201f0170ed419a12390e52175
59598 F20101211_AABLMS aleman_d_Page_065.pro
c58b4598038575388718317c8b1e4723
bfca6d075673c6e047b6e8b3970fa599ce7fd69c
17129 F20101211_AABKKF aleman_d_Page_118.QC.jpg
0ff3898f59ea237db856bc1df9c403bb
0cc7f9fc012dcb51505c9e7faf058fb4a5c870c7
F20101211_AABKJQ aleman_d_Page_011.tif
90aa8d55b2177089642e7f250b72b473
9857823d6773673659d37ef8d2ffc5957c971791
54581 F20101211_AABLMT aleman_d_Page_066.pro
69598ce109e3b53b98a0cd3f17c0958c
c056c6e92024c4789c2b7c137962e0effcdabeea
55409 F20101211_AABKJR aleman_d_Page_141.pro
42cdcae29bcd4758774d0f3ed05afb06
4b49afa4e88b50481dc98595c8f5cef4fc00fc0e
17238 F20101211_AABLNI aleman_d_Page_109.pro
3fcbee55afc4a30c6f82245a2ecd8471
cd6a00e80dca1559ea4def29f883a81f213e7f85
10879 F20101211_AABLMU aleman_d_Page_067.pro
870cbbda9ee4bdee37e0c332d8ce7062
de52254de023009369c4e40a1ff798fc175abdf7
1038953 F20101211_AABKKG aleman_d_Page_103.jp2
95b6f2de30b662adc49e1f0b27e3c48d
da1a70b717110ae2e3625fb1f4030e4f4716e714
F20101211_AABKJS aleman_d_Page_031.tif
7c90307959666b153fee3274de7e4a63
ec9d8177ce1904ea96d3228edc4b3d96517841be
59777 F20101211_AABLNJ aleman_d_Page_111.pro
06e0aa5e7ce0690b2df57e4bae159f90
d6bd5830590e68b48bb5f8fd182770efefa130e3
52490 F20101211_AABLMV aleman_d_Page_069.pro
20852185391f1e293ad1e90bdfe706af
ef43285b390b6e47813a1bff953be6984b67cfe9
F20101211_AABKKH aleman_d_Page_100.tif
e6767874a4f63329acd949846217d776
9751bafbefc7bec8e01fef66fec522362e604174
11841 F20101211_AABKJT aleman_d_Page_003.jp2
1448eaf0105227de564744e074c313d5
1d00016a4623c63e68c07b6535732dd7ac80cc0a
17958 F20101211_AABLNK aleman_d_Page_117.pro
3f30b0e30ac60a8556c191e0b09fe0f5
71e4018053d8d442990880437232ec981772a9c9
56729 F20101211_AABLMW aleman_d_Page_070.pro
b5d4616d06a795b7013462e4b0b6f42e
e1e89d66ec4be1a4ebfdf6d73f9e52aba405eb50
F20101211_AABKKI aleman_d_Page_005.tif
09572e83d10137c98d2e647fcb9ae5c9
4a3b85d253bf194563e88904558282e93f905459
F20101211_AABKJU aleman_d_Page_073.QC.jpg
8e101bca934028645233615bc2a19bda
5720d5ad2bbe4e470a43e2e012d86376892ace22
2317 F20101211_AABLOA aleman_d_Page_013.txt
0ff2a3cffdb683f3a9ad913d68d942e9
df9eb4da24ed3cc11d0a8e4b05bce6a939977385
14006 F20101211_AABLNL aleman_d_Page_118.pro
d8bfa5186f57aceebfdbe3a070f580e4
60bb1241787ca2614e6f0acd58c0131701fd4b69
55608 F20101211_AABLMX aleman_d_Page_072.pro
8b61e997117d673e439e4e6aaae8f908
b52efd3a5c247a6acd83d43a16901ea7fd57a119
6852 F20101211_AABKKJ aleman_d_Page_044.pro
8545659ce5a401bd386f7b0ce94d6e07
dece7fbbada8c289975de46755a7303bd95d3535
F20101211_AABKJV aleman_d_Page_138.jp2
a68adcc94d23df12b2ae6342e0900cad
1984797f335198b8ebed8fffbf0c57abe08c9c6a
2161 F20101211_AABLOB aleman_d_Page_015.txt
9b5f49ce502645b8a1ad6c4f7fe5fc69
61d00797ef7af37b3c3f84c7349df6d86101103e
54934 F20101211_AABLNM aleman_d_Page_120.pro
e91c04c43955f27d28b0581e9d2d4a37
9b00f2b00ebe3d36c5b1e19f7a44634622257e09
27201 F20101211_AABLMY aleman_d_Page_078.pro
7ca51a0510ceb43125f999e5993bfa0e
659885a211605278467fde895961ad4bdefbc147
15209 F20101211_AABKKK aleman_d_Page_105.pro
28e59b2cdf2b652bcb3fc4f20beced72
a6a3069d66834ddf369c6b504543000f537ba5a2
1051947 F20101211_AABKJW aleman_d_Page_038.jp2
61f96a46b5725cd411c4b6d22d3865c9
e1b0e14f089ad27bc975e4a7a6f8e60e0011df52
2352 F20101211_AABLOC aleman_d_Page_017.txt
4192bcf1abc619432b1b0cd46e0c00bc
6228d82c6bf4ad0feddb1174955cbaa9b48aa9dd
41085 F20101211_AABLNN aleman_d_Page_122.pro
c04533640f60803ba250d519cc94a2cc
3c24be8f05e0c1c521df2183cbd0ad5db3e08912
35214 F20101211_AABLMZ aleman_d_Page_084.pro
c6cd1fd09de2858e04be1e7683015694
6e65c57712a42afad30196f870c3583ade8ed5e4
119124 F20101211_AABKKL aleman_d_Page_141.jp2
ff70658a8b8b5fc896334061e2b0cfe8
fd768530524052a733cb30356f04873fb1387de7
16829 F20101211_AABKJX aleman_d_Page_045.QC.jpg
aefb405962eb6faf166c88cfcc975dde
59e7fc9d7d4f972f8d23038a04b5f997ed22803e
1935 F20101211_AABKLA aleman_d_Page_024.txt
7c609a1fb5ca9ed5341dab409f49c7c2
53f52f1430004c0ac2cb444a3be3a80254a7829a
F20101211_AABLOD aleman_d_Page_018.txt
593b6a9f384614736ecc4c4d1d41171f
bbcd8eaa8443ef1816dd5719f022c3793da170a6
57696 F20101211_AABLNO aleman_d_Page_123.pro
d9d9d148bfca036ba0a0161b50804087
277cb7f66efb8ef7b53db9fb44aeddbfa2dcb01a
44238 F20101211_AABKKM aleman_d_Page_024.pro
e16162fe628e90e3983e401b68f96ac0
88426378c3119cda6a8cc887f0daff6878e41636
F20101211_AABKJY aleman_d_Page_109.jp2
17e08502c39a5e6274517605aefa9377
a25529c3d45a1e252f34393bca76ceed8508aa77
F20101211_AABKLB aleman_d_Page_123.jp2
b43ed814912dafc685ddda8ee5096bca
97ca7852835b03523eda02d419a664356c70cacd
F20101211_AABLOE aleman_d_Page_022.txt
d9ddd02114976c389de8812d03341162
df53fee38fd202224ceee65e043add1f06814015
F20101211_AABLNP aleman_d_Page_129.pro
e150d39d3b7c65695bf3d6fffc77940d
4bbe7779cac7f0aca91b06693d6b3ec6c271ba54
6879 F20101211_AABKKN aleman_d_Page_048thm.jpg
120d0417d1d246aa17de8bcd3da6ce95
d0169c4b7c93fab4f202ffefffe52342a7f76b9f
6940 F20101211_AABKJZ aleman_d_Page_063thm.jpg
f2035532a34c086e6df5137fcf0b99d3
da2321d1ab33654f10b61b1d28f3e0efe2902366
1008995 F20101211_AABKLC aleman_d_Page_029.jp2
97e52c654e382ccf34a73772df16b277
18605f562c0a6c06abbbd76381f58cfcbaba3c6a
F20101211_AABLOF aleman_d_Page_025.txt
cc898b918dace1ce5553cf04063c6179
29f3801fd7b3c9f59eb456c6d112c327bfb39f21
30190 F20101211_AABLNQ aleman_d_Page_130.pro
9ad5937a11ea9cc060ca98d8e36aba72
81b7ed1edaee0b37fb2c789138b9372fe943a72d
57758 F20101211_AABKKO aleman_d_Page_106.jpg
857ea7b2a8aae57d5c33edadaac23983
5e02160e404fd608525ae0f820d8fb11b4244136
F20101211_AABKLD aleman_d_Page_050.tif
780e497a2a37bba55b771b9e68d3a554
01afc1dbe3cdeadd865d7a665e9160e77dd46ace
2248 F20101211_AABLOG aleman_d_Page_026.txt
f485705943ab2267f23904eef349d6bb
af302ed893efae61cae164d03e485ce1ddabb89c
47460 F20101211_AABLNR aleman_d_Page_132.pro
81fc171c86268779603ec98e2438f3f0
dd468a9d380cdb40bae19859734381eedea9b977
17620 F20101211_AABKKP aleman_d_Page_030.QC.jpg
4d77f902c1818c058a1dfaa91d29dfcf
fb6d9cea5c2b598af3f1f2fb652083b6a1cd2aa9
F20101211_AABKLE aleman_d_Page_070.tif
38714b53410f34530b00d5b57b006acb
62fd13ea2401469f69d016bb06f9a2d7d8a1b2fe
2173 F20101211_AABLOH aleman_d_Page_036.txt
35aba13ada417e85724ac88b5a908876
0c392f4c5c6022a9ba5eb47f314b09c1bad35de2
52736 F20101211_AABLNS aleman_d_Page_133.pro
88918796d9a05cca6e64eeb4dd9c94fe
2b7610e395991546bee17d6ba0d278f06714eae6
16379 F20101211_AABKKQ aleman_d_Page_040.QC.jpg
7ebef0618edf249b3fe2b5a4f70646c6
e4b3018cfff550f7b8b7c2b864f18ee5bc318bd9
56814 F20101211_AABKLF aleman_d_Page_129.jpg
343217569dc7f89b954b723fd70826ce
1f5e2e1f90d97bc1624bd714799eadb14558bef5
2430 F20101211_AABLOI aleman_d_Page_041.txt
d2019c9d6d8f59dbf5c21a4f1a178d96
d3f230694e2d219e287d74054ebcf7e0388cf58c
55309 F20101211_AABLNT aleman_d_Page_134.pro
668b0bd4edff05c7951f068d80e17f63
8eb49f8b8d92404d720dad7f3091a43e2539c5ee
1899 F20101211_AABKKR aleman_d_Page_074.txt
b00b7f0e5aab0be661002c778a9ab6dc
6f434c25d7e2b9f50bfe07bc828f4dac1ab38b0e
16657 F20101211_AABKLG aleman_d_Page_109.QC.jpg
18a372694fdb9eecdcacf671d854dc10
4a5bdc4c708220c6474a0fe643c5c3e3ecb17987
60257 F20101211_AABLNU aleman_d_Page_146.pro
648bdb7a3016dfcf79122b2b4ab5f219
4213a0097aec0a2f25b8fce23228ce59439861b3
746 F20101211_AABKKS aleman_d_Page_112.txt
ccd8345fa0be68f8095a76e4661caae0
1bab47cf61f3989b1a20c28348c189122890be57
220 F20101211_AABLOJ aleman_d_Page_042.txt
e21f8cab96a9d0e0e22e3890d5f3fcc6
55fe98bd4e74e60bc8b94a4411511a3ca6ceb8eb
53685 F20101211_AABLNV aleman_d_Page_147.pro
45cb8808c36fee5658c51dad556e9b10
417113d1eaa6538a662385f5a9fe76db70fd1e18
4952 F20101211_AABKKT aleman_d_Page_051thm.jpg
44d8784df430d2e3c6e20a9e81bb0072
dbd0545684464433ba6be3a66e3efe99f1a5eb3d
88480 F20101211_AABKLH aleman_d_Page_052.jpg
2d68d6ab130684d99526e11f83edd9e9
54b2d3af3dc054a2e2a98555d25de3a595da7e7f
556 F20101211_AABLOK aleman_d_Page_043.txt
10e3514088e97cd9d528c71b7105e0b9
72ae66a771f95c3dd6a1fc4eb25574784d095a0c
175 F20101211_AABLNW aleman_d_Page_003.txt
b9daeb16515334100590730fc295a420
d0c70cea785dd5f1ea98892020150ba1a97534fd
1051975 F20101211_AABKKU aleman_d_Page_111.jp2
b06a536262db768064ad75183aa5606d
5d3686ac7e6f27a1a97acfdc120030d98e7e33b1
5350 F20101211_AABKLI aleman_d_Page_104thm.jpg
85cb5f1a2161e88a7167dcc7fee730ef
e58c33b043b44b021228e05f99c8fabfa0922dcc
1934 F20101211_AABLPA aleman_d_Page_097.txt
e4cb57853ff1a4e14d93ca8688c9d98a
679ab5c9c482101f9ab48429148a735e17f87690
400 F20101211_AABLOL aleman_d_Page_044.txt
78e8807ee57c1b1260e9ca578c2d28ab
76c1358de45df6ea04954df6cd708dca324f8672
2059 F20101211_AABLNX aleman_d_Page_009.txt
f58567f58d8f961465a687f5171ebc95
8ebadf463e6e01a497072defb431ebedfcc1dcb3
F20101211_AABKKV aleman_d_Page_065.jp2
b0b1ab23bccedeef3b0c0d64fc036384
f1a690ddb196a7c21cf5f3c747691be93fae221a
52177 F20101211_AABKLJ aleman_d_Page_059.pro
c8388af819a62a69943a3d8dc4ccf4ff
89d7c0aea5a8bb8278151f212dd87410bfd5ed3d
1788 F20101211_AABLPB aleman_d_Page_099.txt
6ce54674660672c257bdf67e3a645d49
13f809033f7b05068c922b9b12a97694dd8909ae
1552 F20101211_AABLOM aleman_d_Page_045.txt
0f534bdb6b5466394c3a7fda490bc639
65365b319caa5f667932a9955be23ea6d8f9bff1
1142 F20101211_AABLNY aleman_d_Page_011.txt
05005fec3c55beaf0bc7acbfc90f2248
953349372dda22701c5925a3a3f668ac125b6722
6265 F20101211_AABKKW aleman_d_Page_086thm.jpg
456e76114c6a7a70e3e18ef0b73a377c
7937e78c5eb061f387ae1199d0efa1be36a6c0d9
2246 F20101211_AABKLK aleman_d_Page_001thm.jpg
ed6f08adac9c5bea326ed99a31730d6b
13311bd54241330ddd994062ce0b3114bde6a287
975 F20101211_AABLPC aleman_d_Page_104.txt
5325ed3cda10ddd1c9b695514696a649
ed8e46d962c76f1ecc733b0f445e6334f93fe4cb
2344 F20101211_AABLON aleman_d_Page_048.txt
10e3d08a32cd74c8eac4ed4a1183919c
8c7c9c68501e784c4851d8f847b6e9433a305dfb
2234 F20101211_AABLNZ aleman_d_Page_012.txt
e8ae71bd6f47c626c4524f2299a20815
888b27643fa1fe4e26f1ef92c9ecc04b36d0fd6e
1220 F20101211_AABKKX aleman_d_Page_008.txt
7f998cb6f4adedf310334444103570d0
45de3f073010d16d876852e836b456a98c25e83f
1051934 F20101211_AABKMA aleman_d_Page_052.jp2
272a3b30d6cb7450afca719893641da1
5499c1ad68fc90bae55ef0a4eb56bd9861ee4571
2220 F20101211_AABKLL aleman_d_Page_062.txt
5613a0037d017ff70675bee1b3ed9577
098724892aebfee25fccd909acc29291690e4603
812 F20101211_AABLPD aleman_d_Page_105.txt
cda216782eee6505b9a7034091386c4c
cb8951d524284846b7712c277520263e5ceb6054
F20101211_AABLOO aleman_d_Page_053.txt
9272ae6434ff6fc4b06074e05c5d8f9c
48930d442e8987a4a37198da57f05da94e221516
1357 F20101211_AABKMB aleman_d_Page_046.txt
9871342e15c901018398ac5b527bce64
87c66919fb7cdc758b5f145240aa3a2c99ee7ff8
63394 F20101211_AABKLM aleman_d_Page_060.jpg
136d3696ae3e8c2382cbce1a744c7201
3a7e4833c56a374919195942e682c98dfca5a41e
29510 F20101211_AABKKY aleman_d_Page_005.pro
d0fe36bc32533fe14f0cedf715712509
2fbc35b6549e1c838f2514be51bf7f913d9758de
864 F20101211_AABLPE aleman_d_Page_114.txt
030286ce379ce744d3f42d26fc380b1e
99c9eb7be22f3f3cb27ed384178dd712e6f104c0
2171 F20101211_AABLOP aleman_d_Page_055.txt
3ac090d78947e5f4b476cdb88788aeb1
2b6dd01c50a41efb31eb8ddf418231e01ec972a3
53978 F20101211_AABKMC aleman_d_Page_109.jpg
8ecf2619e48eca6ea053c8d4559b9a47
2a9147a7cb140d210f36903bbe03133d1c8e94eb
6837 F20101211_AABKLN aleman_d_Page_016thm.jpg
9a9d2340e1d03d8f26981744d4a951ea
cfc6c4e40229ead5b92ca77db0f2fb1b2f2ed7e8
1051986 F20101211_AABKKZ aleman_d_Page_063.jp2
4cf911e34cc598f3aa17f09de62ebec1
9fabdfc79bab444fe45377af0d1462b1384d6839
2048 F20101211_AABLPF aleman_d_Page_121.txt
107856b3b61c4ed447da64ae51db1579
940306fcd137fc1e9f2f6f2ba58c03e8f94691d0
1993 F20101211_AABLOQ aleman_d_Page_057.txt
d3d3dc2f0f2c79f52139f6fb001cd782
ebbd66fcae0cf5e19c0a9db6e051d5a9b50f6136
F20101211_AABKMD aleman_d_Page_025.jp2
7ec3b58effa214a5cf014b03eaa01d1d
e5d28b5c24319d0a222e853ee89eeb7bdc3f1bd1
F20101211_AABKLO aleman_d_Page_075.tif
495d7368d31d4f9fa76f3c7c6411b4bd
d790c482913b6e4761c0a02081191e94dce00c5f
1969 F20101211_AABLPG aleman_d_Page_122.txt
b39963f5bd883faff76fa276cb770a45
09e07094f3b9e4d10e7630cab020f2be2b4b2e8d
1589 F20101211_AABLOR aleman_d_Page_058.txt
504be0daeced345ba61c4ad88734d145
1d6ced250441f1311027f01c75e827b8e4bff716
F20101211_AABKME aleman_d_Page_025.tif
273808d9f0ca03b384dbfad0e0095d3a
481c3b2fee7164739a40408ff930a17c40879c87
59038 F20101211_AABKLP aleman_d_Page_051.jpg
6abae9ef8783fab13a505e9629ccb904
3b9662d868d05ab6f3d7fb28d117b92ef9eaeeda
852 F20101211_AABLPH aleman_d_Page_124.txt
f66d7594ba91b25acaee5e02695c9ccf
96d555443beb6e97b98cb15b57d15ef5519e4209
2103 F20101211_AABLOS aleman_d_Page_069.txt
e9601584352d66ce9ceff00cc5e75392
77176bc13a00a430505171c1ca7eec3a01fc3181
F20101211_AABKMF aleman_d_Page_084.jp2
17beaa952c3737a3325744d398e82f63
b74edd7cecd9736cbf939309fae4edab4aedb119
5701 F20101211_AABKLQ aleman_d_Page_114thm.jpg
0c2a656f76958bd8419568f938c4d50a
0be4195d9679cc187e463fbe3f4c532cc27b7d00
2368 F20101211_AABLPI aleman_d_Page_138.txt
af0c75748148b7a79efbba166a81a77b
1c88391bc9e3ccf18374801acff9552dacbb9188
2236 F20101211_AABLOT aleman_d_Page_070.txt
e81d6550a9b9494c724277e8c5746df6
0e8aeb4ad4257e7e37eda24feb6bbc6d4b2182ac
53575 F20101211_AABKMG aleman_d_Page_085.pro
17e32d12732bb2660c19349b642dd931
30e1ec382df950b56ccd5c428c1df56ec27d7acf
63762 F20101211_AABKLR aleman_d_Page_144.pro
49762423d86afa92f445aa59e8000ccc
6af314903fde120c01ef6ffd74344560452d12bf
2622 F20101211_AABLPJ aleman_d_Page_142.txt
4e71d5843d45238d947da6d86710a5e3
610a20e6e2a55e67d53dd5f68fcf52f95e2043b8
2116 F20101211_AABLOU aleman_d_Page_075.txt
d242525a8fb990ffe0e72a6c69f46642
3e59669cefa3c78530d9a172c9166103b0aae835
F20101211_AABKMH aleman_d_Page_081.tif
c69e507d578438a4f02ba641d5b501c4
f174ced253ca7700274120b0871907e7867ca4fb
1051957 F20101211_AABKLS aleman_d_Page_107.jp2
ec8507ad9e4a62ac5609b580e5812cbc
0412b56ff2eefd0ff6044021cb12642c70cbfb53
2126 F20101211_AABLOV aleman_d_Page_080.txt
4e666c3d0d55205c9576c26cddc8c6d8
59eb3b6e278497781e033bf9b32283dfdeb6d352
40589 F20101211_AABKLT aleman_d_Page_038.pro
4c263dba46a50022be8d84575e47be26
4d6c4b779482d57f236e0a1ad86db5d4e4d917f6
2517 F20101211_AABLPK aleman_d_Page_145.txt
d86f191209f8a36f2a94b84174186ee9
6b178f74c4d860c0c6bd5bf5b05e19df946d16cd
2114 F20101211_AABLOW aleman_d_Page_082.txt
77676f0d86a606ceab19682893eb5c5f
630c867514a860e03c9c227a89dbe0cf6bc630a0
6929 F20101211_AABKMI aleman_d_Page_052thm.jpg
3f9a097e14fc4be4fed56879b5015eca
438eb78a55d478680a7288e4576a885aaaa938b9
76385 F20101211_AABKLU aleman_d_Page_070.jpg
458e2a38e4b7dda39df3986eaeb8d34f
d71d35c809b6fd9f1b297e0fdc12a91f462d8d1f
6186 F20101211_AABLQA aleman_d_Page_023thm.jpg
5deb66acbe96dbad3d264c8b7720c1df
c7937cd1aa04c6c4cea643f9f55dfc9a4be52ea3
2413 F20101211_AABLPL aleman_d_Page_146.txt
2074d04c56960c5c4bf09d4a438443f2
a027df655323f9202e8abf24dd75602a4d7e3219
2007 F20101211_AABLOX aleman_d_Page_091.txt
b5c39ba2e3ef377ad9a3ff33c7eccdc7
43128fc05b18a8695655de12e922db05421750b9
59041 F20101211_AABKMJ aleman_d_Page_081.pro
07a9e79c0188468dc017a959ebef9045
46aa8da5bd2a62082eeded60478ea1385bd2fa88
74134 F20101211_AABKLV aleman_d_Page_015.jpg
8c4b2ffa900639e34ee7c768d3ac2259
7da7422cc492874ab709f7ae8cc77a2b67de1e4a
24980 F20101211_AABLQB aleman_d_Page_026.QC.jpg
326409ce5e746c1b6125d14c8a812895
fd573aa1b56f88bcee5a02150fc9b27a6071318f
818 F20101211_AABLPM aleman_d_Page_148.txt
8e28f4a9feb040db6a1020449bb8ac02
6f1e88ca72f3189dc09e285b04f093223ceeced7
2243 F20101211_AABLOY aleman_d_Page_092.txt
c305d099d829df6165a1674a31e5efda
d44f33f810d023399fd5adae69c5aac9f79ffdbe
F20101211_AABKMK aleman_d_Page_147.tif
dcb150b26348890f37ea97da23935abf
4e4bd7fb61d4e2201156dba63c35c5d7c6838466
26313 F20101211_AABKLW aleman_d_Page_089.QC.jpg
bf2e5a8758eb7e6770fcf43e86c2d178
b3988a07231115480487a041091bba1a9e6f88d7
5888 F20101211_AABLQC aleman_d_Page_027thm.jpg
d9959078f28c143209971fbee8139719
1561b661fa477b9ac8f109a49504d89177a0e7aa
7475 F20101211_AABLPN aleman_d_Page_001.QC.jpg
5340877b9b4f7d3c22886179b355b063
c6957928e962af779e1b141b3957eb6aa07420e3
2295 F20101211_AABLOZ aleman_d_Page_094.txt
c05169f81233b0aec966c84a9a5c14f5
d48158fbe9627b839054acb72ea8321c9137f985
2144 F20101211_AABKML aleman_d_Page_079.txt
214d085c54bb9f8ad601f5c70fce56d0
29d63e43309fe143be83d71c485076277d3d8678
79608 F20101211_AABKLX aleman_d_Page_075.jpg
027399602254e84593a2a08c82a20639
366392675c5431bc8ba258f0552dd19887bb7a04
4981 F20101211_AABKNA aleman_d_Page_099thm.jpg
ed19534951c01b4620baa6b8e7b2fae3
f579d25d0c43c8efe07064cc9baf355357aaf79f
17823 F20101211_AABLQD aleman_d_Page_028.QC.jpg
fdff330ccf228441fe751cb6af570bcb
674b9d1c93e650d1f70fe72fcf2d9526b9d6247d
4238 F20101211_AABLPO aleman_d_Page_003.QC.jpg
b699781e9caa2842d196e34833e74c26
8cf5dda988c6af761aae6c3f6f6efc1f8ed27b72
76168 F20101211_AABKMM aleman_d_Page_018.jpg
ef8411b4838222409dcf457cc00a59b7
e938fbc3d298e3c09fc111844a76770b460a14c6
15404 F20101211_AABKLY aleman_d_Page_008.QC.jpg
4c491023554e567618acb33c40cdba9e
df681cc037319d8ef35cf71628e2fa98d0516327
1051981 F20101211_AABKNB aleman_d_Page_032.jp2
88aa9c2d2e85025d396c74b446e8fb9e
2af7edd359dae385fe37b6ad18ea1123ac1cb2e6
5308 F20101211_AABLQE aleman_d_Page_028thm.jpg
b4f7148636ca671b69680708cb2c79af
6aec1186a064963312e9bd86824061fc08de28de
5100 F20101211_AABLPP aleman_d_Page_005thm.jpg
29d325bba4259c8383bc61360e27e9ed
69da3bbd5684c1083757debe567febf59cd4601c
970 F20101211_AABKMN aleman_d_Page_113.txt
41751c1bb8306d2fc72c8077b042d766
106689a74f787b95589ec7e0af49a18d031b51ab
882738 F20101211_AABKLZ aleman_d_Page_130.jp2
dd34cea7e14d3a676fff02484754121b
2a1965e9a5c239087d283baacc5e4498b5b84dc0
33452 F20101211_AABKNC aleman_d_Page_061.pro
6e34598e6dca24a315b84678e2010bd5
dba22ef61f53c570909443ebb776241ec4042755
22488 F20101211_AABLQF aleman_d_Page_029.QC.jpg
091a1c06e2d155508a3dac489b3f897e
b451cbc900d82c4b8900458489861d827fe3edea
5542 F20101211_AABLPQ aleman_d_Page_006thm.jpg
4182737879d6a29e71e19939b3a0258d
d1c4600d728887b691959ae31b671420128fd885
6732 F20101211_AABKMO aleman_d_Page_055thm.jpg
cb7c92093db23141584996c87139fea6
0ef13cd60b8d336b7848e27868f08161c0961410
F20101211_AABKND aleman_d_Page_148.tif
43f87d2c79be24f0b053c85d7733a833
fccf962550a83531a1d69b3c6e9869629eec6cd0
27124 F20101211_AABLQG aleman_d_Page_033.QC.jpg
0d246185fed5f481d39407d283f491d3
c29697c64f4e4e1c9ea287e82057d1ce451dc327
5875 F20101211_AABLPR aleman_d_Page_009thm.jpg
1299f6e1f8c57c635fa5cc30bceea9c0
d435a0cd982745b22a15b57e748105a8c5564ebe
89394 F20101211_AABKMP aleman_d_Page_036.jpg
c7b4418c9d78880b89b103b3179f77e1
441177fbacc4a19a4d33747bf0b7d0c48fc0d5ab
91192 F20101211_AABKNE aleman_d_Page_047.jpg
dd1442f8fb804587f20bfe93deef7f5c
4d9db4a7f9c3817e4a34f7c69ac5d9b17ec096f5
6585 F20101211_AABLQH aleman_d_Page_033thm.jpg
3d667f1fd03fc738f9444703decf771f
0d186f10d4ae39c5dba9c5488f4c7b7d0f08cc11
3545 F20101211_AABLPS aleman_d_Page_011thm.jpg
a8abc285f67788c0257b0ddd4f6d6150
d1442435dc72ad6120244b60aad7bc085d26ce95
6231 F20101211_AABKMQ aleman_d_Page_032thm.jpg
473855f06caababc59dbd5eac55140a8
4d6c8fe373e29542cf3784381a5992bc4f195080
15044 F20101211_AABKNF aleman_d_Page_064.pro
8d72511e95c614dc4e922d124f84801d
1c799e0446741587139d64998a7c1469712d5787
21147 F20101211_AABLQI aleman_d_Page_034.QC.jpg
f2fd5513fb5153a585d0502467ee19ea
03f3cbce6366b7a79e601e414af6a5bb40655a80
26076 F20101211_AABLPT aleman_d_Page_012.QC.jpg
ba6bb2bbadea93746d1b34e4546fa6c1
f5df55edf415d0e6f14e29bc2a70e2204d00bd68
763 F20101211_AABKMR aleman_d_Page_118.txt
f4a10ad75dc13b33393c0dcbe1235376
9e436b64f29fc334e8fa1c755acd797ed7c19c4c
1644 F20101211_AABKNG aleman_d_Page_119.txt
fd46bb673f29e015503d242ad52e58d2
461c1ac4d596ff9d2faa4ed46541b172314a4be7
6594 F20101211_AABLQJ aleman_d_Page_036thm.jpg
0f6a854a6df6cbef4cc81047628c0bb1
9a7bb2cee29a162d606053f0d72a3700431a5e0e
6709 F20101211_AABLPU aleman_d_Page_012thm.jpg
f56a2728c0a36fd75c4a813c3a393422
67a68d1a95cacc2b507ab467e94160bdcab22342
14488 F20101211_AABKMS aleman_d_Page_043.QC.jpg
dfdeabaa27ab71341c28e7c3f26ba7f3
39d7a086f00be8cca8c5153f5a773ac8797ba2c4
2378 F20101211_AABKNH aleman_d_Page_111.txt
9990a6073f963e11df800085680c2c95
7fe0d345dcdc305a87e5fe21ba3d3e26837f00e3
16476 F20101211_AABLQK aleman_d_Page_039.QC.jpg
8d8848ac5c129e87641cacc0bbf445a4
403e04f39b65205b5687f1ec33b4317079fd2d64
24944 F20101211_AABLPV aleman_d_Page_013.QC.jpg
b76b23a770339a51b2be1c81840a68e7
84bce483208d2cd0ac5a7ade99a59e7895a0d74f
2981 F20101211_AABKMT aleman_d_Page_020thm.jpg
0910f4f1491cc9c85615cb1291eedf34
32c6766ec94424774b2569cc9c8a60af44837c23
7123 F20101211_AABKNI aleman_d_Page_004.QC.jpg
497f28037446723c162def27d5e3ec79
356ae0333a4c469f5c4e2f39acf0ba6ec8eb2605
25226 F20101211_AABLPW aleman_d_Page_017.QC.jpg
4fd916ca3420dbc18be8d063f9dd7c4b
cb4dd4ff35b0820909c7cfda1be79c09f4705a06
90434 F20101211_AABKMU aleman_d_Page_065.jpg
b405df4c50d938f75a79a8e222724593
0cfa0566b4a557a09b756be29ddf03fb3c0cea99
6479 F20101211_AABLRA aleman_d_Page_066thm.jpg
2c2317887971c8a424b260d7cdb49e85
52ba4bd10128275678e8eacd992642acf28c867f
28588 F20101211_AABLQL aleman_d_Page_041.QC.jpg
ba90bab50a6572542bffd3a54b7d60aa
fefafd133a4b1fd85416c61972114c7f5bb510b8
F20101211_AABLPX aleman_d_Page_018.QC.jpg
b127a1db9d9cf4dce5c618ebd2c57e0a
154439829da9bd9951306bfcae5f4ebee59cd1e0
23646 F20101211_AABKMV aleman_d_Page_038.QC.jpg
d3268ff31f265fcb533423e894c53fcf
f73f7071b1b4ec7ed0d859d76dd79016c39539a6
6952 F20101211_AABKNJ aleman_d_Page_081thm.jpg
d24af9f79d8b975b84cfed64b3c663ac
db4d257cb0f057b64ceda8a77ccb88c7ee58cf21
8751 F20101211_AABLRB aleman_d_Page_067.QC.jpg
d619135d8900dcc1150e76138b42ad15
68e94bf54cbde531ad65e9242f23494523eb3de0
1845 F20101211_AABLQM aleman_d_Page_042thm.jpg
0d84581a8a82f7432e07d2dba7bbb9b6
1567cefa73852249c71c43bf4624faf429cbfd39
10342 F20101211_AABLPY aleman_d_Page_020.QC.jpg
eaa512531bc806879dc4be74c81ba93e
f986c7904261e76b41f88468374a8be50b071c49
20489 F20101211_AABKMW aleman_d_Page_127.QC.jpg
7d9b1c9b46c7f02d1a618ffe995a1995
8fc7efb8f6b5161e4fa92b7b7cba8082ad23db2a
2155 F20101211_AABKNK aleman_d_Page_034.txt
008ea4a7a9371dbd5a3090b17d040876
142f0172659bbd078fa0ea3b9c26e014deb75090
2945 F20101211_AABLRC aleman_d_Page_067thm.jpg
84dc0f30c6a286d5aa8b58861b97e9a5
e7e1d7df2da1006e7c893401786af053e5bf4c06
3997 F20101211_AABLQN aleman_d_Page_043thm.jpg
31b398ac22524c079a2763e6565e4ddf
56f6f774e8a29e7ac36ae6a4a3767d0da94e8680
25750 F20101211_AABLPZ aleman_d_Page_021.QC.jpg
afb6f501b64d89df664460674480578a
55ff9b9b73a50b74a57f58d148f6a75b808ab9a9
10818 F20101211_AABKMX aleman_d_Page_135.QC.jpg
221ba1e2944760efe52bc58fcea2935f
733d42002642d88955d7d7438614eb816148ee01
5138 F20101211_AABKOA aleman_d_Page_040thm.jpg
7a44ce8b5332678671ae8c56214f0af2
cf48798b85251b300ef876bcffe8594ac8489eba
732718 F20101211_AABKNL aleman_d_Page_035.jp2
9f65212227265b68878ed6713b88c49e
4e4f5dff0bb1118f8bfb3a316cd41c7b2849cfe4
5908 F20101211_AABLRD aleman_d_Page_068thm.jpg
79b6d31f79489ed6547e666b659f8074
744556a033bd587d4dda46ce820a1638284f8cd5
4469 F20101211_AABLQO aleman_d_Page_045thm.jpg
8031d614d7c6524bb2c45725e4ce3925
3a713b184f6ad4651089eec985c9bcd7795a2785
856 F20101211_AABKMY aleman_d_Page_002.pro
e03d2480198659f6eb429eab981d48f1
bcd3666409b61e2320d748d58e65953e33ddad94
1674 F20101211_AABKOB aleman_d_Page_060.txt
13cdce622107770dc65a2d7b84648e73
eb5d67852de57c87fd57da664d05b92fb59dff2d
87128 F20101211_AABKNM aleman_d_Page_033.jpg
1231877e4847370c62ca614e60713a53
d14ac7ae4c4115a207c5627b31fae661b01924ae
25627 F20101211_AABLRE aleman_d_Page_069.QC.jpg
d262ad26b241a7ac1cea975bf1fac10c
4332f519a68fb4f476afbe5ed36d4d2a6389188e
27475 F20101211_AABLQP aleman_d_Page_048.QC.jpg
9df5b526ef65c18f775000b97f4d4bcb
9cf1512bb78994c3511b036dd96e7c359ba13860
5457 F20101211_AABKMZ aleman_d_Page_118thm.jpg
63dec9f9e01364509f71e151c7fd55a1
f0445adc3f6f55d715432b74fb2cba416312cc26
56940 F20101211_AABKOC aleman_d_Page_050.pro
f317f3e529b452750963f48fca04ccec
0214b258808b4b552c260d0cc973279c42592116
5661 F20101211_AABKNN aleman_d_Page_010thm.jpg
2fd707d3c18ee0d318d289f2932c3f04
22a85d71f335fdd1a7f0bebd8981ad92a2754594
24824 F20101211_AABLRF aleman_d_Page_070.QC.jpg
25029ed9eca1565774cecc397b2c1f96
12e887053ae2472395180bcc9f6bad8a2df5c095
26628 F20101211_AABLQQ aleman_d_Page_050.QC.jpg
d2ab955e19c05cdcbdcb65026d74879e
4e2b5c44877f020bd7f3f601f26d9dc3ca1a2b53
24308 F20101211_AABKOD aleman_d_Page_086.QC.jpg
ad6b2bd7da488481b76f0f5fc6a53ca9
97a9f421414033b9e972a3ff6581205c4ed1f332
24631 F20101211_AABKNO aleman_d_Page_001.jp2
d1f86e81bf3e13ea534f0db5ab930509
36bb6df923f175b558bb70775db4d0086a6660e9
5234 F20101211_AABLRG aleman_d_Page_071thm.jpg
f55ccc20d02b8a952e0094253686fad8
6cfdebee29782d1c6ec8fe63469b3b0671fbdc74
6813 F20101211_AABLQR aleman_d_Page_050thm.jpg
9c0027686c40e08be1185c5b7b6ba4b4
5f54f893261a57b6960949eeee7b4fb1be655ae3
918203 F20101211_AABKOE aleman_d_Page_126.jp2
a6784aef2cb101e0ea8ad0fef45cd8ce
718c340870e47a381803036ea6539a401a5e151f
55482 F20101211_AABKNP aleman_d_Page_010.pro
21bba66a1ea3fbbe6da5ce9802c60346
dda9b3a0956bbb88c517ee077bf1470524db8b4c
26688 F20101211_AABLRH aleman_d_Page_072.QC.jpg
03e363824852a75c3bb922e2183d16d6
ebfcf231734dd9684c700bccf636c372b10f8d43
6047 F20101211_AABLQS aleman_d_Page_057thm.jpg
75bf5220ccbb81cc7d9e0314e3d72780
5f7db8f7e587f36990248a92412d095de1ad3b7c
6490 F20101211_AABKOF aleman_d_Page_085thm.jpg
ad934f861326722cfd2a3ab9d4a14185
1e4a6122fcc3f8dd59b7b3be23503e242db5435c
2240 F20101211_AABKNQ aleman_d_Page_033.txt
43377a63963c4422c4fc6a2f00252afb
59762bcd1b059046ae769a5e9f60f740a335f3dd
6667 F20101211_AABLRI aleman_d_Page_075thm.jpg
fca2ca4d78edd1b2fdebc13ca3f5346a
461920b2f0ee13dc21c2592e9429d46b7c446c54
4984 F20101211_AABLQT aleman_d_Page_058thm.jpg
0db62a7c5edff99963788050dd95748d
f9d41d05a854464487dc28d16007448eb7e04f16
33710 F20101211_AABKOG aleman_d_Page_148.jpg
61c5791f2ef20347e01e2bfb062cd642
a3c22383a4f733dddbe40c7fea0f820ba543eebc
47582 F20101211_AABKNR aleman_d_Page_103.pro
e27b457094c0d86d83b53f87aa4615a0
9cafc25f033b2cd907ad9f31fee8aee639a9eb82
25023 F20101211_AABLRJ aleman_d_Page_076.QC.jpg
4dfedacd1830e30c7a47b7952417797e
3c795c3e9a8632b1da3fa5e7b4461890b753961d
24397 F20101211_AABLQU aleman_d_Page_059.QC.jpg
f76fb1c3c4b3c764f415b85fd606245b
9dad4e3670ae4eb97a85a6239c705b5de21526fc
16158 F20101211_AABKOH aleman_d_Page_113.pro
01f3447fef36e43f6de4a7af47700b35
d7f5cc0922d63a02a6f4540fcf68de89bcfbad51
24555 F20101211_AABKNS aleman_d_Page_143.QC.jpg
9f9195594884a50ce9bb2a6224bbbd78
2865756e51b82ab9a18288b6c9fbf235dbcde0d4
15805 F20101211_AABLRK aleman_d_Page_078.QC.jpg
06a81432faa58141b38c0a937a197520
624c727fe7d0c129ae32db65a58ffc3af09ef898
5219 F20101211_AABLQV aleman_d_Page_060thm.jpg
ae0c273ce87b313be5e3953b1b32517b
9d540feb18314e96f6a9a6fef711e07a8456b1fb
91177 F20101211_AABKOI aleman_d_Page_041.jpg
f6ab2db299d605ebffe22d616cc4eccf
30df4cfa87902f1982354182286a4e6a006dd01c
F20101211_AABKNT aleman_d_Page_091.tif
64bde232c1775c4d85ac50634c28e54b
6c6b9aefc4c21e411361f0f6f64c6d6250de1e8e
6235 F20101211_AABLRL aleman_d_Page_080thm.jpg
549a9bbb671891bfa336404507af2139
984b4082b901d6ffc1924de4fbeb49535d69d903
26345 F20101211_AABLQW aleman_d_Page_062.QC.jpg
fbdb1c05791accdcc00656dae153f1df
527c955710d26ef52baeb9c5dacc5b37bd71b7d4
87645 F20101211_AABKOJ aleman_d_Page_120.jpg
69ba0a7d18d9d3c536d1599bbe539bc9
b2510ac2f5a2fac0afa3492fd0ed0bfc34379d39
5134 F20101211_AABKNU aleman_d_Page_100thm.jpg
455ca7062454a7ba922efefb85ee9a80
63d2efdcf88d186753066c53528a3e424da27349
17367 F20101211_AABLSA aleman_d_Page_100.QC.jpg
916b088d2562ceb182e26e0ead56e6b0
950cc4ad3da2d5f7e3329cd154acd09e593c6613
6797 F20101211_AABLQX aleman_d_Page_062thm.jpg
3e8edd5622086f34cd076edd92d9ca67
a8838b4f7a84cb28c8fd672c71aac25c1f04f33b
1527 F20101211_AABKNV aleman_d_Page_087.txt
41808faf5f609db9eddfac4126aed64b
32525b137e1248e69e2e7ea104fb5faa477d4e1f
6377 F20101211_AABLSB aleman_d_Page_101thm.jpg
56bd2a43620dabf0d62658982ea5315c
1bc53075206a57e88dbeff1397b5a66a0bd34e5d
28240 F20101211_AABLRM aleman_d_Page_081.QC.jpg
a0200abf2b007759fd2a69510c74d765
cc5844f00598214f1e358d43f76a5b7c915af738
10335 F20101211_AABLQY aleman_d_Page_064.QC.jpg
1c32fb342e229450c1fd9cccbfd9f1f6
4b9b7be9792efe70d2a841b5312ddf6c5a3d42ec
5433 F20101211_AABKOK aleman_d_Page_002.jp2
92ebcf8112be3687322f38c10a4b654d
b7fa4579a5de711040b186d4c2799d90128b851e
19213 F20101211_AABKNW aleman_d_Page_040.pro
c4891b444de50df2280494872e980d5b
53f7cdd70a8b43d6a7c0f2a312bea57a3a267aeb
6217 F20101211_AABLSC aleman_d_Page_102thm.jpg
11e49e93300163ae98b2ce40739799a0
edfea58fb394780ff6025a5b077669c416d1c335
26638 F20101211_AABLRN aleman_d_Page_083.QC.jpg
d7ef7a8a18a68986c7b8d4164d0935e4
db9e350b3d52b1f87516129444a0e62596dc7dbe
28245 F20101211_AABLQZ aleman_d_Page_065.QC.jpg
74b48db1842a3d81d5f5493dcdc71ef9
351d6b0052de9c44c5719eea9b87876187b4a93e
F20101211_AABKPA aleman_d_Page_096.txt
248f5c8fa39df305f199020a133f6287
d890e8b3bb56c541b3252fc31d042f57b2e812fe
2269 F20101211_AABKOL aleman_d_Page_050.txt
7f54994865c6a983d42c056e5440087b
26e81dffb48339bf1ce4569c61333a0d8e0368d8
F20101211_AABKNX aleman_d_Page_076.jp2
077e9ad22a3df873c71c0315c4c0e19f
da712ce58dc32f852b0ca30efbfda5de9383d408
6105 F20101211_AABLSD aleman_d_Page_103thm.jpg
cab5c0adf7b26932a56abe0d9c1e4dca
7759cf2ecdd6a5fdd7e22f127fe2415101c173e3
5987 F20101211_AABLRO aleman_d_Page_084thm.jpg
96710d9c96e10b8bfc4e66ce9705ea88
843f27430e257bcb73b98c6c0d281906a506dda8
53403 F20101211_AABKPB aleman_d_Page_128.jpg
e5659ba9ef8739373ce198d9ab0d2c4e
8cb128ce2d15da09be48035c3e03a9365ae87e99
911877 F20101211_AABKOM aleman_d_Page_093.jp2
fcb659abf12ab7848199504e23449093
db7cc8f9d868b64ecf5e289fd5d491e27828b9e3
436847 F20101211_AABKNY aleman_d_Page_064.jp2
7958f4e07882eae670f7d99cffb8c69d
9dd2d12a0917bdfb94f14b630f9d8815ab7e812d
17661 F20101211_AABLSE aleman_d_Page_106.QC.jpg
0e6eb5dc0dbd754232f7b982cda06025
9ddb075fa237227638db67fa3ef810443804460c
25836 F20101211_AABLRP aleman_d_Page_085.QC.jpg
37f086ca43adb6e7e9d3a423f38e3084
03ebb7d6aeb6ba3b3a5fd5c67e20af476f058969
1051961 F20101211_AABKPC aleman_d_Page_041.jp2
a25f59490a0b0133d04da2ff55469d95
dcbe6511e0fc40151bbbd419afa7a666a86c0f07
6177 F20101211_AABKON aleman_d_Page_074thm.jpg
cd90f65ea8063e79a51e01b70e8e5a3d
342ed9472c8cd843daaa762bcb931c27fc22beb5
2049 F20101211_AABKNZ aleman_d_Page_088.txt
4637b80ffa6891540dd7745c8aecadf0
5b0fc4d7d582db887dad23e954af9abbee031986
5664 F20101211_AABLSF aleman_d_Page_106thm.jpg
839053867cbae66312fb5b87d941e594
6bc31ccea1acb09547dc6992f3615440fabd1d2c
15407 F20101211_AABLRQ aleman_d_Page_087.QC.jpg
fa7e3da2d0febf449fc2040ccf35f8a8
5b01d919132e93841cae394050849016f6de79b3
16606 F20101211_AABKPD aleman_d_Page_117.QC.jpg
8c621d820c828847e740726f3a48b4f4
1e3cca017299ffe112435978b3a3d9ee718c4dbf
51766 F20101211_AABKOO aleman_d_Page_137.pro
dd1b59b6940e28d34df1bc9896c6f200
c66ce4475dab85b0c61e953fd9ae78d9ae1690cc
16954 F20101211_AABLSG aleman_d_Page_110.QC.jpg
b75ef983c165c85328e9bb3d1d67c07d
6cae36447478498525a1f95a63766dfe0bc3b44a
22388 F20101211_AABLRR aleman_d_Page_088.QC.jpg
7f28a6ab65031c0d8c5d9d3e9d37447e
58058333531d8f9a7d223e69e0df2afba3ba9ce6
1225 F20101211_AABKPE aleman_d_Page_051.txt
6216b92df35fbe3e8be7cb5e13fef491
8b617afc4eb55746e69d0fc4a4a513cc08c7f920
1196 F20101211_AABKOP aleman_d_Page_039.txt
0ee3fa11b43e98c7d70cf2380da2aa22
76601d7086b44bb925ec132251813119eb369f74