<%BANNER%>

3D Deterministic Radiation Transport for Dose Computations in Clinical Procedures

University of Florida Institutional Repository
Permanent Link: http://ufdc.ufl.edu/UFE0022617/00001

Material Information

Title: 3D Deterministic Radiation Transport for Dose Computations in Clinical Procedures
Physical Description: 1 online resource (140 p.)
Language: english
Creator: Al-Basheer, Ahmad
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Nuclear and Radiological Engineering -- Dissertations, Academic -- UF
Genre: Nuclear Engineering Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The main goal of this dissertation was to establish the feasibility of basing megavoltage external photon beam absorbed dose calculations in voxelized phantoms on SN deterministic calculations and pre-calculated electron absorbed dose kernels derived from full-physics Monte Carlo. The SN derived electron absorbed dose kernel method EDK-SN, developed as part of this research, achieves total execution times that are on the order of several times to orders of magnitude faster than conventional full-physics Monte Carlo electron transport methods considering equivalently detailed models and data fidelity. With the rapid movement toward intensity modulated radiation therapy (IMRT), radiation beam intensities have increased dramatically over the past decade, thus heightening the need for further characterization of out-of-field organ absorbed doses, along with their associated biological risks. Assessment of these tissue absorbed doses is complicated by two fundamental limitations. First, anatomic information on the patient is generally restricted to a partial body CT image acquired for treatment planning; consequently, whole-body computational phantoms must be employed to provide the out-of-field anatomy model structure for absorbed dose evaluation. Second, existing methods based on Monte Carlo radiation transport, even with the application significant variance reduction, are quite computationally inefficient at large distances from the primary beam, and point-kernel methods do not properly handle tissue inhomogeneities. Moreover, since absorbed dose are generally tracked in all major organs in the body, variance reduction schemes for Monte Carlo are not all effective in this regard. The outcome of this dissertation is to demonstrate that absorbed dose from high-energy external beams radiation can be accurately computed for whole body and organ-specific absorbed doses. The EDK- SN method implements voxelized phantoms with discrete ordinates (SN) transport computations coupled with directional influences and (pre-computed) full-physics Monte Carlo based electron absorbed dose kernels to yield total body absorbed dose information. This research shows that the deterministic techniques coupled with Monte Carlo based electron absorbed dose-kernels has significant potential for organ absorbed dose evaluation in the clinical management of radiation therapy patients.
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 Ahmad Al-Basheer.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Sjoden, Glenn E.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022617:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022617/00001

Material Information

Title: 3D Deterministic Radiation Transport for Dose Computations in Clinical Procedures
Physical Description: 1 online resource (140 p.)
Language: english
Creator: Al-Basheer, Ahmad
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: Nuclear and Radiological Engineering -- Dissertations, Academic -- UF
Genre: Nuclear Engineering Sciences thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The main goal of this dissertation was to establish the feasibility of basing megavoltage external photon beam absorbed dose calculations in voxelized phantoms on SN deterministic calculations and pre-calculated electron absorbed dose kernels derived from full-physics Monte Carlo. The SN derived electron absorbed dose kernel method EDK-SN, developed as part of this research, achieves total execution times that are on the order of several times to orders of magnitude faster than conventional full-physics Monte Carlo electron transport methods considering equivalently detailed models and data fidelity. With the rapid movement toward intensity modulated radiation therapy (IMRT), radiation beam intensities have increased dramatically over the past decade, thus heightening the need for further characterization of out-of-field organ absorbed doses, along with their associated biological risks. Assessment of these tissue absorbed doses is complicated by two fundamental limitations. First, anatomic information on the patient is generally restricted to a partial body CT image acquired for treatment planning; consequently, whole-body computational phantoms must be employed to provide the out-of-field anatomy model structure for absorbed dose evaluation. Second, existing methods based on Monte Carlo radiation transport, even with the application significant variance reduction, are quite computationally inefficient at large distances from the primary beam, and point-kernel methods do not properly handle tissue inhomogeneities. Moreover, since absorbed dose are generally tracked in all major organs in the body, variance reduction schemes for Monte Carlo are not all effective in this regard. The outcome of this dissertation is to demonstrate that absorbed dose from high-energy external beams radiation can be accurately computed for whole body and organ-specific absorbed doses. The EDK- SN method implements voxelized phantoms with discrete ordinates (SN) transport computations coupled with directional influences and (pre-computed) full-physics Monte Carlo based electron absorbed dose kernels to yield total body absorbed dose information. This research shows that the deterministic techniques coupled with Monte Carlo based electron absorbed dose-kernels has significant potential for organ absorbed dose evaluation in the clinical management of radiation therapy patients.
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 Ahmad Al-Basheer.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Sjoden, Glenn E.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022617:00001


This item has the following downloads:


Full Text





3D DETERMINISTIC RADIATION TRANSPORT FOR DOSE
COMPUTATIONS IN CLINICAL PROCEDURES


















By

AHMAD AL-BASHEER


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

2008





































O 2008 Ahmad Al-Basheer



















c-~~'


d3J\ S
(+YH:
Ir//


/ / **









ACKNOWLEDGMENTS

Any project, no matter how individualized, will almost certainly require input, assistance

or encouragement from others; my proj ect is no exception. I extend my sincerest gratitude to Dr.

Glenn Sjoden. He has been a constant source of thoughtful guidance in pursuing this project.

Because of his input, advice and challenge, I have matured as a researcher and as a student. I

would also like to acknowledge the contribution of my supervisory committee: Dr. Ali

Haghighat, Dr. James Dempsey, Dr. Wesley Bolch, Dr. Ayman Hawari and Dr. William Ditto.

I would also like to take this opportunity to express my sincere gratitude to my classmates

and the faculty members in the Nuclear and Radiological Engineering and Radiation Oncology

Departments for their constructive suggestions and helpful comments. During my research, I had

the pleasure of working with the University of Florida Transport Theory Group (UFTTG). With

the help of coworkers and friends, even the tedium became bearable. Special thanks go to my

family, and mostly my wife Ghadeer, Mom and Dad; they have definitely been encouraging me

for the longest! They will always be dear to my heart. I thank God for all He has done for me.












TABLE OF CONTENTS


page

ACKNOWLEDGMENTS .............. ...............4.....


LIST OF TABLES ........._.._ ..... ._._ ...............8....


LIST OF FIGURES .............. ...............9.....


AB S TRAC T ............._. .......... ..............._ 14...


CHAPTER


1 INTRODUCTION AND LITERATURE REVIEW ................. .............. ......... .....16


1.1 Overview ............... .... .. ......... ...............16......
1.2 Initial Clinical Dose Evaluations .............. ...............17....
1.3 Pencil Beam Transport ................. ...............19................
1.4 Solving the Boltzmann Equation .............. ...............21....
1.4.1 Monte Carlo Method ................. ...............23.......... .....
1.4.2 Deterministic M ethods ................... ...... ........... ...............25......
1.4.2.1 Discrete ordinates method (SN method) ................. ......................25
1.4.2.2 Monte carlo versus deterministic SN methods .................. ...............27
1.5 Dose and Charged Particle Equilibrium............... ......................2
1.5.1 Deterministic Dose Calculations for Diagnostic Application ................... ........30
1.5.2 High Energy Beam Dose Calculations ................. ...............31...........
1.5.3 Limits of Charged Particle Equilibrium ................. ............... ......... ...31
1.6 Outline of This Research............... ...............35


2 CRITICAL DISCRETIZATION ISSUES IN 3-D DISCRETE ORDINATES
MEDICAL PHYSIC S SIMULATION ................. ...............36................


2.1 Code Systems ................. ...............37........... ....
2.2 Simulation Geometry .............. ...............38....
2.2.1 PENTRAN Model ................. ...............3.. 8......... ....
2.2.2 M onte Carlo M odel .............. .... ...... ....... .. ........3
2.3 Comparison of Deterministic and Monte Carlo Results ................ ............ .........39
2.3.1 Investigation of Angular Quadrature .........._.._.._ .....__ ......._.. .........39
2.3.2 Investigation of Spatial Discretization (Meshing) .............. .....................4
2.3.3 Investigation of Cross-Section Libraries:.. ...49............
2.4 Absorbed Dose Calculations ................. ...............52......... .....
2.5 Parallel Decomposition ................. ...............54...............
2.6 Discussions and Findings ........_................. ............_........5











3 WHOLE BODY DOSIMETRY SIMULATIONS WITH THE UF-B SERIES
PHANTOM USING THE PENTRAN-MP CODE SYSTEM FOR LOW ENERGY
PHOTONS ............ ..... ._ ...............57....


3.1 Methodology .................. .. ...............58.
3.2 UF Series B Pediatric Phantoms .............. ...............59....
3.3 PENTRAN-MP Code System ............ ..... ._ ...............61...
3.3.1 Pre-processing Codes.................. ..........___......__ ............6
3.3.1.1 The collapsing code (GHOST-3-D)............... ..............6
3.3.1.2 The mesh generator code (PENMSH-XP) ................. ............... ....61
3.3.1.3 S ource spectra and cros s- secti ons generate on ................ ..................62
3.3.2 Transport Calculation............... ..............6
3.3.3 Post-Processing Codes ................. ..................... ........6
3.3.3.1 Data management utility (PENDATA) .............. .....................6
3.3.3.2 Dose calculation code (3-D-DOSE) .............. ....................6
3.4 Computational Example............... ...............64
3.5 Computational Results .............. .. .. ........... ......... .. .......6
3.5.1 Flux Comparison of Deterministic and Monte Carlo Results ................... ........67
3.5.2 Dose Calculations and Dose Volume Histograms (DVH). ............... ... ............70
3.5.3 Parallel Computation............... ..............7
3.6 Discussion and Findings .............. ...............76....

4 ELECTRON ABSORBED DOSE KERNELS TO ACCOUNT FOR SECONDARY
PARTICLE TRANSPORT IN DETERMINISTIC SIMULATIONS .............. ...............79


4.1 Introduction to The EDK-SN Method .............. ...............79....
4.2 Monte Carlo Based Dose Kernels................. ......... ....................8
4.3 EDK Determinations Based on The MCNP5 Monte Carlo Code ............... ... ........._...82
4.3.1 Computational Model............... ...............83.
4.3.2 MCNP5 Tallies Comparison ................. ...............83................
4.4 EDK -SN Code System Methodology .............. ...............86....
4.4.1 N et Current............... ..... ............8
4.4.2 MCNP Based Electron Kernels............... ...............87
4.4.3 Euler Angles............... ...............89.
4.4.4 Ordinate Proj section ................. ...............92.____ ....
4.4.5 Final Dose Calculations .............. ...............93....
4.5 External Source Benchmark............... ........ ...............9
4.6 Comparison of Deterministic and Monte Carlo Results ..............._ ............_.........96
4.7 Parallelization of EDK-SN Code ................. ......... ...............99.....
4.8 Discussion and Findings .............. ...............101....

5 APPLICATION OF EDK-SN TO MODELS WITH HETEROGENITIES .........................103


5.1 Heterogeneous Tissue-Equivalent Phantoms ......____ ..... ... .__ ........_......103
5.1.1 Monte Carlo Simulations .............. ...............103....
5.1.2 PENTRAN Model ................. .......... ...............107....
5.1.3 EDK-SN Absorbed dose Comparison............... ..............10











5.1.4 Correction Factor Approach ................. ...............109........... ..
5.2 Computational Phantom of 15-Year Old Male ................ .............................111
5.2.1 EDK-SN Absorbed Dose Calculations ................. ............... ......_. ...114
5.2.2 Ray-Effects ................ ...... ......_.._ ....... .. ....... ... ....... ...............1 6
5.2.3 MCNP5 Energy Deposited Tally (F6) and Pulse Height Tally (*F8)
Results ................. ...............117____.......
5.3 Discussion and Findings ............ _...... ._ ...............118..

6 SUMMARY, CONCLUSIONS AND FUTURE WORK .............. ..... ............... 12

6.1 Conclusions and Findings .............. ...............121....
6.2 Future W ork ............ _...... ._ ...............123...

APPENDIX


A HIGH ENERGY PHOTON TREATMENT (IN CONDENSED ELECTRON
TRANSPORT) ................. ...............124................

B MCNP5 PHNTOM INPUT DECK .............. ...............133....

LI ST OF REFERENCE S ................. ...............136................

BIOGRAPHICAL SKETCH ................. ...............140......... ......










LIST OF TABLES


Table page

1-1 Comparison between Monte Carlo methods and deterministic techniques.............._.. ....28

1-2 Approximate attenuation of photons with a layer of water equal to the maximum
range of secondary charged particles28 .........._.... ........_....._. ............3

1-3 Approximate thickness of water needed to achieve various attenuation amounts
compared to maximum energy electron ranges generated by the same beam ...................32

2-1 Total number of directions, cumulative problem time** required for a single energy
group, P3 scattering, 126,000 fine mesh cells using 12 processors on a parallel
cluster ................ ...............54..__.._.......

3-1 Computed tomography image sources for the development of the UF pediatric
phantom series .............. ...............60....

3-2 Total number of directions, Cumulative Problem Time required for different energy
groups, P3 Scattering, and 189k fine mesh cells............... ...............76.

5-1 Comparison of selected organ absorbed dose rate (MeV/g. s) calculated using
MCNP5 pulse height tally with (photon, electron mode) and EDK-SN for the UF
hybrid 15-year-old male phantoms for a chest 8 MV X-ray flat weighted source ..........115

5-2 Prostate absorbed dose rate (MeV/g.s) calculated for the test problem using EDK-SN
and MCNP5 using pulse height tally (*F8) with (photon, electron mode) at different
MCNP5 Monte Carlo running times .....__._.. ... ..._.... ...............116.

5-2 Absorbed dose rate (MeV/g. s) calculated for the test problem using EDK-SN and
MCNP5 using pulse height tallies (*F8) with (photon, electron mode) .......................117

5-3 Absorbed dose rate (MeV/g. s) calculated for the test problem using MCNP5 using
pulse height tally *F8 with (photon, electron mode) and MCNP5 energy deposited
tallies (F6) ........._ .......... ...............118....

A-1 Ratios of photon interactions in water for range of energies ..............._ ............ .....124










LIST OF FIGURES


FiMr page

1-1 Comparison between thicknesses required to attenuate different percentages of
primary photons and Rcs range for maximum energy secondary electrons
produced by the same beam ................. ...............33.____.....

1-2 Comparison between thicknesses required to attenuate different percentage of
primary photons and average range for maximum energy secondary electrons
produced by the same beam ................. ...............33.____.....

2-1 60C0 unit (left: Mesh distribution generated by PENTRAN package; Right: MCNP5
Cell/Surface schematic input) ................ ...............38........... ....

2-2 60C0 y-rays 3-D Groups 1,2 3 and 4 scalar flux computed by PENTRAN with the
CEPXS gamma cross section library ................. ......... ...............40.....

2-3 S12 (Level Symmetric) and S42 (Legendre-Chebychev) quadrature first octants (with
21 and 231 directions/ octant respectively) .............. ...............41....

2-4 60Co y-ray 3-D Group 1 scalar flux distribution for x-level depths at 0 cm, 5 cm, 10
cm and 15 cm, computed by PENTRAN with the CEPXS gamma cross section
library ................. ...............42.................

2-5 60Co y-ray 3-D Group 1 scalar flux distribution for x-level depths at 0 cm, 5 cm, 10
cm and 15 cm, computed by PENTRAN with the CEPXS gamma cross section
library; an S42 angular quadrature (1848 directions) with P3 Scattering anisotropy ..........43

2-6 Scalar flux distributions using S12, S22, S32, and S42 with P3 Scattering anisotropy in
PENTRAN using the BUGLE-96 gamma library .............. ...............44....

2-7 60C0 model spatial meshing, axial level............... ...............47.

2-8 Scalar flux comparison of variation in the interval of 3-D mesh cells containing air;
fine meshes of 1 cm size (along each axis) in air proved to be best in providing high
detail flux distribution............... ..............4

2-9 Scalar flux distributions generated by PENTRAN for a single energy group of each
library using the MCNP5 solution as the reference case. ................. .................5

2-10 Relative deviation of energy group 1 discrete ordinates results with Monte Carlo
results ............... ...............52........_ ......

2-11 Absorbed dose rate comparison between SN method using kerma approximation and
MC techniques using F6 tally, MC statistical uncertainty were less than 3% on
average ................. ...............53........ ......










3-1 PENTRAN-MP Code System Structure Flowchart ................. .............................58

3-2 The UF Series B pediatric phantoms series ................ ...............60..............

3-3 Values of the mass energy-absorption coefficient, IdenlP, and the fitting function as
a function of photon energy, for Lung (ICRU-44), soft tissue (ICRU-44) and cortical
bone ................ ...............64....... ......

3-4 UF-Series B phantom (11 years old male) with (398 x 242 x 252 voxels); the
corresponding PENTRAN "back-thinned" models with (132 x 80 x 84 voxels) and
72 materials (upper right), and another with (79 x 48 x 50 voxels) and 66 materials
(low er ri ght) .............. ...............66....

3-5 Transversal view of the 11i-year-old male Series B original phantomcollapsed into
189k voxels and 4 materials, collapsed into 900k voxels and 4 materials (lower left)
and collapsed into 900k voxels and 72 materials............... ...............6

3-6 x-ray 3-D Group 1, 3, 5 and 7 scalar flux computed by PENTRAN with the CEPXS
cross section library, S32 angular quadrature (1088 directions) with P3 Scattering
ani sotropy ................. ...............68.._._._.......

3-7 Flux distributions using S32 with P3 scattering anisotropy in PENTRAN using the
CEPXS library vs. MCNP5 Monte Carlo results ................. ...............68........... ..

3-8 Angular flux distributions using S32 with P3 scattering anisotropy in PENTRAN
using the CEPX S library vs .............. ...............69....

3-9 Angular flux distributions using S32 with P3 scattering anisotropy in PENTRAN
using the CEPXS library vs. MCNP5 Monte Carlo results for energy groups 1, 3, 5
and 7 along the z axis at different depths in the phantom............... ...............69

3-10 Simulation methodology for 3-D absorbed dose computations using PENTRAN-MP
code system ................. ...............71.................

3-11 CT view of the 11i-year-old male Series B original phantom collapsed into 189 k
voxels and the correspondent absorbed dose distributions 900 k voxels and the
correspondent absorbed dose distributions .............. ...............72....

3-12 CT view of the 11i-year-old male Series B original phantom collapsed into 189 k
voxels and the correspondent absorbed dose distributions 900 k voxels and the
correspondent absorbed dose distributions .............. ...............73....

3-13 CT view of the 11i-year-old male Series B original phantom collapsed into 189 k
voxels and the correspondent absorbed dose distributions 900 k voxels and the
correspondent absorbed dose distributions .............. ...............74....

3-14 Simulation methodology for 3-D organ absorbed dose computations using
PENTRAN-MP code system .............. ...............74....










3-15 Dose Volume Histograms (DVH) computed based on PENTRAN (S32,P3) transport
results using the CEPXS library vs. DVHs computed using MCNP5 Monte Carlo
results .........._...._ ...............75.... ..._ .....

4-1 Procedure to produce EDK based on Monte Carlo Calculations............... ..............8

4-2 Monte Carlo model used for comparisons between MCNP5 pulse height tally, energy
deposited tally and cell flux tally employing dose conversion factors (DE, DF)..............84

4-3 Comparison between F6 (p, e mode), F6 (p mode), and F4 ( DE, DF) tallies in
MCNP5, statistical uncertainty was less than 1%............... ...............84...

4-4 Comparison between F6 (p, e mode) and *F8 (p mode) tallies in MCNP5, statistical
uncertainty was less than 1%.............. ...............85...

4-5 Comparison between *F8 (p, e mode) and *F8 (p mode) tallies in MCNP5, statistical
uncertainty was less than 1% .............. ...............85...

4-6 Deterministic calculation will provide the angular information explicitly ................... .....87

4-7 For each energy group, using Monte Carlo radiation transport, a pencil beam of
mono-energetic photons interacts at the center of a water phantom ................. ...............88

4-8 Euler angles, zyz convention. Similar conventions can be obtained by following of
the following order in rotation (zyz, xyx, xzx, yzy, yxy)............... ...............89..

4-9 EDK-matrix after rotation to direction cosines A: (-0.5, 0.35, 0.8) and B: (0.25,-
0.40,-0.88) relative to the reference coordinate system (0, 0, 1) .............. ....................91

4-10 First octant" of S42 equal weight angles used for proj ecting the EDK-SN.................... .....93

4-11 A flow-chart for the EDK-SN Code System .............. ...............94....

4-12 External Source impinging on block of water ................. ...............95..............

4-13 External Source 3-D absorbed dose distribution for 2 Energy groups computed by
EDK-SN based on PENTRAN-MP A. EG(7.5-8.0Mev) and B. EG(0.5-1.0Mev) .........._.97

4-14 Energy group absorbed dose rate distributions (in Mev/ g.sec) along the X-axis
using the EDK-SN method based on PENTRAN-MP ......____ ... ......_ ................97

4-15 Energy group absorbed dose rate distributions (Mev/ g.sec) along the y-axis using
the EDK-SN method based on PENTRAN-MP .............. ...............98....

4-16 EDK-SN Total absorbed dose rate distributions along the X-axis compared to MCNP
*F8 tally derived absorbed doses ................. ...............98......___ ...

4-17 Total absorbed dose rate distributions of EDK-SN along the Y-axis compared to
M CNP *F 8 tally ................. ...............99................










5-1 Three Monte Carlo phantoms used in the ab sorbed dose study ................. ................. .1 04

5-2 Monte Carlo derived total absorbed dose rate distributions along the X-axis
computed using *F8 MCNP5 tally for three different phantoms .............. ..................105

5-3 Monte Carlo derived total absorbed dose rate distributions along the X-axis
computed using *F8 MCNP5 tally for three different phantoms .............. ..................106

5-4 Monte Carlo derived total absorbed dose rate distributions along the X-axis
computed using *F8 MCNP5 tally for the _SBLLBSS_ phantom ................. ...............106

5-5 Three SN phantoms equivalent to the one used in the Monte Carlo simulation. .............107

5-6 EDK-SN heterogeneous SBLLB SS phantom total absorbed dose rate distributions
along the X-axis compared to MCNP *F8 tally............... ...............108.

5-7 EDK-SN Homogenous phantom total absorbed dose rate distributions along the X-
axis compared to MCNP *F8 tally............... ...............109.

5-8 Continuous slowing down approximation range (Rcsda) for electrons in Cortical
Bone, Soft Tissue and Lung as a function of energy ................. ......... ................110O

5-9 Continuous slowing down approximation range (RCSDA) ratio comparison for
electrons in Cortical Bone, Soft Tissue and Lung as a function of energy ................... ...11 1

5-10 Frontal views of 3-D rendering of the 50th percentile UFH-NURBS 15-year male
and female ................. ...............112................

5-11 Simulation methodology for EDK-SN computations using PENTRAN-MP code
system ........... ..... ._ ...............113...

5-12 Absorbed dose contribution of each energy group is calculated via the EDK-SN
procedure, then folded to calculate the total absorbed dose delivered to the 15 year
old male phantom from 8 MeV flat weighted source ...........__.......__ ..............115

A-1 Total mass attenuation coefficient for photons in water, indicating the contributions
associated with phoelectric absorption, Compton scattering and electron-positron
pair production ................. ...............124................

A-2 Sketch of Compton scattering, photon of energy hv collide with an electron. The
photon' s incident forward momentum is hvl/c, where c is the speed of light in
vacuum ................. ...............125._._._.......

A-3 Compton effects differential cross section per unite angle for the number of electrons
scattered in the direction 8e............... ...............128..

A-4 Pair Production in the Coulomb force field of the nucleus. ..........._... ........_.._.........130











A-5 Pair Production in the Coulomb force field of the nucleus. ........._. ..... ..._._...........132









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

3D DETERMINISTIC RADIATION TRANSPORT FOR DOSE
COMPUTATIONS INT CLINTICAL PROCEDURES

By

Ahmad Al-Basheer

August 2008

Chair: Glenn Sjoden
Maj or: Nuclear Engineering Sciences

The main goal of this dissertation was to establish the feasibility of basing megavoltage

external photon beam absorbed dose calculations in voxelized phantoms on SN deterministic

calculations and pre-calculated electron absorbed dose kernels derived from full-physics Monte

Carlo. The SN derived electron absorbed dose kernel method EDK-SN, developed as part of this

research, achieves total execution times that are on the order of several times to orders of

magnitude faster than conventional full-physics Monte Carlo electron transport methods

considering equivalently detailed models and data Eidelity.

With the rapid movement toward intensity modulated radiation therapy (IMRT), radiation

beam intensities have increased dramatically over the past decade, thus heightening the need for

further characterization of out-of-Hield organ absorbed doses, along with their associated

biological risks. Assessment of these tissue absorbed doses is complicated by two fundamental

limitations. First, anatomic information on the patient is generally restricted to a partial body CT

image acquired for treatment planning; consequently, whole-body computational phantoms must

be employed to provide the out-of-field anatomy model structure for absorbed dose evaluation.

Second, existing methods based on Monte Carlo radiation transport, even with the application

significant variance reduction, are quite computationally inefficient at large distances from the










primary beam, and point-kernel methods do not properly handle tissue inhomogeneities.

Moreover, since absorbed dose are generally tracked in all major organs in the body, variance

reduction schemes for Monte Carlo are not all effective in this regard.

The outcome of this dissertation is to demonstrate that absorbed dose from high-energy

external beams radiation can be accurately computed for whole body and organ-specific

absorbed doses. The EDK- SN method implements voxelized phantoms with discrete ordinates

(SN) transport computations coupled with directional influences_and (pre-computed) full-physics

Monte Carlo based electron absorbed dose kernels to yield total body absorbed dose information.

This research shows that the deterministic techniques coupled with Monte Carlo based electron

absorbed dose-kernels has significant potential for organ absorbed dose evaluation in the clinical

management of radiation therapy patients.









CHAPTER 1
INTRODUCTION AND LITERATURE REVIEW

1.1 Overview

With the growing implementation of radiation applied in different therapeutic and

diagnostic medical procedures, comprehensive and accurate absorbed dose computations are

fundamental to precisely understand the impact and outcome of such procedures. In practice, the

clinical outcome in radiotherapy can be significantly improved with enhancements provided by a

treatment planning systems (TPS) Current TPS have been based on augmenting clinically

measured absorbed dose distributions. Such methods yield only approximate solutions, and are

limited in rendering a precise solution. Detailed solutions are achievable by solving the

Boltzmann transport equation, which describe the behavior of neutral particle radiation transport

throughout the phase space of a problem. Often absorbed dose calculation systems employ

relatively simplistic yet fast algorithms to predict the patient absorbed dose distribution from

"first principles," using simplified radiation transport models. Overall, these techniques are not

capable of accurately describing high fidelity whole body detailed radiation transport through the

human body, particularly as new procedures in medical physics are developed.

In recent years, Monte Carlo techniques have proven to be capable of representing very

complex geometries in a rigorous manner. This gave the Monte Carlo technique a great

advantage over previous methods in rendering accurate estimates of tissue absorbed doses,

especially when simulating human anatomy and yielded accurate prediction of absorbed dose

delivered in various medical physics procedures. However, this accuracy is achieved at a high

cost: extremely long running times are typically necessary to achieve a statistically precise

solution. This limits the practical use of Monte Carlo simulations for clinical procedures even

with High Performance/Parallel Computing (HPC) 3









Solving the Boltzmann equation directly using deterministic techniques can be

significantly more efficient than using the Monte Carlo methods, and therefore it can be used to

improve the quality of radiotherapy planning. Relative to the amount of information generated,

and with the use of HPC resources, such techniques usually require significantly lower running

times (relatively to Monte Carlo) to provide detailed and precise solutions. Moreover, the rapid

proliferation of Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) enable

human anatomy to be represented using regular images consisting of rectangularly (hexahedral)

shaped voxels in a regular matrix. These new developments make Cartesian geometry

deterministic techniques ideal for medical physics applications, especially given the demand for

high resolution absorbed dose distributions.

This research is focused on the development of EDK-SN, a first-of-a-kind capability to

assess organ absorbed doses incurred in radiation therapy (RT) applications using anatomical

patient models coupled with 3-D deterministic discrete ordinates (SN) radiation transport

methods. Absorbed doses delivered as a result of secondary charged particles will be linked

using electron kernels to account for secondary electron effects. This methodology is proposed

for regions in the phase space where Charged Particle Equilibrium (CPE) is not satisfied, and

subsequent absorbed dose estimation using a collisional kerma is not valid.

1.2 Initial Clinical Dose Evaluations

In the field of radiation therapy, absorbed dose distributions were historically first

approximated without doing any explicit radiation transport at all, but rather by extrapolating

from measured results taken for a particular energy, field size, and source-to-surface distance

(SSD) on a given machine. An extrapolation was performed by modifying the absorbed dose

that was experimentally determined under the reference parameters by ratios representing the

change between specific parameters and the parameters of interest. For example, if absorbed









dose values were measured in water, a table of the percentage of the maximum dose as a function

of depth (on-axis) in the water could be deduced. A similar table could be generated for off-axis

ratios (OARS) by measuring lateral profiles at a variety of depths. The absorbed dose at a given

point P in the patient could then be estimated by knowledge of the maximum dose for the

treatment, and applying a percent depth dose (PDD) correction for the depth of P in the patient,

and an off-axis correction for the distance to P from the axis at that depth4

Tables of PDDs were made available for various sources (incident energies), field sizes,

depths in the phantom, and SSDs. To modify the dose for an SSD that was different from the

reference SSD, the ratio of inverse distances squared would be applied. For a change in field

size, tables of field-size-factors would be subsequently applied. To compensate for the curved

surfaces of a patient, as opposed to the flat surface of the phantom, isodose lines could be shifted

along rays proj ected from the source.

It is important to point out that these techniques are often quite successful for predicting

absorbed dose in a homogeneous volume. In fact, in some cases they may be superior to

alternative simplified methods (for homogeneous volumes) because they are based on

measurements from a particular machine; thus, they circumvent the problem of how to model the

photon distributions emerging from the source and incident on the patient/phantom, which can be

a significant obstacle in accomplishing computational radiation transport based models.

The real challenge for these empirical-based methods, then, lies in correcting the

homogeneous water absorbed dose distribution to account for the many inhomogeneities present

in the human body, including air cavities, bone, dense muscle mass, and sparse lung tissue.

One way of addressing this problem is to apply "equivalent-thickness corrections" to the

water measurements. These corrections were based on compressing or stretching isodose lines by









factors that were derived from the relative photon/electron density in the path through tissue as

compared to that in water4'5. These methods, at their best, were useful for obtaining depth dose

information in the center of a broad beam, for a medium consisting of homogeneous layers. Even

then, these approaches do not account for the fact that photons interact not only with electrons,

but also with the nucleus. For example, the elastic scattering cross section depends on the atomic

number of the material. The clinical implication of this is that not all materials behave like

scaled-density water; real bone, for example, scatters photons to greater extent than water of

equivalent bone density due to the presence of Ca and other high Z elements not found in liquid

water. The effect of this was better addressed in the "Pencil Beam" method.

1.3 Pencil Beam Transport.

The Pencil Beam method sums the dose distribution from individual small diameter rays

called pencil beams. A pencil beam is made up of particles which pass through a differential

cross sectional area, 3x~y The off-axis dependence of the dose distribution for each pencil

beam is described by thick-target multiple scattering'. The on-axis dose information is obtained

from measured depth-dose data.

The calculation of absorbed dose assumes that the patient or phantom can be represented

by a stack of slabs of different material types. Each individual slab is homogeneous and infinite

in lateral extent. Assuming a single pencil beam has normal incidence on such geometry, the

absorbed dose at a position (x, y, z) can be expressed as

D(x, y, z) = f (x, y, z)g(z) (1-1)

where the first factor is the off-axis term, and the second factor is the central-axis term. The off-

axis term may be represented by a Gaussian of the form










1 -(x2 2)
f (x, y, z) = 2exp( 2 )(1-2)


where o, is the angular spread about the axis including the penumbra. The on-axis term may be

expressed in terms of its relationship to the depth dose in matter, go(z4f)


g~z)= gozz (1-3)


where SSD is the source-to-surface distance, and ze, is the effective depth in water, assuming

the linear stopping power of the material at depth : is relatively independent of electron energy

for normal body tissue.

The quantity go(z4f) can be related directly to a depth dose curve measured in a water

phantom, assuming a uniform incident beam, by the Eq. 1-4

go (z)I~xX (x x )2 + y_ y 2 L d 14
Do (x, y, z) = 2 x x'y 14


where the limits of the integral depend on the dispersion of the field at depth z.

An example of pencil method used for clinical photon radiotherapy is an application of the

Ahnesjo pencil beam model6. In this method, combination of measured data and Monte Carlo

calculated convolution kernels are used to characterize clinical photon beams and eventually

calculate the absorbed dose. The strength of the Ahnesjo pencil beam method is that it is fast

enough for clinical use, on the order of minutes. It also uses measured data for input that may be

specific to the treatment unit. However, because this method applies simple corrections using

pencil beams in homogenous matter, it is not accurate as a complete coupled charged-particle-

photon Monte Carlo simulation in heterogeneous media. Moreover, it does not fully represent the

directional component of the scattered radiation.









Aside from the geometric assumptions of infinite homogeneous slabs, the pencil beam

model does not handle changes in scatter from lateral heterogeneities; deviation of up to 5% can

be expected for low energy photon beams, and 14% for high energy photon beams8,9

1.4 Solving the Boltzmann Equation

At the most fundamental level, radiation transport simulation involves determining the

distribution of the particles (in physical space, direction, time, and energy) after they have

traversed some distance into a medium. Formally, a way to begin this discussion is by writing

the general equation for particle transport from kinetic theory of gases. This theory does not

explicitly treat individual particles, but rather describes the aggregate behavior of a continuous

particle distribution in a statistical manner. Let n be the expected number of particles within a

differential volume element d3r at position r with a velocity (which is directly related to its

kinetic energy) that is within a differential velocity d3v, at time t. Then, if a general solution is

found for the "particle density function" n(r, v, t) in phase space, then the transport problem is

solved.

When transporting the radiation through an arbitrary volume V, bounded with a surface

area S, the rate of change of number of particles is due to flow out through the surface of V,

collision events within V, or sources (scattering and independent sources) of particles located

inside of V. (all the terms of rate of change; commonly the first two terms are losses and the last

term is production)





chan ge in )= to flow + to collisionsI + to sources
nat time t houhSin V in V









This balance of neutral particles is cast as the linear Boltzmann Transport Equation (BTE).

The BTE is an integro-differential equation which describes the behavior of neutral particles in

terms of spatial, angular, and energy domains as they interact in a system; the steady-state

transport equation is given in Equation (1-5):


dE 'WrR dn' o-, (,',E 'a) Yr, ', ')qE,0E (15

0 4xr

where y(r7, OE)d'rdEdai is the angular flux with energy between E an~d E dE, at position r

within the volume element dF~, and in direction aZ within the solid angle daZ. The scattering

term, defined with the double integral term, represents the sum of the particles scattered into

dEdaZ from all the dE 'dnZ' after scattering collisions represented by the double differential

cross-secti on Gs (F,ER 4 E, iZ' 4 6) q (r-, 6,E ) represent the external source.

The left side of Equation (1-5) represents streaming and collision terms (loss), and the right

side represents scattering and independent sources (gain). Since it describes the flow of radiation

in a 3-D geometry with angular and energy dependence, this is one of the most challenging

equations to solve in terms of complexity and model size. For rendering a deterministic

computational solution for a large problem in a reasonable time, it is necessary to utilize a robust

parallel transport algorithm and a high performance computing system. All the methods

described below are different approaches to solving this equation. For cases with charged particle

transport, the Boltzmann and Fokker-Plank formalisms (Boltzmann-FP) 10are generally used. A

simplified form of this equation is usually used, which is:









R V~(rR, E)+ o(r, E )~(rR, E)


dE dn' o- (r,0'- OE'4 E) YlI r, O',E')
0 4xr (1-6)
8[S (r, i',E)Y(r, 0',E ')]
+q(r, 1, E)
dE

However, the full Boltzmann-FP equation is extremely challenging to solve

deterministically .

1.4.1 Monte Carlo Method

Monte Carlo techniques are a widely used class of computational algorithms for the

purpose of statistically modeling various physical and mathematical phenomena. In general,

individual probabilistic events are simulated sequentially, where probability distributions

governing particle interactions are statistically sampled to describe the total phenomenon. In

principle, for each case sampled, there are four required inputs; namely, a random number

generator, source characteristics, a simulation phase space, and probability distributions

governing the possible outcomes. Results of individual particle histories are generated using

pseudo-random numbers to sample the probability distributions. This process is repeated many

times, until the statistical precision required for the solution is achieved, based on the law of

Large Numbers and the Central Limit Theorem"l. Therefore, the Monte Carlo technique is an

effective method for indirectly solving the Boltzmann radiation transport equation via statistical

methods, and can readily treat the full physics of coupled photon-electron transport. Simulations

for particle histories are repeated many times for large number of histories, where the computer

code tracks the desired physical quantities, means, and uncertainties (variances) for all of the

simulated histories.









One of the most important issues limiting the use of Monte Carlo techniques for clinical

absorbed dose calculations is the long execution time required to reduce the statistical variance

associated with the calculated absorbed dose at each differential volume. At the same time, the

outcome of any radiotherapy procedure depends solely on the accuracy of the radiotherapy

treatment planning system, so that a small change in the absorbed dose delivered (A 5%) can

result in a dramatic change in local response of the tissue (f 20%)12,13


In order to reduce the statistical variance a, a ac associated with Monte Carlo, large


numbers of histories N are required, and consequently this leads to higher running time to yield

acceptable results. Improved variance reduction techniques may reduce the running time.

However, such techniques are to be used with caution, as improper application of statistical

variance reduction, method applied un-conservatively may cause wrong answers. Also, variance

reduction is often tied to one objective; this can be problematic when faced with determining a

dose at all locations in the phase space.

The increasing use of voxelized human body geometries has created challenges in many

aspects of applying Monte Carlo simulationS14. With increasing use of modern technologies such

as CT and MRI in vivo imaging, anatomical definition has shifted from using simplified surface

equations to segmented tomographic images in order to truly reflect the anatomy of a specific

human body. The segmented images are represented in the form of a large uniform 3-D matrix of

hexahedral voxels, where each voxel represents a constant material with a tag number (for

example, CT number). One digital image may contain billions of voxels, making it very

challenging be fully represented explicitly using Monte Carlo, particularly considering large

models size and number of tally sites, adding significant overhead to the computation. Some

special data management treatments are needed for Monte Carlo simulations to make









computation times more tractable, including voxel down-sampling and parallel computing

resources.

1.4.2 Deterministic Methods

Several different methods have been developed for numerically solving the Boltzmann

equation" ~l. In general these methods depend on discretizing the energy, angular, and spatial

domains of the Boltzmann transport equation.

1.4.2.1 Discrete ordinates method (SN method)

The most common deterministic approach currently used for photon transport is the

discrete ordinates method (SN method) developed in detail by Carlson over 50 years agol6. The

discrete ordinates equations are derived by discretizing the BTE over phase space elements (i.e.,

spatial-energy-direction variables) to construct discrete-variable cell particle balances. Equation

(1-7) is the standard Legendre expanded Cartesian multigroup integro-differential formulation of

the BTE, and x, y, z are the spatial variables along each axis in the Cartesian coordinates.






g= =1 =0 k=1 -I+k


[ 1~ r(x,y,z)cos(kp)+0b' (x,y,z)sin(kp) )+q (x,y,z,pi,p) (1-7)

Other variables in Equation (1-7) are:

pu, r, correspond to the direction cosines along each axis for each ordinate

Y, is the energy group angular flux

o-, is the total group macroscopic cross section

0-, a,,i~ is the lth moment macroscopic scattering cross section for group g to g









;,~ is the lth flux moment for group g


at~~,~1 and # ; are lth, kth associated cosine, sine flux moments for group g

P(pu) and Plk (pU) are the Legendre and Associated Legendre polynomials.

ql (x ,y z, p, cp)is independent source for group g


The SN methodology is a widely accepted deterministic scheme in yielding numerical

solution to neutral particle radiation transport problems, and can be applied to medical physics

problems, provided that discretization and truncation errors can be reduced to an acceptable level

over the problem space. Depending on the amount of data needed per floating point operation,

the SN method is more efficient than the Monte Carlo method for many problems, and with

proper discretization produces an accurate solution". The SN method is widely used in nuclear

engineering to obtain a numerical solution of the Boltzmann transport equation. The method

solves the BTE for a discrete number of ordinates (directions) with associated weightsl6. The

combination of discrete directions and weights is referred to as quadrature set l. A drawback of

the SN method is that with the limited number of directions involved, in certain situations, this

may lead to so called r~ay-e~ff~cts, which appear as unphysical flux oscillations. In general, this

behavior occurs for problems with highly angular dependent/localized sources and/or fluxes, or

when the source is localized in a small region of space, in low density or highly absorbent media.

This can make medical physics applications, which often require small mesh sizes, relatively

large models, and low density materials, suffer severe ray-effects.

The ray-effects can be alleviated naturally by the presence of a distributed isotropic source,

which tends to flatten the flux in the angular domain. However, in medical physics, the photon

transport problems may exhibit significant ray effects. This effect becomes worsen in a low









scattering medium such as air, or in a strongly absorbing medium such lead, where the photon

flux usually is typically strongly angular dependent. In such situations, either a higher order

quadrature set (many directions) is required or some ray-effects remedies must be appliednl~l

(such as Taylor Projection techniques and adaptive differencing schemeS19). More discussion on

discretization and ray-effects are presented in Chapter 2.

1.4.2.2 Monte carlo versus deterministic SN methods

The most fundamental difference between the Monte Carlo methods and the deterministic

SN transport method is that the SN method solves the transport equation for the average particle

behavior throughout the problem phase space, while the Monte Carlo method, the terms of the

transport equation are solved statistically by simulating the individual particles and recording

some aspects (tallies) of their average behavior, but only at specific tally locations requested by

the user.

In the discrete ordinates method, the phase space is divided into many small meshes, in

which the cross section is assumed to be constant. In the medical physics applications, the model

geometry, in this case, the patient is represented using regular voxel-based grid generated from a

segmented CT scan data or other digital imaging technique. The types of grids can then be easily

represented in an input deck for the SN calculations, making both Monte Carlo methods and

deterministic techniques capable of representing the model of study accurately.

Another advantage of the MC methods is that they can easily utilize continuous energy

interaction data, whereas deterministic calculations use multigroup or energy group averaged

data. However, photon attenuation coefficients in the energy range of interest for radiotherapy

(RT) are quite smooth with little structure (in contrast to neutron cross sections); and are

accurately represented by multigroup data. Recent research we conducted has shown that

multi-group deterministic methods produce highly accurate absorbed dose distributions in photon










transport calculations. Similar conclusions were drawn by Daskalov et. al and Williams et. al21-

23. Table (1-1) shows a comparison between Monte Carlo and Deterministic techniques24

Table 1-1. Comparison between Monte Carlo methods and deterministic techniques
Term Deterministic Monte Carlo,
1. Geometry Discrete! Gilobally Exact
Discretized
2. Energy treatment- cross section Discrete Exact
3. Direction Discrete!Truncated series Exact
4. Input preparation Varies Simple
5. Computer memory required Large mllrg
6. Computer time Sma'lll Large
7. Numerical issues Con vergen~ce Statistical uncertainty
8. Amount of information Large Limited
9. Parallel computing Complex Trivial



1.5 Dose and Charged Particle Equilibrium

Absorbed dose, D, can be defined as a non-statistical quantity in terms of a statistical

quantity, energy imparted s The absorbed dose to matter at a point P is given by

D = df /dm (1-8)

where r is the expectation value of E, the energy imparted by ionizing radiation to matter of

mass m in a finite volume y, and

E = R,,, Roz,, + eQ (1-9)

in which

R,,, = the radiant energy incident on the volume y,

Ro,, = the radiant energy emerging from V, and

ceQ= the net energy derived from rest mass within V (negative for the reverse process).

where the radiant energy R can be defined as the energy of particles (excluding rest energy)

emitted, transferred, or received










Collisional kerma, Kc can be defined in terms of the related net energy transferred, Sn ,26,27

and R25 as.

dE" /dm (1-10)

The energy-transferred in a volume Vis

Et;' = (R~n), (Rozer)"" on R + c Er, R', (1-11)

in which

(R~n), = the radiant energy of uncharged particles entering the volume y,

(Rozer) onr= the radiant energy of uncharged particles leaving V, except that which

originated from radiative losses of kinetic energy by charged particles while in V.

Rr = radiant energy emitted, where u indicates uncharged particles.

Charge particle equilibrium (CPE) can be established in a volume when the energy carried

into the volume V by charged particles is balanced by the energy carried out of the volume V by

the charged particles. Note that, CPE is not established for thin materials irradiated by photons.

The range of secondary electrons produced is larger than the material dimensions, so that much

of the electron energy is lost simply by electrons leaking out of the material.

Under the conditions of CPE, Attix demonstrated that when the energy imparted equals

energy transferred, r = 8,,, collisional kerma Kc is equal to the total dose Dar,2 Eq(11).




CPE~iP(-2











where 'F is the eonegy fluece~, and (ie": is the mass energy absorption coefficient for


7 ray energy E in medium Z. This result is very important, since it allows us to calculate the

measurable dose D based on the calculable quantity Kc -

1.5.1 Deterministic Dose Calculations for Diagnostic Application

For low energy beams, such as the one used in diagnostic applications, the concept of

charged particle equilibrium is of a great importance as a mean of equating absorbed dose D to

collisional kerma Kc -

For a volume V, containing a smaller volume v, irradiated with an indirectly ionizing

radiation field (i.e., Photons and Neutron fields), there are four basic causes for CPE failure in

volume v:

In-homogeneity of the atomic composition within volume V.

In-homogeneity of density in V.

Non-uniformity of the field on indirectly ionizing radiation in V.

Presence of a non- homogeneous electric or magnetic field in V.

These four situations will disrupt CPE by preventing charged particles of a given type and

energy leaving the volume v to be replaced by an identical particle of the same energy entering27

In practice, CPE conditions are met if the attenuation of the primary beam is negligible in

the maximum range of the secondary electrons generated by the beam. As indicated by Table (1-

2) for a layer of water of 5 cm thickness, this scenario is plausible in the limit of diagnostic

medical physics applications, where the energies fall within 10 to 200 keV, and CPE conditions

are met for water/tissue volumes less than 5x5x5 cm3, and consequently D is equal to Kc under

such conditions.










Table 1-2. Approximate attenuation of photons with a layer of water equal to the maximum
range of secondary charged particles28
Primary Radiation Energy Maximum Secondary Photon Attenuation (%i) inl
(MeV) Electrons Range (cm) Maximum Electron Range

0.1
0.01387 0.23%
1.0
0.43 3.00%
4.933 10.04%
9.305 15.%

1.5.2 High Energy Beam Dose Calculations

Based on the previous discussion, as the energy of the ionizing radiation increases, the

penetration power of the secondary charged particles increases more rapidly than the penetration

power of the primary radiation in a given volume. This leads to CPE failure, caused by the small

number of charged particles produced at deeper depths in volume V, relative to the number of

charged particle produced at initial points along the primary radiation field direction. This is

indicative of significant attenuation of the primary indirectly ionizing radiation.

As a result of this phenomenon, higher numbers of charged particles will be generated

close to the entrance compared to the rest of the volume, leading to CPE failure considering dose

in that volume. The degree of this failure becomes progressively large for higher energies, so that

D no longer equals to Kc A more detailed discussion on this topic is presented in chapter 4.

1.5.3 Limits of Charged Particle Equilibrium

Assuming that a maximum energy is transferred from a photon to an electron, we can use

this assumption to investigate a CPE requirement for a water phantom region, as defined earlier.

Considering the continuous slowing down approximation range of the electron (REsn), and

average range of the maximum energy electrons as function of energy.



Ross =IT dTL (1-13)











o 1 tt (dt(1-14)

j: dN (t )dtd No 0


where

dT
-is the electron stopping power (MeV cm2/g
p dxC

To is the starting energy of the electron (MeV)

dN (t )
tf (t) = is the differential distribution of farthest depth of penetration


N (t ) is the number of particles penetrating a slab of thickness t

Through inspection of these relationships described in Figures 1-1 and 1-2, one can

decide what kind of tolerance is acceptable in applying a collisional kerma approximation for a

dose calculation based on a particular voxel size, i.e. to determine CPE approximation limits.

This also has been demonstrated in Table 1-3; for example at 2 MeV the CPE computed dose

will be accurate to approximately 5% using a 1 cm voxel, assuming the CSDA range applies.

Table 1-3. Approximate thickness of water needed to achieve various attenuation amounts
compared to maximum energy electron ranges generated by the same beam

P hotona Range (cm) at which CPE is Average" RewA
Energy conserved by e- range
(Mle 1 2%0 3% 5% (cm) (Cm')
0.3 0.085 0.17 0.256 0.431 0.056 0.085
0.6 0.112 0.225 0.34 0.572 0.15 0.225
1.0 0.142 0.286 0.43 0.726 0.286 0.43
2.0 0.203 0.41 0.617 1.04 0.617 1.04





Maximum electron energy CSDA
range in liquid water


1.00E+00


1 .00E-01

-*RCSDA range of max. energy e-
1.00E-02
Thickness where 1% att.
Y ~ IThickness where 2% att.
S1.00E-03
_c: Thickness where 3% att.
-*-Thickness where 5% att
1.00E-04
1.00E-02 1.00E-01 1.00E+00 1.00E+01
Energy(MeV)

-1. Comparison between thicknesses required to attenuate different percentages of
primary photons and Rcsn range for maximum energy secondary electrons
produced by the same beam


i~-


-*-Ave.range of rnax. energy e-
-.Range where% atten.

Range where 2% atten.
Range where 3% atten.
- Range where 5% atten.


Figure 1-2. Comparison between thicknesses required to attenuate different percentage of
primary photons and average range for maximum energy secondary electrons
produced by the same beam

With proper model discretization of the angle-energy-spatial transport phase space, and at

low photon energies yielding primary and secondary electron interactions with small electron


Figure 1


Average range of maximum energy
electron in liquid water


EE


E 1.00E-03
a
2 1.00E-04



o 1.0E0
cn0


0OE-02


1.00E-01 1.00E+00
Energy (MeV)


1.00E+01









transport paths relative to a spatially discretized mesh grid interval, 3-D SN methods can be

implemented to directly yield very accurate dose distributions. Therefore, the deterministically

derived dose using a kerma approximation will be accurate up to the range of electron path

lengths comparable to ranges smaller than the photon spatial SN mesh grid interval in anatomical

models (typically representative of the anatomical data voxel size).

As mentioned earlier, at higher photon energies, where electron ranges exceed the SN

mesh grid interval, this is not the case, and interactions yielding energetic electrons must account

for the mechanism of electron transport through the problem phase space to deliver an absorbed

dose after representative electron transport distances. This work will set the foundation for state

of the art 3-D absorbed dose computations for high energy external beams to account for

interactions yielding energetic electrons. The EDK-SN methodology will account for electron

transport in the phantom by applying a pre-computed electron absorbed dose kernel calculated to

a very low variance

It is worth comparing the EDK-SN methodology, proposed here, with the traditional

convolution-superposition technique, which applies a database of "energy deposition kernels,"

calculated using Monte Carlo techniques to determine a absorbed dose29,30. We note that these

kernels can be used to describe the spread of energy about the primary photon's point of

interaction in the phantom, however, these have limited use since the convolution-superposition

technique is performed without regard to the incident photon angular component. While, in

contrast, the EDK-SN procedure is based on a detailed, full physics Boltzmann transport solution

with direction and flux information directly available for coupling with the EDKs along a

propagated photon direction of travel.









1.6 Outline of This Research

The method developed in this dissertation, EDK-SN, includes implementing voxelized

phantoms with discrete ordinates (SN) transport computations coupled with multi-directional and

(pre-computed) electron absorbed dose kernels (EDK). Particular care was taken when

generating the EDKs using full-physics Monte Carlo calculations (with the MCNP5 code) and

tallying the results using pulse height tally (*F8) tallies in MCNP5 to extremely low statistical

uncertainties. The application of the pre-computed EDKs was based on the photon flux and net

current vector for the "dose driving voxel" (DDV) calculated via the discrete ordinates parallel

code PENTRAN with a voxelized phantom. Chapter 2 introduces code systems, and presents is a

study of critical discretization issues in 3-D discrete ordinates medical physics simulations

benchmarked with the Monte Carlo. The discretization of the phase space variables in the

transport equation will be discussed, along with the absorbed dose calculation comparisons.

Chapter 3 discusses whole body dosimetry simulations with the UF-B series phantoms. It also

presents the creation of the PENTRAN-MP code system. Chapter 4 elaborates further on CPE,

and compares absorbed dose calculations using different absorbed dose tallies in the MCNP5

code. It also introduces the EDK-SN methodology and the parallel strategies. In Chapter 5, we

apply the new EDK-SN methodology to heterogeneous phantoms, including the whole body 15

year old phantom and compare the results to the full physics (Monte Carlo MCNP5 *F8) tally.

Chapter 6 summarizes the research and draws conclusions on the objectives accomplished;

aspects for future development are also highlighted.









CHAPTER 2
CRITICAL DISCRETIZATION ISSUES INT 3-D DISCRETE ORDINATES MEDICAL
PHYSICS SIMULATION

Discretization of the BTE must be well understood in solving this equation for medical

physics problems. As discussed, the fundamental difference between the SN and Monte Carlo

methods is that the SN method directly solves the BTE and requires phase space discretization,

while the Monte Carlo method applies a statistical approach to represent the radiation transport.

It is important to explore discretization issues associated with the SN method. To account for

discretization effects, 3-D flux distributions were computed in a water phantom exposed to 60C0

y-rays using the PENTRAN parallel SN code and compared to analogous models implemented

using the continuous energy MCNP531 Monte Carlo results with ENDF/B-VI data libraries. A

major issue affecting the accuracy of the SN method is selecting an appropriate discrete set of

directions, or the "angular quadrature set" for which the transport equation is to be solved.

Choosing a quadrature set that does not include enough discrete directions to represent the

physics can lead to "ray-effects." The ray-effects are numerical artifacts sometimes exhibit in the

SN computations when there is high radiation streaming and little scattering. With a finite

number of directions considered in each fine mesh cell for the radiation emanating at points

distal from the source; it often results in the "spines" at points distal from the source. In this

work, a number of studies have been conducted through comparison with the Monte Carlo

simulations to determine suitable quadrature sets required to reasonably limit the ray-effects in a

proposed radiotherapy model that includes radiation transport in air to still yield high accuracy.

Another set of simulations were accomplished to address another difficulty in performing the SN

simulations with regard to spatial discretization. Selection of an adequate 3-D spatial mesh size

to represent the different geometric features of the model is also critical; at the same time,

selection of a differencing scheme capable of representing the physics within tractable memory









storage and time limitations is also a must. This issue has been investigated, particularly in the

low density portions of the model (mostly in air). Finally, we compare the performance achieved

with three cross section libraries available from the nuclear engineering and physics

communities. These multi-group gamma-ray libraries were derived from the BUGLE-

96321ibrary, the CEPXS package33, and the SCALES package.

2.1 Code Systems

PENTRAN (Parallel Environment Neutral particle Transport) is a 3-D discrete ordinates

code that solves the Boltzmann transport equation on distributed parallel computers. It is capable

of complete, hybrid phase space decomposition with up to a 97% parallel fractionl9,34 I

executing a parallel transport calculation on each distributed node, PENTRAN performs discrete

ordinates sweeps only on the local portion of the phase space assigned. As problem sizes

increase, more machines may be needed to scale up to accommodate the total parallel memory

demand. The inherent parallel scalability of PENTRAN with advanced numerical treatments

(including block adaptive mesh structures and adaptive differencing schemes) makes it ideal to

apply to challenging problems in medical physics. The PENTRAN code is part of a code system

that includes pre- and post-processors, and other utilities to manage large parallel datasets.

For the purpose of comparison to the Monte Carlo methods, the well known MCNP5

Monte Carlo code developed by Los Alamos National Laboratory is used to simulate a 60C0

beam for an equivalent unit. MCNP5 is used to calculate the scalar flux along the central axis of

the model. For these studies, cell flux mesh-tally (F4) and energy deposition volume tallies (F6)

are defined in the Monte Carlo model in a way that is analogous to the Cartesian fine mesh cells

used in the PENTRAN.









2.2 Simulation Geometry

Because the purpose of our investigations was to identify critical discretization issues, the

models we considered were simplified to enable us to focus on proper numerical discretization.

2.2.1 PENTRAN Model

We considered a box of lead containing a 60C0 source and a water tank for use in multiple

benchmarking applications (Figure 2-1). As we mentioned earlier, this model consisted of four

key components; a 60C0 source (2x2x2 cm3), a lead source housing 4 cm thick, an air layer, and

a homogeneous water phantom (30x30x30 cm3) at a source-to-surface distance (SSD) of 34 cm.

A 3-D spatial distribution was generated by the PENMSHXP35 meSh generating code in the

PENTRAN system. The model was divided into 5 z-levels at 0 cm, 30 cm, 60 cm, 64 cm, and 66

cm; each z-level was divided equally into 9 "coarse" meshes containing various numbers of

"fine" spatial mesh cells; a total of 126,000 fine mesh cells were used in the SN problem.














2000000 Coacco000




0 15 1 fl
"1e 30 30 c


Figure 2-1. 60C0 unit (left: Mesh distribution generated by PENTRAN package; Right: MCNP5
Cell/Surface schematic input)









2.2.2 Monte Carlo Model

The MCNP5 Monte Carlo model geometry was defined to be equivalent to that used in the

PENTRAN model. The 60C0 source was distributed isotropically over multiple equi-probable

sources cells in one energy group. Source spectra were defined in a consistent manner with the

various multi-group libraries considered. Volumetric cell flux F4 mesh-tallies and energy

deposited F6 tallies were used in the Monte Carlo geometry description that was equivalent to

the discretized SN volumes. The MCNP5 F6 tallies were used along the central axis of the model

across the lead box, air region, and water phantom. All simulations were performed using a PC

cluster composed of 16 Dual Intel Xeon processors at 2.4 GHz with 2 Gb/processor, for a total

memory of 32 Gb. MCNP5 tallies along the central axis were obtained using variance reduction

techniques. The cell fluxes converged, on average, to within 3 % for the reference case.

2.3 Comparison of Deterministic and Monte Carlo Results

Calculations were performed to compare the SN and Monte Carlo results for the scalar

fluxes along the model central axis. Figure 2-2 depicts a 3-D scalar flux distribution computed by

PENTRAN for the reference case for different energy groups. The highest energy group (group

1) is forward peaked with some ray-effects along the central axis. However, as the radiation

down scattered to lower energy groups, it demonstrates more isotropic behavior evident in

energy groups 2 and 3. This simulation was based on the use of an S42 angular Legendre-

Chebychev quadrature (1848 directions/mesh) with P3 Scattering anisotropy using the CEPXS

library. This highlights the fidelity of the information available in a converged deterministic SN

computation.

2.3.1 Investigation of Angular Quadrature

Medical physics applications represented in SN models often require small mesh sizes,

relatively large models, and low density materials. These problems can lead to severe ray-effects,









which appear as unphysical flux oscillations in the scalar flux distribution unless treated

properly. In general, this behavior occurs for problems with highly angular dependent sources

and/or fluxes, or when the source/sink is localized in a small region of space, in low density or

highly absorbent media.












Grou 1 rou 2 rup3Gru

Fiur 2-2 600yry rus12,3ad4saa xcmue yPNRNwt h
CEPXS gamm crs eto irr,(S2aglrqartr (88drcin)wt
P3i Scteigaiorp hr sd
Tomtgt this rbehg nua udaueoresaeotnnee.Fgr -
deit a sceai ofteodnt est itiuin soitdwt 1 0e-ymti

(rotatonall lee-ymtic) and S4 LeenreChbyhe quaraur sets on 1/to ui






thgue origi ando p-rojec toD e rus1 nd at thea squar "pont"plo ted inteFiue BeyndRA S20, Legede


Chbyhe quadratue sets polmus beg nused due t neative wreighs tate occur ine standrd lee-







symmetric sets36. Particular attention must be made in deriving quadrature sets to enable

preservation of Spherical harmonics (Legendre) scattering moments and numerical quadrature

orthogonality relationships.





S12 S42

Figure 2-3. S12 (Level Symmetric) and S42 (Legendre-Chebychev) quadrature first octants (with
21 and 231 directions/ octant respectively)

Using a uniform 1 cm mesh (for reasons discussed later) in the air region, PENTRAN was

used to calculate the gamma flux distributions in the system for several different quadrature sets.

PENTRAN provides the option of readily using any quadrature that could be prescribed by the

user. The flux distributions produced by an initial computation, using S12 leVel-Symmetric

quadrature with P3 Scattering anisotropy and 126,000 fine mesh cells~ is shown in Figure 2-4.

The ray-effects in the S12 COmputation are evident when compared to scalar fluxes in Figure 2-5

using S42 quadrature, with a similar mesh structure and P3 Scattering (the reference case). While

there is evidence of small ray effects in the reference case far from the source, the differences

between Figures 2-4 and 2-5 is dramatic. We also found that there was a small difference (on the

order of several percent, depending upon location) in flux solutions computed with S42 between

P3 and Pi computations, but little difference noted in the problems tested between P3 and P5

anisotropy, and so P3 Was implemented for all other computations. To further investigate the

effects of raising the quadrature order, unbiased quadrature sets were tested using S22, S32, etc, up

to S42 with P3 aniSotropy. A comparison of the PENTRAN computed fluxes in Group 1 using

selected quadratures and equivalent MCNP5 calculations are given in Figure 2-6.





60


N 40


20


~ "1
,o'~-~


~;51~~


Y~cm)


Ylcm)


0 cm


5cm


60


A 40


20


0


60


N 40


20


0


~~ "


Y (cm)


Y (cm)


10 cm 15cm

Figure 2-4. 60Co y-ray 3-D Group 1 scalar flux distribution for x-level depths at 0 cm, 5 cm, 10
cm and 15 cm, computed by PENTRAN with the CEPXS gamma cross section
library; an S12 angular quadrature (168 directions) with P3 Scattering anisotropy was
used as a reference case
























1
r15 O
V(c,! O


1
I-15 o
Y(cm) O


Ocm 5cm





60 I. 60


40 40


20~ --: I 20


0 34 0 30
30 15 0 L 30 1
Y1e 0 0e


10 cm 15 cm

Figure 2-5. 60Co y-ray 3-D Group 1 scalar flux distribution for x-level depths at 0 cm, 5 cm, 10
cm and 15 cm, computed by PENTRAN with the CEPXS gamma cross section
library; an S42 angular quadrature (1848 directions) with P3 Scattering anisotropy











As shown in the Figure 2-6, the PENTRAN results with S42 Legendre-Chebychev


quadratures are in good agreement with the MCNP5 results, and differences between the two are

within the statistical uncertainty of the Monte Carlo in air and the water phantom distal from the


source. Overall, this figure shows that increasing the angular quadrature order results in a


progressive improvement in solution accuracy, and that a substantial number of directions are


required for proper representation in air region between the source and the water phantom. We

further note that while the S12 leVel-Symmetric quadrature is quite adequate to represent neutron


scattering in nuclear engineering neutronics and reactor physics applications, it does not produce


accurate results here. For a general problem with a high degree of angular dependence, the ray-


effects must be largely mitigated by employing a large number of directions, i.e., a high-order


angular quadrature, From our analysis, a minimum of S32 Legendre-Chebychev quadratures

should be considered. Investigation of alternative, less computationally intensive methods for


mitigating the ray-effects in problems of this type has shown that ordinate biasing methodologies

can also be ofbenefitl9


1.OOE+10-


1.OOE+08-
-HSn22
Sn32
-MSn42
1.OOE+06 MMCNP5


1.OOE+04-


1.OOE+02-


1.OOE I O
O.OOE+OO 3.OOE+01 6.OOE+01 9.OOE+01
Distance (cm)


Figure 2-6. Scalar flux distributions using S12, S22, S32, and S42 with P3 Scattering anisotropy in
PENTRAN using the BUGLE-96 gamma library









2.3.2 Investigation of Spatial Discretization (Meshing)

In addition to proper resolution of the angular variable, the spatial variable must be

appropriately discretized to represent the materials and geometry of the system to yield the

solution to the BTE. In PENTRAN, a block adaptive Cartesian three-dimensional hexahedral

mesh arrangement is used to define the spatial-material distribution of the system, i.e. the system

is partitioned into coarse mesh blocks containing fine mesh cells having a uniform cross-section

in each fine mesh cell. The solutions to the SN equations are obtained by sweeping along each

cell and discrete direction on the unit sphere; this process is usually referred to as a "transport

sweep" ". In discretizing a spatial mesh, it is often necessary to examine, using successive

computations, the effect of the coupling between angle, mesh size, cross section, and the spatial

differencing scheme used to solve SN equations. Here, we will briefly highlight how spatial

differencing schemes are applied in the SN method.

Equation (1-6), is integrated over the volume ~,],k = 7x ]y kz to yield volume and

surface averaged flux terms. Rearranging, we can obtain the zeroth moment balance equation

(omitting group sub scripts for brevity):



Ax Ay Az

Equation (2-1) contains seven unknowns. Three of which, entrant fluxes ("in") are

assumed to be known. Also, the collective cell averaged angular volumetric source q, (which

includes scatter and fixed sources) is assumed to be known from previous source iteration.

Therefore, the average angular flux YA. and the exiting ("out") surface angular fluxes are

unknowns, where these values are derived using a set of "auxiliary equations" which enable









coupling of the average and boundary angular fluxes. Linear weighted schemes use the auxiliary

formulations in Equation (2-2):


Your 1 nx~a1))(2-2a)


Yo1 (;,cY.yb 1)) (2-2b)



Yon, n=(c- (2-2c)

The standard Diamond Differencing (DD) scheme results when a=%/, b=%/, and c=%/ in

equations (2-2a, 2-2b, and 2-2c); the DD scheme is second order accurate, but may lead to

negative solutionsll. Petrovic and Haghighat developed the Directional Theta-Weighted scheme

DTW37 where a can be expressed as




qA + mis+(;; iy m in
Ax Ay z


Ay Az mo~ix

Similar weights can be derived along the y- and :-axes for "b" and "c", respectively. DTW

can remedy the unphysical oscillations inherent in solutions rendered by the DD scheme, and it is

always positive. In cases of high angular streaming, exponential auxiliary equations may be

needed. These are implemented into PENTRAN to represent streaming behavior of the angular

flux across the spatial cell. The exponential Directional Weighted (EDW)38,39 can be employed in

this situation with the following auxiliary formulation:


Y,;;(x, y, z) =acexp(ilP4(x)/|p~, ,|)exp(Aglp(y)/|q ,|)exp(ilPkz~ n ~1
(2-4)









Multiple differencing schemes in PENTRAN have an adaptive structure between

successive schemes of increasing accuracy within a coarse mesh containing many fine meshes

called "adaptive differencing strategy (ADS)": if one begins with the Diamond scheme

(Diamond Zero (DZ)), the detection of a negative flux-fix-up causes selection of the Directional

Theta-Weighted (DTW) scheme, and if weights (as in Equation 2-3) are beyond certain values,

selection of the Exponential Directional Weighted (EDW) scheme is made. In recent PENTRAN

versions selection of the Exponential Directional Iterative (EDI) method is made as an

improvement over EDW. This ADS enables the code to automatically select the most appropriate

differencing scheme within a particular coarse mesh. We examined the effect of different sizes of

mesh cells with various differencing schemes in our model. Figure 2-7 depicts the fourth z-level

mesh structure (plotted at hexahedral cell centers) relative to the full-size model; note that

sparser meshing was used in the surrounding air region.



















Figure 2-7. 60C0 model spatial meshing, axial level 4

The size of the fine mesh cells in the air region of the problem was of particular

importance due to low density; smaller meshes yield high accuracy, but can lead to more

pronounced ray effects; larger meshes mitigate the ray-effects, but may be inaccurate. Figure 2-8









shows a comparison between Eine meshes side size of 1 cm, 2 cm and 5 cm in the air region with

the DTW scheme fixed in air. These results demonstrate that a 1-cm mesh in air is necessary to

maintain accuracy. Based on these observations, we designed our PENTRAN model as follows;

the model is 30x30x70 cm3, and consists of 45 total coarse mesh blocks arranged among 5 axial

levels, containing a total of 126,000 Eine meshes. In recent PENTRAN versions, selection of the

Exponential Directional Iterative (EDI) method is made as an improvement over EDW39. The

geometry of the model has been represented using the block adaptive meshing capability of the

PENTRAN code that allows the user to use a higher mesh density in the regions of interest.

Axial level thicknesses were determined based on major components of our model, and the X-Y

spatial dimensions of coarse meshes (containing various densities of Eine meshes) were

uniformly 10 cm on each side. The DTW differencing scheme was locked for the regions with

low density material (air regions), and the adaptive differencing scheme (with adaptive upgrade

potentially up to the EDW differencing scheme) was prescribed for higher material density

regions. A mesh coupling scheme, Taylor Proj section Mesh Coupling (TPMC)40, was used to link

discontinuous meshes using first order projections as radiation sweeps through the problem. The

Cartesian mesh sizes for the full size model were selected to preserve component volumes, also

taking into account the optical thickness of each material. The Eine-mesh intervals in different

problem regions are: Source- 0.5 cm; lead- 0.5 cm; water- 1.0 cm; Air- 1.0 cm. Smaller mesh

sizes better represent curved boundaries and irregular structure in the model, but require larger

memory and may lead to longer running time (depending upon the spectral radius of the

equations). Smaller mesh sizes were selected for the source region of the problem, where the

particle flux has a strong gradient. Mesh sizes in the air can be made larger near the edges of the

problem where a more detailed solution is not required.













1.00E+10


1.00E08 --*-5m Air mesh
-*2cm Air mesh
-*-1cm Air mesh
1.00E+06 -I ---MCNP5

L..
1.00E+04


1.00E+02


1.00E+00
0.00 30.00 60.00 90.00
Distance (cmr)


Figure 2-8. Scalar flux comparison of variation in the interval of 3-D mesh cells containing air;
fine meshes of 1 cm size (along each axis) in air proved to be best in providing high
detail flux distribution

Using the ADS described earlier with different mesh sizes, we were able to achieve a

converged solution. Figure 2-8 depicts the solution behavior versus mesh size in the air region


(even though there is very little optical thickness or attenuation by the air); this can be explained

by the fact that larger mesh sizes in the air do not adequately propagate the radiation from the

source, and therefore result in solutions with significant truncation error, particularly near the

source region, while using smaller mesh sizes in air can provide good solutions with fine detail

over the problem domain.

2.3.3 Investigation of Cross-Section Libraries:

After satisfying the requirements for properly representing angular quadratures and spatial

discretization, the use of an appropriate cross section library is also important in solving


problems in medical physics. The scalar fluxes resulting from the deterministic calculations

using different cross-section libraries have been examined (Figure 2-9) and were compared with









similar values generated using MCNP5. To best present comparisons with the Monte Carlo

results (since Monte Carlo statistical errors were lowest (<3% relative) along the central axis),

we consider a slice through the central axis of the model. It is important to note that even with

the use of volumetric "mesh tallies" that yield global fluxes across a grid spanning the Monte

Carlo geometry, the statistical uncertainty associated with each mesh tally site, particularly those

distal or off-axis from the source region, can routinely exceed 40% or more, limiting their utility

in practical applications.

Figure 2-10 shows that for this particular problem, the relative differences in scalar fluxes

between all deterministic solutions and the Monte Carlo reference increased adjacent to the

gamma source. Also, for this problem, deterministic results using BUGLE-96 and CEPXS

demonstrated good results in this water phantom, and results were within the statistical

uncertainty of the Monte Carlo method. Subsequent investigation of these solutions could not

attribute this behavior to grid density, quadrature, or anisotropic scattering order. We did

determine that the relative fluxes of the Monte Carlo results (converged to <3% relative error),

particularly near the source, were strongly dependent upon the precise method by which the

geometric volume tallies were defined, and this could easily account for the differences

computed near the source. On the other hand, cross section libraries such as CEPXS provide

flexibility in defining energy group width not available in BUGLE-96. In general, differences

can be attributed to the energy group structure of the libraries. Results here demonstrate the

importance of library benchmarking exercises for specific problem applications. This may also

suggest that in some cases the use of an equivalent surface source fluence prescribed on the

entering phantom surface may be more practical to implement for dosimetry in the phantom,

discussed in the next section.













BUGLE-96 PE NTRAN vs ENDFIB-VI MCNP5
1.00E+10-

-o- MCNP5



1DEO1.00E+06 -PNR4


1.00E+04-


1.00E+02-


1.00E 100
0.0 30.0 60.0
Distance (cm)


A




C EPX(S PENIT RAN i ENDFW-VIIMONP5 SCALE PENITRAN vs ENDF B-UI MONIP5
1.00E+10 1.00E+10


1.00E+08 cr 1.00E+08-
PEIITRAl

1.00E+06 \I 1.00E+06


1.00E+04 -II1.00E+04- .


1.00E+02 -I 1.00E+02


1.000+ 00 1 00E+00
000 30.00 60 0 000 3000[ 60100
Distance~cm] Distance~cm)


B C


Figures 2-9. Scalar flux distributions generated by PENTRAN for a single energy group of each
library using the MCNP5 solution as the reference case. These fluxes span the central
axis in the three layers of the model. In general, the deterministic results yield a very
good match to the Monte Carlo results





Relative Deviation of PENTRAN from MCNP5


50-


40 -


~30




20-



0


- Bugle-96
-*- SCALES
+ CEPXS


30 60
Distance (cm)


Figure 2-10. Relative deviation of energy group 1 discrete ordinates results with Monte Carlo
results. Results using PENTRAN with BUGLE-96 were overall best in the water
phantom, and differences were within the Monte Carlo statistical uncertainty

2.4 Absorbed Dose Calculations

For the energy beam used in this study, the maximum electron range produced by the

primary photons is approximately 0.6 cm (as computed by the Tabata-Ito- Okabe (TIO) formula,

estimate for electron range, 41), which is less than the size of the smallest fine mesh used in the

water phantom. This allows electrons produced to deposit their energy within the fine mesh,

achieving CPE within the phantom volume. By applying this collisional kerma approximation

for absorbed dose computation based on the fluxes calculated using the SN method with

PENTRAN, we compared the SN based absorbed dose estimates with absorbed doses calculated

based on MCNP5 track length (F6) tallies, which estimate the energy deposited averaged over a

given cell volume and are equivalent to the collisional kerma, this is shown in Figure 2-11.



















-* MC


Water Phantom Dose Rate
Alona The Z axis


1.00E+04


1.00E+03


a01.00E+02
O
1.00E+01


1.00E+00


0 5 10 15 20 25 30
Depth(cm)

A



Dose Rate Along The X-axis Dose Rate Along the X-axis

1.00+04(5 cm Depth) 1.00E+04 (15cm Deph



1.00E+01 .0E0


1.00E+00 ~-i~bC~~~bbb 1.00E+00

B CU
Fiur 2-1 bobdds aecmaio ewe L mto sn em prxmto
an C ehnqesuin 6 aly C ttitca ncrait wr ls than3%o
averageb a
Forn th nryrneue nti rbew eosrthtteasre osert
cacultduighem appoxmato forteSmthduigheefenecs(P









iused in1 radiothrapythesecodr ectosgnratedmaio ewe can ehave larng e rmane erxceingthfine




mesh intervl cosqunl voaing the requiprxiatonfremhe nt fetor CPE, nd subequently inalidatin









the collisional kerma approximation as a basis for absorbed dose computation. In order to use

deterministic techniques in high energy photon dosimetry applications, some special treatment

for secondary electron transport is therefore required; this will be solved using the EDK-SN

method.

2.5 Parallel Decomposition

For solving the 60C0 model problem in a parallel environment, different domain

decomposition strategies were considered. In Table 2-1, we present the CPU times and memory

requirements for the PENTRAN calculations using different quadrature sets. Parallel

decomposition results for solution of the 60C0 problem model are also presented in Table 2-1

based on a constant number of meshes of 126,000 with P3 aniSotropy for a single energy group

and different angular quadrature orders. From the table data, it can be observed that increasing

the number of directions scales up running times to the processor workload. More work on each

machine lowers the relative impact of message passing synchronization and latency, relative to

additional computational work, as evidenced where eleven times added workload only required

9.3 times as much computational effort (referring to the S42 calculation). This is a useful

illustration that emphasizes that with parallel computation, added model fidelity may not actually

cost additional wall clock time to yield a more robust solution.

Table 2-1. Total number of directions, cumulative problem time** required for a single energy
group, P3 scattering, 126,000 fine mesh cells using 12 processors on a parallel cluster

~Angular Numb~er Parallel Decomposition Cumnulative
quadrature of Proc~essors Iterations Problem T~ime Time
(relative # (minutes) ratio*
workoad)Angle E~nergy Space

S12 (1.0) 12 4 1 3 6 3 1
S22 (3.2) 12 4 1 3 7 11 3.2
S32 (6.5) 12 4 1 3 7 22 7.3
S42 (11.0) 12 4 1 3 7 28 9.3

* Directions and time ratio are referred to the Sez quadrature set. ** Non dedicated running tune









2.6 Discussions and Findings

In this chapter, we highlighted recent studies examining the discretization requirements

of the deterministic SN method compared with the Monte Carlo method with application to

dosimetry problems. In doing so, we chose to simulate an idealized 3-D 60C0 model using the

PENTRAN code. Since radiation transporting in such applications encounters little interaction in

low density materials, such as air, attention was given to the quadrature level and mesh size

distribution needed for solution accuracy through the air and in a water phantom. As expected,

radiation transport in this gamma ray application is highly directional, and that a traditional SN

quadrature, such as an S12 leVel-Symmetric set (168 total directions), considered robust for many

neutron reactor physics applications, is insufficient for modeling gamma rays in the 60C0 model.

This limitation was overcome by the use of the higher levels of quadrature, up to S42 (1848 total

directions per cell) with Legendre-Chebychev quadrature sets and P3 anisotropy. We also noted

that at least an S32 angular quadrature leads to the elimination of most ray effects and scalar

fluxes within the statistical uncertainty of Monte Carlo. Moreover, model mesh sizes were

selected to preserve system volumes while taking into account the optical thickness of each

material. Smaller Cartesian meshes do better represent the curvature of the model, but they

require higher quadrature orders, and therefore larger memory and longer running times. Even in

optically thin air regions, a mesh cell interval of on the order of 1 cm was necessary to yield an

accurate solution, and more cells were necessary where the particle flux encountered strong

gradients; cell intervals may be relaxed near boundaries where detail is not required. Also of

importance was the flexibility of the ADS in PENTRAN, permitting more appropriate spatial

differencing schemes. Scalar flux distributions obtained from PENTRAN agreed with the Monte

Carlo prediction when simulating equivalent geometry models of the 60C0 model with three

different multigroup cross-section libraries. Still, realistic representations of the energy spectra









for dosimetry applications may require fine energy group structures; it is clear the applicability

of the cross section library should be investigated for each problem considered. Development of

a standardized multi-group cross-section library for these kinds of applications would be of

benefit to minimize potential inconsistencies in the application of currently available libraries.

The results of parallel deterministic PENTRAN code were obtained within comparable running

times to parallel MCNP5 Monte Carlo calculations for tally sites adj acent to the source. Overall,

we demonstrated that with an adequate quadrature set, cell mesh interval, and cross-section

library, it is possible to produce results with the PENTRAN discrete ordinates code that agree

with the Monte Carlo prediction. A major advantage of the SN method is that it provides a

detailed, accurate flux distribution at thousands of cells throughout the system, while the Monte

Carlo method is best for providing accurate values at a few selected points near the source. Both

methods can be advantageous in these types of applications, provided the level of SN

discretization is adequate.

This chapter laid out the foundations necessary for accurate SN discretizations in general,

although only a homogenous phantom was considered. The need to implement heterogeneous

media in the SN simulation is essential for any medical physics applications. In the next chapter,

we set the foundation for the new PENTRAN-MP code system. In this system, we implement a

state of the art series B phantom into our methodology by down-sampling these phantoms into

accurate phantoms to generate an equivalent PENTRAN input. Moreover, we calculate the

absorbed dose by applying a kerma approximation for diagnostic procedures. Then, the EDK-SN

method is developed, presented, and applied in subsequent chapters.









CHAPTER 3
WHOLE BODY DOSIMETRY SIMULATIONS WITH THE UF-B SERIES PHANTOM
USING THE PENTRAN-MP CODE SYSTEM FOR LOW ENERGY PHOTONS

This work describes our 3-D SN Medical Physics code system PENTRAN-MP in support

of ongoing development research in large scale 3-D absorbed dose computationS42. In the

PENTRAN-MP code system, Figure 3-1, an initial high-resolution data phantom is obtained

from state of the art 3-D whole-body voxelized human phantoms based on Computed

Tomography (CT) images43. A pediatric patient (an 11-year-old male) 3-D CT image is then

down sampled (collapsed) into a coarser one. The collapsed model was cast into a 3-D spatial

distribution and sub-divided into coarse meshes containing fine meshes with constant densities.

SN calculations were performed using PENTRAN and Monte Carlo results were generated with

MCNP5. To accurately compare our deterministic results with the Monte Carlo, we also

generated an equivalent MCNP5 31 model implementing continuous energy ENDF/B-VI cross-

section data libraries. Post processing in the PENTRAN-MP code system includes seamless

parallel data extraction using the PENDATA code, followed by the application of the 3-D-DOSE

code to determine absorbed dose in each voxel of the 3-D phantom based on fitted mass-energy

absorption coefficients 44, hence yielding a whole-body absorbed dose distribution. This chapter

is presented as follows: Sections 3.1, 3.2 and 3.3 highlight the methodology for accomplishing

our deterministic calculations. Section 3.4 presents a computational human phantom example

implementing the methodology of PENTRAN-MP and an equivalent MCNP5 model. Section 3.5

compares the Monte Carlo and deterministic solutions and analyzes the parallel performance of

the PENTRAN. Finally, Section 3.6 discusses the findings of this Chapter.















Voxelized Human
Phantorm forr Transporrt
Simu nations


PENTRAN SN Output for
EDK-SN Calculatiorns


Figure 3-1. PENTRAN-MP Code System Structure Flowchart

3.1 Methodology

In this work, we must determine the radiation transport flux and absorbed dose in a

computational tomographic (or voxelized) phantom using deterministic techniques. Central to

this is the application of detailed anatomical models in which we can render accurate absorbed

doses. Therefore, coupling of these anatomical models with robust 3-D deterministic transport

simulations can yield critical organ and whole body absorbed dose profiles from patient-specific

radiotherapy treatments, including absorbed dose delivered to collateral tissue located away from

the exposure site. To properly attribute absorbed doses in our models, we must set up a

discretized phase-space based on a detailed phantom for the material spatial distribution. Due to









interest in fully characterizing diagnostic absorbed doses to younger patients, for this

presentation, we chose a pediatric phantom. Also, to ultimately render absorbed doses in the

phantom wherever we desire, there are a number of codes that constitute the PENTRAN-MP

code system that include preprocessing to establish an x-ray source, appropriate interaction cross

sections, phantom spatial mesh, followed by the radiation transport calculation, and post-

processing computation data to yield absorbed doses and absorbed dose volume histograms, etc.

The details of this methodology are presented in the paragraphs below.

3.2 UF Series B Pediatric Phantoms

The UF Advanced Laboratory for Radiation Dosimetry Studies (ALRADS) has a well

established background in work with detailed human phantoms and phantom data. Following the

development of their CT-Contours segmentation software, and the construction of their first

tomographic dosimetry model (UF Newborn), the ALRADS research team focused on

constructing an expanded series of pediatric tomographic phantoms (Series B phantoms), this

time using live patient CT images from the Shands image archive. A total of five additional

phantoms have been constructed from CAP (chest abdomen pelvis) and head CT scans of

pediatric patients examined at the University of Florida (UF) Shands Children's Hospital.

Patient inclusion criteria included those patients displaying normal or near-normal anatomy at

their respective age and gender (e.g., avoidance of regions of tumor, edema, organ enlargement)

Table 3-1. While the previous phantoms preserved the body dimensions and organ masses as

seen in the original patients who were scanned, comprehensive adjustments were made for the

Series B phantoms to better match International Commission on Radiological Protection (ICRP)

age-interpolated reference body masses, body heights, sitting heights, and internal organ

masses





Table 3-1. Computed tomography image sources for the development of the UF pediatric
phantom series
Phantom Gender Fused Body iImage


Vaxel dimensions (mm)A
0.86 x 0.86 x 3.00 289
0.90 x 0.90 x 5.00 351
1.16 x 1.16 x 6.00 322
0.94 x 0.94 x 6.00 398
1.18 x 1.18 x 6.72 394


rrayr size
x 180 x241
x 207 x211
x 171 x 220
x 242 x252
x 193 x 252


UF 9 months
UF~ 4 years
UF 8 years
UF 11 years
UF 14 years


Figure 3-2. The UF Series B pediatric phantoms series (Newborn, 9-months, 4-years, 8-years,
11-years and 14-years old patients)43

For this work, a pediatric patient (an 11-year-old male, depicted in Figure 3-2) is voxelized

with high resolution using 0.94 mm x 0.94 mm x 6.00 mm voxels as a case study. This geometry

data was then collapsed into a 4.7 mm x 4.7 mm x 30.0 mm voxelized model using GHOST-3-

D, described in section 3.3. Subsequently, using PENMSH-XP, the collapsed model was cast









into a 3-D spatial grid distribution and sub-divided into coarse mesh grids, each containing a

corresponding number of Eine mesh cells.

3.3 PENTRAN-MP Code System

Currently, the PENTRAN-MP code system is composed of 3 levels of calculation,

including pre-processing algorithms that include the GHOST-3-D phantom voxel collapsing

(back-thinning) code, the DXS x-ray source generation code 45, the PENMSH-XP 3-D mesh

distribution and input generator code, and the CEPXS code (a product of Sandia National

Laboratories) for generating multigroup cross sections 33. The radiation transport calculations are

carried out in PENTRAN, and post-processed using PENDATA to gather parallel distributed

data, and 3-D-DOSE to render user specified absorbed dose data.

3.3.1 Pre-processing Codes

3.3.1.1 The collapsing code (GHOST-3-D)

The GHOST-3-D, developed by Ghita and Al-Basheer, is a code used to input CT-scan

binary data obtained from a high resolution voxelized human phantom. There are several

constraints in simulating mega-voxel models, including balancing limited computer memory

(even with the benefit of parallel memory storage, as in PENTRAN) with proper discretization,

etc. To optimize accuracy and computational work in representing the voxelized phantom, we

developed a code, GHOST-3-D, that provides an equivalent model with a reduced number of

meshes (subj ect to user-derived accuracy targets), overcoming the mentioned difficulties.

3.3.1.2 The mesh generator code (PENMSH-XP)

PENMSH-XP35 by Yi and Haghighat is a 3-D Cartesian-based mesh generator that

prepares discretized material and source distributions for the PENTRAN particle transport code.

PENMSH-XP is capable of performing two different methodologies in generating PENTRAN

input files. In the first methodology, a physical model is partitioned into slices along the z-axis,









referred to 'coarse z-levels', and then each z-level is partitioned into single coarse levels of 3-D

x-y-z fine mesh grids. Predefined shapes can also be used to generate the desired shapes in a

particular model. A second methodology is based on two binary input files, one for the phantom

geometry, which can be established from real CT data, and the other containing the source

density for each voxel/fine mesh, which can be used to specify a fixed-source distribution in the

model. By combining the two files, PENMSH-XP directly generates an input deck for the

PENTRAN transport code.

3.3.1.3 Source spectra and cross-sections generation

Based on the bremsstrahlung and characteristic parametric functions of the model proposed

in 1991 by Tucker, Barnes, and Chakraborty 46 the DXS code, principally developed by Monica

Ghita, generates x-ray spectra in the diagnostic range (50-140 keV), according to the input

parameters (tube potential, anode angle, type and amount of filtration), and accommodates them

to any energy group structure. Optionally, as part of the PENTRAN-MP package, DXS also

generates the input files for CEPXS (Sandia National Lab), the x-ray cross-section generator

code, corresponding to a desired energy group structure and PN order. Group-dependent cross-

sections rendered by CEPXS are then stripped from the output file and adapted to a format

suitable for the PENTRAN deterministic calculations using another code developed by Ghita,

GREPXS47

3.3.2 Transport Calculation

Transport calculations were performed using the PENTRAN SN and MCNP5 Monte Carlo

as discussed in Chapter 2.









3.3.3 Post-Processing Codes

3.3.3.1 Data management utility (PENDATA)

Immediately following a transport calculation, a data management utility, PENDATA,

seamlessly gathers parallel stored data automatically following a parallel PENTRAN run, and

provides several options for the user in stripping results from parallel output files, including data

extractions from binary file storage which is an essential utility for a code with parallel I/O.

3.3.3.2 Dose calculation code (3-D-DOSE)

The 3-D-DOSE code, developed by Al-Basheer is used to render absorbed doses based on

photon fluence results derived from the PENTRAN transport calculations. In doing so, a

methodology for determining absorbed dose in the phantom based on photon fluence was

developed for situations where electron transport effects do not require specific treatment. CPE is

assumed to be established in a volume when the energy carried into the volume by charged

particles is balanced by the energy carried out of the volume by the charged particles.

In the example simulation in Section 3.4, the maximum energy used is 90 kV, for which

the range of the secondary electrons produced is smaller than smallest voxel dimensions used in

the model. Moreover, the atomic number of the material Z is relatively low, excluding major

production of bremsstrahlung radiation, thus depositing all the photon energy in the voxel,

achieving CPE. Consequently kerma is equal to absorbed dose in this case. We note this is not

true for regions in close proximity to the surface of the model using higher energies due to

absorbed dose buildup, where corrections for secondary electron transport may be quite

significant, especially for high energy beams. After extracting 3-D scalar fluxes from the

phantom model we apply the 3-D-DOSE code. Based on fitted mass-energy absorption

coefficients 44, Shown in Figure 3-3, and the scalar fluxes from the PENTRAN calculations, 3-D-


















Soft tissue (ICRU-44)
1000 +1CRU-44
-rFIT
100





1.00 E-03 1.00E-0n 1.00E-01 1.00 E+00 1.00E+01 1.0+2

0.1



Energy(Mev)


Cortical Bone (ICRU-44)
1 00E+03
lCRU-44
'uI 1 00E+02 -I FIT


0 1 00E+01



S10 -03 1.00E-02 ~1.00E-01 1.00 E+00 1.00E+01 1.0+2
1 00E-01



Energy(Mev)


DOSE calculates absorbed dose in each voxel yielding a whole-body absorbed dose distribution

and dose-volume histograms for critical organs of interest.


Energy(Mev)


Figure 3-3. Values of the mass energy-absorption coefficient, PdenlP, and the fitting function as
a function of photon energy, for Lung (ICRU-44), soft tissue (ICRU-44) and cortical
bone (ICRU-44)

3.4 Computational Example

For this work, a pediatric patient (an 11i-year-old male) was voxelized with high resolution

using 0.94 mm x 0.94 mm x 6.00 mm voxel size (398 voxel x 242 voxel x 252 voxel) as a case










study. This geometry data was then collapsed into a 4.7 mm x 4.7 mm x 30.0 mm voxel size (79

x 48 x 50 voxels) model using GHOST-3-D. Subsequently, using PENMSH-XP, the collapsed

model was cast into a 3-D spatial distribution and sub-divided into six coarse mesh z-levels

containing five coarse meshes in x-y domain, and these contained 6320 fine mesh cells (Figure

3-4, 3-5). An isotropic volumetric (17.8 cm x 3.65 cm x 30 cm) x-ray source was placed above

the upper left side of the phantom.

Although the model has relatively small fine mesh sizes, using S32 Legendre-Chebychev

quadrature, the adaptive differencing, and CEPXS cross-sections with P3 aniSotropy, we were

able to eliminate any unphysical oscillations that may result in the deterministic solution. Using

4, 6, and 8 energy groups, the SN model was executed on 16 Opteron processors and was

completed in 1.8, 2.1 and 3.8 hours, respectively.

With the general purpose radiation transport code MCNP5, we simulated an equivalent model to

the one simulated by PENTRAN. The collapsed 3-D binary files composing the UF voxel

phantom was converted to ASCII formatted matrix files and then incorporated into a MCNP5

lattice defined input. An input deck example for the MCNP5 model is presented in Appendix B.

Volumetric flux (F4) mesh tallies were used in the Monte Carlo geometry and were

equivalent to the discretized SN volumes all over the model. All simulations were performed

using the parallel cluster Einstein composed of 16 processors at 2.4 GHz with 4 Gb/processor,

for a total of 64 Gb of parallel memory used for general purposes. Parallel MCNP5 results in the

source region converged with a statistical error less than 4% in the source region on average.





Figure 3-4. UF-Series B phantom (11 years old male) with (398 x 242 x 252 voxels); the
corresponding PENTRAN "back-thinned" models with (132 x 80 x 84 voxels) and
72 materials (upper right), and another with (79 x 48 x 50 voxels) and 66 materials
(lower right)



































Figure 3-5. Transversal view of the 11i-year-old male Series B original phantom (upper left),
collapsed into 189k voxels and 4 materials (upper right), collapsed into 900k voxels
and 4 materials (lower left) and collapsed into 900k voxels and 72 materials (lower
right)

3.5 Computational Results

3.5.1 Flux Comparison of Deterministic and Monte Carlo Results

Comparison of the SN and Monte Carlo results for the scalar fluxes along the source central

axis and in the source region along the z axis at different depths for different energy groups were

made. Figure 3-6 depicts a 3-D scalar flux distribution computed by PENTRAN for the reference

case. This simulation was based on the use of an S32 angular Legendre-Chebychev quadrature

(1088 directions/mesh) with P3 Scattering anisotropy using the CEPXS with 8 energy groups.







































G2(70-80)key
1 81E+06

1 51E+06
MCNP5
1 21+06- *PENTRANI


9.10E+05

610E+05


1.00E+04 -*
0 5 10 15 20 25
Y(cm)


G4(50-60)key
6.01E+06

5101E+06 .,

4.01E+06 i"IMCNP5
x -PENTRAN
q3.01E+r06

2.01E+06

1.01E+06

1.00E+04
0 5 10 15 20 25
Y(cm)


G1(80-90))key
6.01E+05


4.51E+05
-MCNP5

30E0 ~PENTRAN


1 51E+05


1 00E+03 "
0 5 10 15 20 25
Y(cm)


G3 (60-70) key
3.51E+06

2 81E5+06

2 11E+0 i MN
k -PENTRAN
1.41E+06

7.10E+05

1.00E+04__
0 10 20 30
bicm


50


20'e" ,-.,2


Figure 3-6. x-ray 3-D Group 1, 3, 5 and 7 scalar flux computed by PENTRAN with the CEPXS
cross section library, S32 angular quadrature (1088 directions) with P3 Scattering
amisotropy


Figure 3-7. Flux distributions using S32 with P3 scattering anisotropy in PENTRAN using the
CEPXS library vs. MCNP5 Monte Carlo results for energy groups 1, 2, 3 and 4 along
the source central axis. Monte Carlo uncertainties were less than 5%, 4%, 4%, and
3% respectively






























G7(20 30)key


G6(30-40)key
7.51E+06

6.01E+06

4.51E+06*PETA

S3.01E+06 '

1.51E+06

1.00E+04
0 10 20 30
Y(cm)


2.00E+05 G 8(10-20)key


1.50E+05 .MCNP5
-PENTRAN

9.00E+05


i5.00E+04 ;


0.00E+0 6 ..
0 5 10 Y~cm)15 20 25


Y=166cm X=13Decm

1.41E+06

1,21EI06





4.10E*05 is 5n 1
: IleG:

0i 00E*0 2.04140E0 .0E0 .0*110E
Z, .cm)G


Z=43.5cm Xr13.0cm

1.01E407
0.10E+06 ,
8.10E*06

7.10E+06*Mc8s



3.10E+0 a9 G,


000000t 2.0E0140E0160E 1 0E+1 .0E+2



Z(cm)


G5(40-50)key


6.01E+06
5 01E+06

3.01E+06

201E+06
1 01E+06


* MONPS,
* NPENTAN


O 5 10
Y~cm)


15 20 25


-M,;Npf
* PENTRAN


i1.76E+06 -


1.00E+04 ; "****
O 5 10 Y(cm) 15 20 25


Figure 3-8. Angular flux distributions using S32 with P3 scattering anisotropy in PENTRAN
using the CEPXS library vs. MCNP5 Monte Carlo results for energy groups 5, 6, 7
and 8 along the source central axis. Monte Carlo uncertainties were less than 2%, 3%,
6%, and 30% respectively


Figure 3-9. Angular flux distributions using S32 with P3 scattering anisotropy in PENTRAN
using the CEPXS library vs. MCNP5 Monte Carlo results for energy groups 1, 3, 5
and 7 along the z axis at different depths in the phantom. Monte Carlo uncertainties
were less than 5% at y=4.3 cm, 6% at y=13.5 cm









As shown in (Figures 3-7, 3-8, and 3-9), the PENTRAN results with S32 Legendre-

Chebychev quadrature are in very good agreement with the MCNP5 predictions in the source

region; MCNP5 mesh tallies were converged to <6 % relative error, and typically < 2% for most

of the results for the reference case (using 8 energy groups). Overall, this figures show that

deterministic techniques are capable of producing accurate results within the statistical error of

Monte Carlo methods. Since the uncertainty is proportional to 1/ (number of histories), to

achieve an equivalent-uncertainty Monte Carlo simulation with converged mesh tally

information 70 cm away from the source in energy group 1, we estimate over 200 hours are

needed for MCNP5 executed on 16 processors to achieve fluxes to within 5% for this group;

longer times would be required for lower energy x-rays.

3.5.2 Dose Calculations and Dose Volume Histograms (DVH).

Using 3-D-Dose, 3-D absorbed dose distributions were calculated in each voxel in the

whole model of the 11-year old male phantom model for broad-beam external photon field.

Figure 3-10 shows the stages of calculation in the PENTRANP-MP code system that results in 3-

D Dose distributions for the entire phantom. Figures 3-11, 3-12, and 3-13 show selected slices of

the phantom and the corresponding absorbed dose distribution, while Figure 3-14 depicts

selected organs with correspondent 3-D Absorbed dose distributions.





































Figure 3-10. Simulation methodology for 3-D absorbed dose computations using PENTRAN-
MP code system. Left: UF series B voxelized phantom, middle: correspondent
PENTRAN input, right: 3-D absorbed dose distribution after irradiation with a x-ray
source


































Figure 3-11i. CT view of the 11i-year-old male Series B original phantom collapsed into (a) 189 k
voxels and the correspondent absorbed dose distributions (b) 900 k voxels and the
correspondent absorbed dose distributions


































Figure 3-12. CT view of the 11i-year-old male Series B original phantom collapsed into (a) 189
k voxels and the correspondent absorbed dose distributions (b) 900 k voxels and the
correspondent absorbed dose distributions
































Figure 3-13. CT view of the 11i-year-old male Series B original phantom collapsed into (a) 189
k voxels and the correspondent absorbed dose distributions (b) 900 k voxels and the
correspondent absorbed dose distributions


Figure 3-14. Simulation methodology for 3-D organ absorbed dose computations using
PENTRAN-MP code system










Dose volume histograms (DVHs), for a selected number of organs giving the percentage of

tissue volume (organ) that receives an absorbed dose greater than the value indicated on the

abscissa, were calculated and compared to those results from the equivalent Monte Carlo model.

As anticipated, DVH results are in good agreement for the SN and Monte Carlo radiation

transport simulations (Figure 3-15). However, Monte Carlo calculations can require substantial

run times for low precision at "out of field" distances in the phantom, particularly as photons

down-scatter to lower energies. The Monte Carlo results computed values will eventually

converge with increased run time as more particles reach sites distal to the source.




DVH (Left Lung) DVH-(RIGHT LUNG)
In-field Out-of-field
100.00% k I100.00%


75.00% i 1 75.00%


S50.00% DVH(L Lung)-SN 50.00% -t DVH-(R Lung)-SN
SDVH(L Lung)-MCNP DVH-(R Lung)-MCNP

25.00% i L 125.00%


0.00% II ... .00 ,****
0.00 0.40 0.80 1.20 1.60 2.00 0.00 0.10 0.20 0.30 0.40 0.50
Dose (mGy) Dose (mGy)

a. b.
Figure 3-15. Dose Volume Histograms (DVH) computed based on PENTRAN (S32,P3)
transport results using the CEPXS library vs. DVHs computed using MCNP5 Monte
Carlo results. MCNP5 Monte Carlo results required 16 hours 16 processors compared
with the PENTRAN-MP SN solution, requiring <1.9 hours on 12 processors for the
whole-body computation

3.5.3 Parallel Computation

For the UF 11 year old male phantom problem, a decomposition strategy was used to assist

the parallel performance. In Table 3-2, we present the cumulative problem time and memory









requirements for deterministic calculations discussed in previous sections using different

quadrature sets.


Table 3-2. Total number of directions, Cumulative Problem Time required for different energy
groups, P3 Scattering, and 189k fine mesh cells
Number SN Number Parallel Cumulative
of (relative of Decomposition Problem Time
Energy workload) Processors (hours)
groups A E S

4 42 (2) 12 4 1 3 4.1
6 24 (1) 12 8 1 2 2.1
8 32 (2.32) 12 4 1 3 3.69
8 32 (2.32) 16 8 1 2 3.06


Parallel decomposition results for solving the phantom are presented in Table 3-2 based

on a constant problem phase space of 189,000 cells with P3 aniSotropy and different numbers of

energy groups numbers and several angular quadratures. "A, E, S" refers to the angular, energy

group, or spatial decomposition applied to solve the problem using parallel computing. From the

table data, it can be observed that increasing the number of directions requires longer running

times that essentially scale with processor workload.

3.6 Discussion and Findings

We have demonstrated a methodology incorporating a discrete ordinates (SN) approach

that is an effective alternative to the Monte Carlo method for accurately solving the transport

equation over an entire human phantom. Models here were based on the UF series B pediatric

phantoms developed by the University of Florida ALRADS group. We simulated the complete

phantom discretized using 900k/186k voxels, and demonstrated that the SN method in the

PENTRAN-MP code system, developed as a team effort, readily provides accurate whole body

fluxes. Radiation transport throughout the phantom encounters few interactions in low density









materials, such as air. Attention was given to the quadrature level, using at least an S32 Legendre-

Chebychev angular quadrature, yielding 1088 total directions per mesh-cell; we were able to

accurately obtain a solution analogous to the MCNP5 Monte Carlo calculations. Also, the tally

results from the eight energy groupsrepresenting the spectrum and using the CEPXS multi-group

cross-section library, were compared to the continuous energy MCNP5 Monte Carlo results

using the ENDF/B-VI data libraries.

Overall, with an adequate quadrature, mesh size, and cross-section library, it is possible to

produce results with the PENTRAN discrete ordinates code that agree within the statistical

uncertainty of the "on-axis" MCNP5 Monte Carlo calculations for the proposed model. Parallel

deterministic PENTRAN results were obtained within comparable running times to parallel

MCNP5 Monte Carlo calculations for tally sites adjacent to the source.

A major advantage of the SN method is that it provides a detailed, accurate flux

distribution at thousands of cells throughout the system, while the Monte Carlo method only

provides highly accurate values for selected points near the source. Though the use of

volumetric "mesh tallies" can provide global fluxes across a grid spanning the Monte Carlo

geometry, the statistical error associated with each mesh tally site that would require, according

to estimates, more than 200 hr for tallies 70 cm away from the source to converge to less than

5%., limits their utility in practical applications. An equivalent Monte Carlo calculation to obtain

the flux distribution with reasonable statistical sampling error over the entire problem space may

require orders of magnitude longer running times compared to parallel deterministic SN

calculations.

While more tests are needed, the recently developed Exponential Directional Iterative

(EDI) SN differencing scheme was excellent in dealing with fine inhomogeneous material









densities inherent in medical physics problems, properly handling numerical radiation transport

physics for diagnostic energy beams.

Overall, for the energy ranges used in the problem, the secondary electrons need not

specifically be modeled using charged particle transport methods to accurately determine global

absorbed dose distributions. This is due to the fact that for an SN calculation with an appropriate

numerical fidelity (evidenced by the excellent agreement with Monte Carlo), secondary electrons

that result from photon interactions in the phantom have equivalent ranges smaller than the

minimum SN spatial grid interval used to represent the anatomical structure. Alternatively, at

higher photon energies (not considered here), this is not the case, and charged particle transport

and corresponding interactions yielding energetic electrons must be considered over the problem

phase space to yield an accurate absorbed dose distribution. In Chapter 4, a new EDK-SN

approach is presented to account for higher energy beams and secondary electron absorbed dose.









CHAPTER 4
ELECTRON ABSORBED DOSE KERNELS TO ACCOUNT FOR SECONDARY PARTICLE
TRANSPORT INT DETERMINISTIC SIMULATIONS

In previous chapters, Charged Particle Equilibrium (CPE) and its limitations were

presented, and that in cases with high energy photons, corrections for secondary electron

transport may be significant. As mentioned, this is readily treated in Monte Carlo codes, yet quite

difficult to treat explicitly in deterministic codes due to the large electron ranges and added

numerical complexities in reaching convergence in photon-electron transport problems The

goal here is to produce a methodology to accurately estimate organ absorbed doses and attribute

whole body absorbed doses incurred from irradiations characteristic of external beam therapy

using the CT-based anatomical patient geometries in 3-D deterministic discrete ordinates (SN)

radiation transport models. Accurate absorbed doses will be rendered using pre-computed, high

resolution Monte Carlo based electron kernels. In doing so, the flux and the current provided by

the SN method will be used to both scale and project the absorbed dose using the Monte Carlo

based kernels, and map it to the surrounding voxels. Hence, this approach preserves high fidelity

and accuracy, accounting for complete photon transport, with the final absorbed dose delivered

from electron transport propagated by coupling with 3-D deterministic photon transport; this is

what makes this approach novel. In this chapter, water phantom results using a 0 to 8-MeV step

uniform beam will demonstrate that absorbed doses can be accurately obtained within the

uncertainties of a full physics Monte Carlo simulation.

4.1 Introduction to The EDK-SN Method

Various modalities are applied in absorbed dose calculations for high energy photon

beams, including empirical models that are based on measurements made in the clinic, Of

theoretically based models, such as the application of Monte Carlo techniques3'48. Alternatively,

as mentioned earlier, deterministic methods have been implemented to assess absorbed dose










calculations20,21, although experience has shown that 3-D deterministic electron transport is quite

challenging in order to yield a converged solution. To properly treat 3-D electron transport

physics deterministically, yet still achieve reasonably fast and accurate whole body computation

times for simulations with high energy photons, we have developed angular-energy dependent

transport "electron dose kernels," or EDKs that can be coupled to SN solutions. These kernels

were derived via full physics MCNP5 Monte Carlo electron transport simulations, and are

applied based on deterministic photon solutions over the problem phase space, thereby

accurately accounting for the absorbed dose that is ultimately rendered by charged particle

electron transport. This is possible because the PENTRAN discrete ordinates code preserves

detailed photon flux and current data on a voxel-by-voxel basis, so that accurate whole body

absorbed doses may be rapidly achieved for high energy photon sources by performing a single

deterministic SN multigroup photon calculation on a parallel cluster. This is followed by linking

the SN-derived photon fluxes and net currents to our Monte-Carlo-based Electron Dose Kernels

to account for a full-physics absorbed dose; we call this the 'EDK-SN' methodology for

determining 'total body dosimetry'. This 'total body dosimetry' enables dose to be computed

across the entire phantom phase space, wherever it is required, presented in the following

sections.

4.2 Monte Carlo Based Dose Kernels

Using Monte Carlo calculations, electron absorbed dose kernels for pre-determined photon

energy groups were generated; energy was deposited in a collective "cloud" of voxels (i, j, k)

composed of tissue, where absorbed doses attributed to these voxels are directly caused from

incident primary photons in a given energy group that propagate electrons from voxel, (i',jk')

the 'dose-driving voxel' or DDV. When producing these 'electron absorbed dose kernels,' the

photons having energies lower than the limits of the energy group are cut off as they are









generated to prevent a cascading contribution to the EDK for lower energy groups. These kernels

were then applied along the direction of the net current and scaled by the magnitude of the

photon flux on a fine mesh basis provided by the PENTRAN discrete ordinates code, as well be

explained in later sections. Figure 4-1 illustrates the DDV, with an indicated direction of photon

current. The Electron absorbed Dose Fractional kernel in voxel (i, j, k), given as the EDF,, due to

a primary photon group, can be determined in terms of the initial photon energy for a particular

energy group g. By partitioning the energy deposited in voxel (i~j,k) into multiple energy bins

that are aliased to the SN energy group structure, we can predict the fractional electron absorbed

dose kernel contribution per unit photon flux per source particle:

ED)F (i, j, k )= EDKg (i, j, k ) #MCg (i ', j ', k ')(41

Therefore, the SN transport-guided absorbed dose rate can then be obtained for any voxel

location (i, j, k) by summing the product of the EDF,(i~j,k), the particle normalization factor (P),

and the SN scalar flux #(i, j~k),g divided by the voxel mass M (i, j, k):


D~ij~k= s[ED, (i j~) ]f~i j~ksm l/Mi, ~k)(4-2)

This procedure will enable the detailed information to be obtained on the absorbed dose

deposited in the model by scaling the EDKs by the photon flux and projecting them by along

photon net current vectors computed for each voxel. This will directly account for secondary

electron physics in each voxel to achieve an accurate 'total body' attributed absorbed dose.

The particle normalization factor P in equation 4-2 correctly scales the number of voxels

linked using the EDK, and is set by the deterministic grid spacing. We note that this factor is

specific to the grid used to estimate the EDK-SN. Consequently, to determine this factor, the

absorbed dose is calculated through the EDK-SN model and should be scaled to the absorbed










dose calculated via a Monte Carlo full physics absorbed dose in an equivalent model. This P

factor is a kernel-specific factor, and once determined, it can be used for any execution

implementing the corresponding electron kernels, provided grid spacing is consistently used.


Photons

















Figure 4-1. Procedure to produce EDK based on Monte Carlo Calculations

4.3 EDK Determinations Based on The MCNP5 Monte Carlo Code

In the MCNP531 Monte Carlo code, there are three different methods by which one can

tally to yield absorbed doses to be used for integrated organ absorbed dose calculations; these are

achieved by using either a cell flux tally (F4), an energy deposited tally (F6), or a pulse height

tally (*F8), and each has a different basis for computing the absorbed radiation dosel4,31. To

illustrate this, we compare absorbed doses computed using the F4, F6 and *F8 tallies,

respectively to highlight the differences in these computations, and establish a procedure that

ensures that our electron kernels are calculated accurately, and the Monte Carlo transport and

subsequent absorbed doses determined as a benchmark are completely based on a detailed

physics treatment.









4.3.1 Computational Model

To highlight the differences between the different tallies, used for absorbed dose estimation,

in the MCNP5 code, we considered a box of 11x1Ix1 1 cm3 COmposed of a water cube containing

a lxlxl cm3 8 MeV forward peaked X-ray source placed at the center of the cube Figure (4-2).

The size of this model allows complete deposition of secondary electrons produced from the X-

ray source. A flux mesh-tally, using "DE" absorbed dose energy and "DF" absorbed dose

function cards, was generated to cover the entire model. Moreover, since no mesh tally is

available for *F8 and F6 tallies in the MCNP5 code, a 3-D tally distribution of cells, equivalent

in purpose to the MCNP5 F4 mesh tally, were implemented in the model. This model, with the

implemented tally sites, was used for highlighting the differences between the different tallies,

and consequently was used for reporting a scaled full physics absorbed dose.

4.3.2 MCNP5 Tallies Comparison

The source in the center of the model was applied to generate a pencil beam impinging

perpendicularly, along the z-axis, on the box of water. As shown in Figures 4-3, 4-4, and 4-5, the

absorbed dose is highly forward peaked, with little dose resulting from back-scattered radiation.

Absorbed doses resulting from the use of an F4 volume flux tally method, using a "DE" dose

energy and "DF" dose function cards, compared to an F6 tally using both the photon-electron

mode (p, e mode) and photon only (p mode) were all in agreement, since the two tallies depend

on similar physics in predicting absorbed dose. These results are depicted in Figure 4-3. On the

other hand, when comparing absorbed doses computed using a *F8 using a fully coupled photon

electron transport mode (p, e mode) and the F6 derived dose tallies, the differences were

significant, as depicted in Figure 4-4. This is expected, since F6 tallies are analogous to F4 tallies

in MCNP, and can be directly aliased to assumptions equivalent to CPE, while *F8 tallies are

determined based on full charged particle transport physics, as discussed below31





Figure 4-2. Monte Carlo model used for comparisons between MCNP5 pulse height tally,
energy deposited tally and cell flux tally employing dose conversion factors (DE,
DF). The source is located at the dose driving voxel DDV


0 202 v G ui l U l a~ illll IIvIFIl v utig~



S0.15--E



o. 10
Cell
fact


8 0.05



0.00
0 5 10 15
Depth (cm)


Figure 4-3. Comparison between F6 (p, e mode), F6 (p mode), and F4 (DE, DF) tallies in
MCNP5, statistical uncertainty was less than 1%









Dose rate using MCNP5 different tallies
3.00E-01

2.50E-01

2.00E-01 -*- iI Energy deposited tally using
(p,e mode)"
o -a-Pulse height tally using (p,e
1.50E-01 .-Y mode)

S1.00E-01

g 5.00E-02

0.00E+00 g
0 5 10 15 20 25
Depth (cm)

Figure 4-4. Comparison between F6 (p, e mode) and *F8 (p mode) tallies in MCNP5, statistical
uncertainty was less than 1%

I ODose rate (p~e mode Vs. p mode)


1


Figure 4-5. Comparison between *F8 (p, e mode) and *F8 (p mode) tallies in MCNP5, statistical
uncertainty was less than 1%


0.30

0.25

0.20

0.15

0.10

0.05

0.00


-*Pulse hight tally with (p,e mode)
-.Pulse hight tally with (p mode)


~--3Ccrl


101
Depth (cm)









When comparing *F8 (p mode) transport and *F8 (p, e mode) transport, (Figure 4-5) shows

significant differences in the outcomes for the same simple model. This directly results from

accessing different calculation modes in MCNP5 with regard to electron physics. If electron

transport is turned on (mode p, e), then all photon collisions except coherent scatter lead to

electrons that are banked for later transport. Alternatively, if electron transport is turned off

(achieved by omitting the 'e' on the MCNP5 Mode card), then a "thick-target bremsstrahlung"

model (TTB) is used. This model generates electrons, but assumes that they are locally "slowed

to rest". The TTB production model contains many approximations compared to models used in

actual electron transport. The choice of which tally is to be used and the detailed physics

treatment by which the Monte Carlo code is executed is essential to the success of the EDK-SN

method. In conducting computations for generation of the EDKs, we used the full detailed

physics treatment, implementing a *F8 (p, e mode) tally, as opposed to the alternative methods

that lead to less robust electron physics, essentially employing a collisional kerma. A detail

discussion on photons interactions in matter is presented in Appendix A.

4.4 EDK -SN Code System Methodology

4.4.1 Net Current

For any SN calculation, the PENTRAN discrete ordinates code preserves angular information

explicitly in parallel data storage arrays. This allows one to calculate Jser and to apply the pre-

computed electron absorbed dose kernel along the net current direction, Figure 4-6, where:



J =J i^+J 7+J. k
net nx ny =(4-3)



J, = J J-,
(4-4.a)









(4-4.b)

(4-4.c)


J = J


(4-5)


Where ~ju is the photon net current unit vector.


Figure 4-6. Deterministic calculation will provide the angular information explicitly

4.4.2 MCNP Based Electron Kernels

When generating these electron kernels, a single beamlet of mono-energetic photons, for a

certain energy group, interacts at the center of a water phantom, as discussed in the previous

section. For the purpose of this study, we generated the EDKs for 16 photon energy groups

covering the energy spectrum from 0 to 8 MeV. Based on our previous study, as discussed in

chapter 1, these energies will violate CPE for the model considered. Moreover, the width of the

energy group is set to be 0.5 MeV, minimizing the effect of energy discretization in an

acceptable SN running time (we note that energy group discretizations can be optimized for better


Jnet

I~-~et









time performance, although this is not considered here). Kernels are computed and then scaled

by photon flux in the DDV in dose kernel voxels (projected from the DDV) along the net current

direction of the deterministically computed beams of same energy group (Figure 4-7). It is

important to note that for each energy group calculation, the Monte Carlo model was set to cutoff

photons with energies lower than the lower limit of the energy group being modeled. This is

important to prevent any cascading contribution to the EDK to lower energy photons.


SAngulatr Phonton Flux








Fiur 47.Fo ec eery ropusn Mont Crlo raiatiotrnptaecibamf
mono-energetic photons interact atte ete f atrphno







canr 47 also b epaced n g various tsuetpsand eontergies relatives to a spot eific dietio of poo





beam determined from solution of the Boltzmann transport equation using PENTRAN. This will

permit complete attribution of the absorbed dose ultimately due to electron straggling, where the

EDK coupled with SN (EDK-SN) provides an "equivalent electron absorbed dose look-up" for

each photon "beamlet" in a discrete ordinates computation. However, to achieve precise

solutions using the EDK-SN, the SN calculation should be carried out with great attention to

phantom model representation and meshing, proper quadrature sets, cross section libraries, and









SN differencing schemes; these issues were investigated and presented in Chapter 2 in assessing

what critical discretizations are required to yield accurate photon transport soluationS42

4.4.3 Euler Angles

Euler angles, described in (Figure 4-8), provide a means of representing the spatial

orientation of any frame of angular space as a composition of rotations from a given reference

frame. The translations afforded by Euler angles were implemented to enable projections of the

EDKs for a collection of cells aliased to a single reference pencil beam vector, as modeled in the

Monte Carlo simulations (as in Figure 4-2) to generate EDKs to an arbitrary net current vector

specifying the direction of the photons derived from SN computations. In the following

discussion, we denote a fixed system in lower case (x,y,z) and a corresponding rotated system in

upper case letters (X Y,Z). We use the term "Euler Angle" for any representation of three

dimensional rotations of the electron kernel absorbed dose matrix that aligns with the direction of

the net current for each DDV.







y(2) Y
casel sh op sn cs




0oc 0ia 1 csin 0 cosip 0os 0.o 1



Figure 4-8. Euler angles, zyz convention. Similar conventions can be obtained by following of
the following order in rotation (zyz, xyx, xzx, yzy, yxy)

Where :









a is the angle between the y-axis and the line of nodes .
p is the angle between the z-axis and the Z-axis .
y is the angle between the line of nodes and the Y-axis.

The above description works for the z-y-z form. Similar conventions are obtained just

renaming the axis (zxz, xyx, xzx, yzy, yxy). There are six possible combinations of this kind, and

all of them behave in an identical way to the one described before. Just imagine three rotations

over the set of axis forming the rotation matrix R.

[R ]= [Rz (a)][R, (/7)][Rz (7)] (4-6)

Where R is the rotational element matrix

Depending on the convention used for the calculating these angles, they can be related to

the angles for the net current obtained from the PENTRAN run. Assuming that rotated matrix

indices to be known, we generated the vector A for each point from the center of the matrix,

where the dose values of this vector form the matrix dose point to be projected onto the SN

matnx :

[A '] =[R][A ] (4-7)

[R -] [A '] =[R1] [R ][A ] (4-8)

[A ] =[R ][A '] (4-9)

When calculating the original un-rotated A vector for each point, we interpolate the vector

location, in 3-D, for the corresponding dose value based on the original EDK matrix calculated

with the Monte Carlo model. This generates a 3-D translated (by rotation) electron absorbed dose

kernel matrix corresponding to a particular direction in space. For the current code system, this

process was performed for 1848 directions, generating 1848 files for each energy group. An

example of a rotation of the EDK from a reference (Monte Carlo) EDK is depicted in Figures 4-

9.a and 4-9.b.

































V1


A























V1

B
Figure 4-9. EDK-matrix after rotation to direction cosines A: (-0.5, 0.35, 0.8) and B: (0.25,-
0.40,-0.88) relative to the reference coordinate system (0, 0, 1). Dose Driving Voxel
is located at (0,0,0)









4.4.4 Ordinate Projection

In the discrete angle approximations, it is assumed that the EDK-SN is applied on distinct

equal weight angles 04~(8, 9)a(u,r,,(). To generate these angles, the following relations

should hold for any given octant:


AB = (4-10)
order

The Norde in any 3-D EDK-SN quadrature corresponds to the number of levels from each

direction cosine on the unit sphere, and there are Ndir= Norde (Norde + 2) ordinates on the unit

sphere, with Ndir, oct oder rder+ 2)/ 8 in each octant, with Norder/2 distinct direction

cosine values. Figure 4-10 shows Norde 42 used rendering 231 direction per octant.



8= (2i -1) (4-11)


S=( )(4-12)
2 2i

where i = N orde

For any unit angular direction vector 0Z the ordinate must lie on the unit sphere:

pU2 +2 +2 = 1 (4-13)

pu = Sin (0)Cos (p) (4-14.a.)

r = Sin (0)Sin (9) (4-14.b.)

5 = Cos (0) (4-14.c.)




















0.5 Of 00,,0 UU r
S0O 000 00 0 U
CO 0 00







1 1

Figure 4-10. First octant" of S42 equal weight angles used for proj ecting the EDK-SN


4.4.5 Final Dose Calculations

The EDK-SN model specific portion, described in Figure 4-11, serves as the critical link in the

EDK-SN system to accumulate the absorbed dose in each fine mesh. The algorithm determines

the accumulated energy deposited in each voxel for each photon energy group, based on the SN

net current vector and is scaled to the appropriate dose by the photon flux. Finally, the code

accumulates all the absorbed dose contributions from all the energy groups in to a 3-D total

absorbed dose matrix.

Figure 4-11 shows a flow chart representing the EDK-SN methodology. An energy dependent

EDK was generated using high resolution direction proj section sets and a MC pre-computed EDK

for a single direction. These kernels were generated over the unit sphere of ordinates via the

Euler Angles described in Section 4.4.3. Concurrently, when generating the EDK, a

FORTRAN90 module will be generated for later use in the dose calculation portion of the












PENTRAN SN MCNP Electron High Re~solution
Calculations Dose Kernel (ED=K) i n Di re ctio n
One Direction for Projection Set



Electro n Dose
Generation Code!
Using Euler Angles




Correspondent Delta
Module to Define
1 ~Discreet Angles




Parallel EDK- SN Code for 3D-
Dose Calculations EDK Sets for
Each EG and
Each Material




3D- Dose Organ
Di stribtutio~n Do3se -PrCoue
Model Specific


Figure 4-11. A flow-chart for the EDK-SN Code System

methodology. The function of this module is to define a AO on which the net current vector will

be discretized into.

For each specific model, after executing the deterministic calculation using PENTRAN, we

generate 3-D fluxes, current vectors, and material distributions for each energy group for a

particular model. These 3-D data distributions, along with pre-computed EDKs, will then be










implemented to calculate the 3-D absorbed dose distribution, for each energy group, scaled by

the flux along the current vector direction. Finally the EDK-SN will accumulate these individual

energy group dose contributions into a 3D total absorbed dose distribution. In Section 4.7, we

describe the parallel code algorithm.

4.5 External Source Benchmark

An external source problem for benchmarking purposes is presented to demonstrate the

EDK-SN methodology and provide a direct comparison with a full-physics Monte Carlo solution.

The model considered was a homogenous block of water (45 x 45 x 45 cm 3) with a 15 x 15 x 1

cm 3 Source (defined for 16 energy groups spanning from 0.0 to 8.0 MeV cast as a flat-weighted

energy distribution source term). A 3-D spatial mesh distribution was generated by the

PENMSH-XP mesh generating code in the PENTRAN systeml9. The model was then sub-

divided into 3 z-levels; each z-level was then divided equally into 9 coarse meshes, each coarse

mesh contained 3375 fine mesh cells; therefore, a total of 91,125 fine mesh cells were used in the

complete SN problem as shown in Figure 4-12.A.








20g


40 4



A. B

Figure 4-12. External Source impinging on block of water (left: Mesh distribution generated by
PENTRAN package; Right: MCNP5 Cell/Surface schematic input)









For the purposes of direct comparison to the Monte Carlo method, the MCNP5 Monte Carlo

code is used to simulate an equivalent external source problem as shown in Figure 4-12.B. For

validation of the EDK-SN approach, the equivalent MCNP5 model is used to calculate the

absorbed dose along the central axis and across the model for these studies using MCNP5 pulse

height tallies (*F8), with full physics transport (p, e mode) to allow direct comparison of fluxes

and absorbed dose values for the Cartesian fine mesh cells used in PENTRAN-MP and the EDK-

SN approach.

4.6 Comparison of Deterministic and Monte Carlo Results

Using a uniform 1 cm 3 meSh, PENTRAN was used to calculate the flux distributions and

the net current for each Eine mesh cell in the system. Based on the PENTRAN results and the

pre-calculated electron kernels, the EDK-SN code was used to produce 3-D absorbed dose rate

distributions for all energy groups. The absorbed dose rates for energy groups (7.5-8.0 MeV) and

(0.5-1.0 MeV) are depicted in Figures 4-13A and 4-13B, respectively. These Eigures provide a

clear insight of the 3-D absorbed dose behavior for these two different energy groups. For

example, the high energy photons indicate more forward peaked absorbed dose distributions (as

expected), and, while still forward peaked, the lower energy photons indicate more isotropic

behavior, also as expected.

Figures 4-14 and 4-15 depict the absorbed dose rate for each energy group (over a total of 16

energy groups) calculated by EDK-SN along the central axis, and laterally across the problem

respectively. Alternatively, Figures 4-16 and 4-17 show the MCNP Monte Carlo total absorbed

dose along the central axis and laterally across the problem after 1000 minutes using 16

processors of same PC cluster. The statistical uncertainty 20, on average, of the *F8 MCNP5

Monte Carlo tally was 6.0% for the study along the central axis (x-axis), and 4.0% along the y-














































Figure 4-13. External Source 3-D absorbed dose distribution for 2 Energy groups computed by
EDK-SN based on PENTRAN-MP A. EG(7.5-8.0Mev) and B. EG(0.5-1.0Mev)


Dose rate along the X-axis


1 1111 1 _


-.x


I ~0 10 20 Xtcm) 30 40 50


Figure 4-14. Energy group absorbed dose rate distributions (in Mev/ g.sec) along the X-axis
using the EDK-SN method based on PENTRAN-MP


axis, 2 cm away from the source, in the source region. On the other hand, the average absolute


relative differences between the EDK-SN absorbed dose calculation and the full physics MCNP


derived absorbed doses were 3.7% along the central axis (X-axis), and 2.5% along the Y-axis,


2cm away from the source.


6.00E-05


5.00 E-05



S4.00E-05



> 3.00E-05


S2.00E-05
o


-*- D(7~.5-B.0Mev)
-e D(7 0-7 5Mev)
D(6 5-7 OMev)
D(B 0-6 6Mev)
~- --(5 5-60OMev)

D(4 0(.-4 5Mev)


D (1 5-2 OMev)

D(1 O-1 5Mev)

D(O 5-1 OMev)

D(D 0-0 5Mev)


-


`,





x-


- -- -.~ -. .


-


1.00 E-05 -


0.00E+00 -












Dose-rate(along Y-axis)


4.00E-05

3.50E-05


S3.00E-05

~2.50E-05

S2.00E-05

1.50E-05

o 1.00E-05

5.00E-06

0.00E+00


--+- D(7 5-8 OMev)
--* D(7 .0-7 5Mev)
--b-- D(6 5-7 OMev)
D(6 0-6.5Mev)
-- D(5 5-6.0Mev)
-*--D(5.0-5.5Mev)
-- D(4 5-5.0Mev)
----(4 0-4 sMov)
---- D(3 5-4 OMev)
-o-- D(3.0-3 5Mev)
- D(2 5-3 OMev)
~-D(2 0-2 5Mev)
SD(1 5-2 OMev)
D(1 0-1 5Mev)
D(0 5-1 OMev)
tD(O O-o SMev)


0 10 20 30 40 50
Y(cm)


Figure 4-15. Energy group absorbed dose rate distributions (Mev/ g.sec) along the y-axis using
the EDK-SN method based on PENTRAN-MP


Dose rate along the X-axis


4.0E-04

3.5E-04
--Dose using MnCNP5
S3.0E-04 pulse height tally(*F8)
[" 1 I I-*EDK<-SN

6 2.5E-04

r 2.0E-04

1 .5E-04

ij 1.0E-04

5.0E-05

0.0E+00
0 10 20 30 40 50

X(cm)


Figure 4-16. EDK-SN Total absorbed dose rate distributions along the X-axis compared to
MCNP *F8 tally derived absorbed doses. MCNP uncertainties (20) average (6.0%) in
the Source region. The average absolute relative difference between the two methods
was 3.7%






























I vY(cm)

Figure 4-17. Total absorbed dose rate distributions of EDK-SN along the Y-axis compared to
MCNP *F8 tally. MCNP uncertainty (20) average (4.0%) in the Source region.
Average absolute relative difference (2.5%)

Although small, less than 1%, the ripple apparent in lateral absorbed dose, shown in Figure

4-15 and Figure 4-17, can be explained by the mesh structure used in projecting the absorbed

doses along the collective region of electrons, and low level ray effect, in the SN calculation,

causing absorbed dose to oscillate

4.7 Parallelization of EDK-SN Code

EDK-SN is designed to run on distributed memory architectures, where each processor is

an independent unit. This system is a distributed memory computer system, where the processing

elements are connected by a network and tare highly scalable. The interconnection scheme

among the processors is fundamental for distributed memory architectures because it affects, in

part, the performance of the system. For cluster-type architectures, the cluster using identical

processors (matched for speed) connected with an Ethernet local area network to enable data

transfer among the processor units.









To achieve maximum speedup with the EDK-SN methodology, parallel programming

methods were implemented in the EDK-SN code using energy group decomposition via the

Message Passing Interface (MPI)49 library. The MPI library is meant to provide an essential

virtual topology, process synchronization, and communication functionality between a set of

processes (that have been mapped to nodes/servers/computers) in a language independent way,

with library specific syntax (bindings), plus a few features that are library specific.

The EDK-SN code has been designed in ANSI FORTRAN-90 (to take advantage of

dynamic array allocation) and parallelized for operation on distributed memory, multiple

instruction/ multiple data (MIMD) machine architectures using initial standard Message Passing

Interface protocols (MPI-1). All Input/ Output (I/O) is performed by each processor in parallel

(as required) to include input processing, initialization, processing, and Eile output.

The EDK-SN code carries out the computations with electron dose kernels based on the

scalar flux scaling for each energy group independently Consequently, the decomposition used is

energy domain decomposition. The basic philosophy of this approach is to decompose energy

groups to each processor, and carry the total angular (net current vector) and spatial domain

information on each processor.

The main advantages of the EDK-SN parallel algorithm is that appreciable speed up can be

obtained to determine the dose in any voxel throughout the spatial domain. Since little

communication is required between the processors, n processors would solve the problem nearly

n-times faster than a single unit. Energy group absorbed doses are partitioned for the entire

problem, and partitioned energy group assignments are scaled to n processors for coupling

photon transport information with EDKs to produce group absorbed doses over the spatial

domain. An MPI REDUCE call is made to collect all absorbed doses across processors. Since









this work focused on the use of 16 energy groups for a flat weighted 8 MeV high energy source

term, the parallel EDK-SN computation could be parsed to a maximum of 16 parallel processors.

Near linear speedups were achieved.

The parallel code follows the following steps in executing the EDK-SN methodology for

total dose calculation:

Initiate the code on all processors for sets of energy groups, up to one group per
processor
Sort the locally parallel-allocated fine mesh distributions into a global monotone
increasing mesh distributions
Create Energy group dose matrix aliased to the location of the DDVs
Translate material tags into densities for later calculation
Project the Group-dependent EDK based on the group current vector and material at the
DDV
Scale group dose using group scalar flux, correcting for densities at material interfaces
Use MPI-REDUCE to sum all doses for the phase space across the parallel processing
grid

4.8 Discussion and Findings

We examined a new EDK-SN methodology of coupling pre-computed Monte Carlo based

electron absorbed dose kernels (EDKs) with the discrete ordinates (SN) solutions to achieve

accurate absorbed doses for megavolt energy photon scenarios, since charged particle

equilibrium fails and full photon-electron physics must be accounted for to fully attribute

absorbed dose. The EDK-SN methodology was compared with the Monte Carlo method and

shown to give results consistent within the statistical uncertainty of the absorbed dose computed

using Monte Carlo calculations.

By following the EDK-SN approach, the absorbed dose applied to tissue from any

incident external high energy photon beam (pencil, fan, areal, etc) can therefore be readily

determined based on a coupling of the deterministic SN photon transport solution from the

PENTRAN code using the EDK-SN methodology. As a result, this research renders the first









application of integrated electron transport kernels coupled to the SN photon transport

methodology to properly account for the total absorbed dose in high energy photon irradiation

scenarios. A fundamental difference between the deterministically guided EDK- SN technique

and the existing convolution-superposition technique is the fact that our EDK-SN approach is

based on a full physics Monte Carlo generated kernel projected using an accurate SN transport

solution that provides a net current direction and flux scaling to scale the absorbed dose using the

spectrally dependent EDKs. Overall, based on the water phantom problem examined here, with

an adequate discretization using the SN method coupled to properly computed energy dependent

electron kernels, it is possible to produce absorbed dose results with the EDK-SN method that are

consistent with Monte Carlo calculations. A major advantage of the EDK-SN method is that it

provides a detailed, accurate absorbed dose distribution wherever required throughout the

system, while incorporating the full physics electron transport, and is generally much faster than

Monte Carlo. Since PENTRAN is based on a Cartesian geometry, CT voxel data can readily be

mapped onto the problem geometry to define the tissue and organ geometry structure.

Subsequent discussion will account for material density variations (encountered at tissue-bone

interfaces) where we will apply the EDK-SN methodology to a whole body voxelized phantom to

rapidly achieve a full physics whole body, 'total body' absorbed dose.









CHAPTER 5
APPLICATION OF EDK-SN TO MODELS WITH HETEROGENITIES

This chapter focuses the many aspects that make up the EDK-SN methodology in order to

apply it to a voxelized phantom, validated side-by-side using full physics Monte Carlo

calculations. The SN calculations are performed based on a full photon transport calculation

using deterministic techniques in a heterogeneous phantom. This eliminates any need for any

photon corrections for tissue heterogeneities, since the SN solution yields the complete solution

to the Boltzmann transport equation. Moreover, because the EDK-SN calculation is performed

over the full spectrum of the radiation for a discrete number of energy groups, we can thereby

generate separate sets of electron absorbed dose kernels to account for the different densities

present in the phantom model. In this chapter, we will address applying the EDK-SN method to a

heterogeneous phantom, including the application to a new state of the art human phantom.

5.1 Heterogeneous Tissue-Equivalent Phantoms

Three 'slab based' phantoms were simulated, using the EDK-SN method, as a prelude to

more complex calculations for full human phantoms. Each phantom had a flat weighted 0 to 8

MV x-ray source impinging on 45x45x45 cc soft tissue (S) phantom. Two of these phantoms

have slabs of 1-cm-thick discs of materials representative of bone (B), or lung tissue (L) (Figure

5 -1).

5.1.1 Monte Carlo Simulations

In order to compare the accuracy of the EDK-SN methodology, a series of Monte Carlo

simulations were also performed using the radiation transport code MCNP5. An X-ray flat

weighted spectrum 8 MV used as input to MCNP5 to guarantee no specific preference was

giving to any particular energy group. Cell estimates of tissue absorbed dose were calculated in

MCNP5 by tallying the energy deposited by the photons and their secondary electron within 1










cm3 VOxels, equivalent to the size of the EDK-SN voxel and the voxel size to be used in the SN

calculation.


X-ray Beam




SSSSSSS SSBBSSS SBLLBSS












Soft Tissue Lung Tissue Cortical Bone

Figure 5-1. Three Monte Carlo phantoms used in the absorbed dose study. Phantom SSSSSSS is
a homogeneous soft-tissue (' S') phantom; phantoms SSBBSSS and SBLLBSS are
heterogeneous phantoms consisting of combinations of soft and bone tissue ('B'), and
soft (' S'), bone ('B'), and lung ('L') tissue-equivalent materials, respectively

After running the three phantoms for 100 minutes, Figure 5-2, using MCNP5 photon,

electron mode and pulse height tallies (*F8) the results uncertainties (20) were 20% on average.

A 'continue run' was executed for the three phantoms for 400 minutes more, for 500 minutes

total, reducing the Monte Carlo uncertainty (20) down to 10% on average (Figure 5-3). However,

the Monte Carlo computed difference in the deposited dose between the two tissues is within 2%,

making it impossible to observe any difference between the different tissues in the phantoms,

especially for therapeutic photon beams; this is shown in (Figure 5-4). The estimated time

required to reduce Monte Carlo uncertainty to (20) 2% on-axis was 12500 minutes; hence, the

time required to reduce the Monte Carlo uncertainty off-axis to 2% will be much greater.










Therefore, this is an example that highlights that tissue heterogeneities can be difficult to discern

in these problems using the Monte Carlo method.

As explained previously, one of the most important issues limiting the use of the Monte

Carlo techniques for clinical absorbed dose calculations is achieving a statistically reliable result

in a reasonable time. This problem is even greater when a photon-electron mode is required,

when CPE is not applicable, causing computer running times needed to reduce the statistical

error to increase rapidly. Improved variance reduction techniques may reduce the computation

times. However, such techniques are to be used with caution, as improper application of

statistical variance reduction may cause wrong answers. Moreover, variance reduction

techniques may not reduce the time required to reduce the variance to acceptable levels,

especially when accurate absorbed dose estimations are needed over the entire problem.


100 Minutes (Pulse height tally using photon-
4.00E-04 -electrn mode)

3.50E-04 --*-- SBLLBSSS
Ilu -o- SSBBSSS
(0 3.00E-04

as 2.50E-04-


1.50E-04-


ur1.00E-04 -

5.00E-05-

0.00E+00
O 10 20 30 40 50
X(cmR)

Figure 5-2. Monte Carlo derived total absorbed dose rate distributions along the X-axis
computed using *F8 MCNP5 tally for three different phantoms. MCNP5 uncertainty
(20) average (20.0%), 100 Minute run





I ~X(cm)

Figure 5-3. Monte Carlo derived total absorbed dose rate distributions along the X-axis
computed using *F8 MCNP5 tally for three different phantoms. MCNP uncertainty
(20) average (10.0%), 500 Minutes run


Pulse height tally using photon-electron mode
(SBLLBSS)
4.00E-04
3.5
3.50E-04
-*-MC-Dose
a 3.00E-04 -
-6- Material 2.5
E 2.50E-04 n
52.00E-04 Z 2~

1.50E-04 1.5



5.00E-05 -g

0.00E+00
0 10 20 30 40 50
X~rcm)

Figure 5-4. Monte Carlo derived total absorbed dose rate distributions along the X-axis
computed using *F8 MCNP5 tally for the _SBLLB SS_ phantom. MCNP uncertainty
(20) average (7.0%), 1000 Minutes run


500 Minutes (Pulse height tally using photon-
electron mode)


4.00E-04 -

3.50E-04 -

3.00E-04 -

2.50E-04 -

2.00E-04 -
~.50Ea4-

1.50E-04 -


5.010E-05

0.00E+00 -


-*-- SBLLBSSS
-( SSBBSSS
-8- SSSSSSS









5.1.2 PENTRAN Model

As shown in Figure 5-5, we simulated equivalent three phantoms to those simulated with

Monte Carlo. The heterogeneous models consisted of three key components; an X-ray source

(15 x1 x15 cm3) with flat weighted 8 MV source, divided into 16 energy groups, a soft tissue

phantom (45x45x45 cm3), eqUIValent layers of bone and lung tissue identical to that used in the

Monte Carlo simulations. Models were divided into 3 z-levels; each z-level was divided equally

into 9 coarse meshes containing (15 x15 x15) fine mesh cells for each coarse mesh; a total of

91,125 fine mesh cells were used in the SN problems. Although lower quadrature set (S32) aef

sufficient to produce accurate SN results for similar calculations (Chapter 1), to assure minimum

ray-effects on the SN calculations, these simulations were based on the use of an S62 angUarT

Legendre-Chebychev quadrature with P3 Scattering anisotropy using the CEPXS library.


















SSSSSSS SSSBSSS SBLLBSS


Figure 5-5. Three SN phantoms equivalent to the one used in the Monte Carlo simulation.
Phantom SSSSSSS is a homogeneous soft-tissue phantom; phantoms SSBBSSS and
SBLLBSS are heterogeneous phantoms consisting of combinations of soft and bone
tissue, and soft, bone, and lung tissue-equivalent materials, respectively










5.1.3 EDK-SN Absorbed dose Comparison

The SN calculation was performed based on a full photon transport calculation using

deterministic techniques in the heterogeneous phantom. In principle, this eliminates any need for

any correction for tissue heterogeneities which affect the different scattering components in

differing ways. Moreover, the EDK-SN calculation was performed for material specific absorbed

doses. This allowed the use of the pre-computed electron absorbed dose kernels for the different

densities present in the phantom model.


Pulse height tail using photon-electron mode
(SBLLBSS)
4.00E-04

3.50E-04
-+- MC-Dose 3
3.00E-04 n-> Myaterial
-EDK-SN2.
2.50E-04 -r


1.50E-04 1.5

S1.00E-04 1.


5.00E-05 -I I a9

0.00E+00
0 10 20 30 40 50
X(cm)

Figure 5-6. EDK-SN heterogeneous SBLLBSS phantom total absorbed dose rate distributions
along the X-axis compared to MCNP *F8 tally. MCNP uncertainty (20) average
(8.0%) on-axis. Average absolute relative difference (4.1%)

For the phantoms used in this study, calculations were performed to compare the EDK-SN

and Monte Carlo pulse height tally results for the absorbed dose deposited along the model

central axis. As shown in the Figures 5-6 and 5-7, the EDK-SN results are in very good

agreement with the MCNP5 results, and differences between the two are within the statistical

uncertainty of the Monte Carlo calculation. The Monte Carlo uncertainties, on the central axis of









the source, were 4% on average while the average absolute difference at the same cells was

3.7%.


ED]K Va. Pulse height tally using photon-electron
4 00E04 -mode (SSSSSSS)

S3.50 E-04

n 3 00 E-04-
E -e EDK-SN
Ea 2.50 E-04
~5;-*-MC-Dose
2 00 E-04-

QL 1.50 E-04

2u 1.00 E-04-
a 5 00 E-05

0.00 E+00- ,,
O 10 20 30 40 50
X(cm )

Figure 5-7. EDK-SN Homogenous phantom total absorbed dose rate distributions along the X-
axis compared to MCNP *F8 tally. MCNP uncertainty (20) average (6%) on-axis.
Average absolute relative difference (3.7%)

We note that at the interface lung region, Figure 5-6, the EDK-SN derived absorbed dose

indicated a lower absorbed dose than Monte Carlo. This can be related to the application of the

lung electron kernel over the bone and soft tissue regions. This phenomenon was mitigated by

applying a correction factor for voxels located at the interface of different material densities.

5.1.4 Correction Factor Approach

The range of a charged particle of a given type and energy in a given medium is the

expectation value of the path-length that it follows until it comes to rest. On the other hand, the

RCSDA TepreSents the range in the continuous slowing down approximation (CSDA). Figure 5-8

gives the CSDA range RCSDA for electron in cortical bone, soft tissue and lung tissue for electron

energies between 0.0 MeV to 17.5 MeV. Figure 5-9 depicts a comparison between the ratio of










the CSDA ranges for cortical bone to soft tissue, lung to soft tissue, and cortical bone to tissue.

One can notice that the behavior of electrons in the different tissues is quite similar for the same

electron energy. This implies that there is a possibility of applying a simple correction factor

within different tissues to account for the different densities in the phantom for each energy

group.

Unlike other methodologies, this is possible due to the dependence of the EDK-SN on a full

photon physics transport calculation accomplished by the SN computation to provide the

scattered photon information and a directional (current) component.


Figure 5-8. Continuous slowing down approximation range (Rcsda) for electrons in Cortical
Bone, Soft Tissue and Lung as a function of energy











Resda Co mparison
(Cortical Bone, Soft Tissue, Lung)
1.20E+00


1.00E+00


8.00E-01

-*-- Tissue/Bone
6.00E-01 ---Tissue/Lung
u --6- Bone lung
4.00E-01


2.00E-01


0 .00E+ 00
0 5 10 15 20
Electron Energy(MeV)


Figure 5-9. Continuous slowing down approximation range (RCSDA) ratio comparison for
electrons in Cortical Bone, Soft Tissue and Lung as a function of energy

The key to this approach is that photon transport is carried with the 3-D- SN code

PENTRAN. The use of such code provided us with the directional component of the photon

transport in the homogenous medium. By having this information we were able to project the 3-

D electron kernel along the net current direction. The assumption of a homogenous medium in

photon transport in a heterogeneous problem, often used in other methods, may lead to an

underestimation of the scattered component of the photon beam, and misrepresentation of its

directional component, especially between the different organs of the human body (Figure 5-9),

leading to misrepresentation of the absorbed dose.

5.2 Computational Phantom of 15-Year Old Male

A new set of stylized phantoms have been developed by the UF Advanced Laboratory for

Radiation Dosimetry Studies (ALRADS). These phantoms represent a totally new class of

whole-body patient hybrid phantoms so named as they retain the non-uniform scalability of










stylized phantoms, while maintaining the anatomic realism of segmented voxel phantoms with

respect to organ shape, depth, and inter-organ positioning to great accuracy. One study showed

that these phantoms match (to within 1%) to that of a newborn of the ICRP Publication 89

reference newborn child.

In this study, hybrid computational phantoms representing reference 15-year male body

anatomy and anthropometry is used, shown in Figure 5-10. This phantom was matched, by the

ALRADS group, to anthropometric data and reference organ mass data provided by the ICRP,

where it matched to within 1% overall"


Figure 5-10. Frontal views of 3-D rendering of the 50th percentile UFH-NURBS 15-year male
and female

The availability of hybrid phantoms as a replacement for voxel-based phantoms provides

the needed platform for constructing phantoms representing the unique internal and external









anatomy of individual pediatric patients. However, this process would be complicated

tremendously if each individual organ of the phantom had to be adjusted to match patient-

specific organ volumes as seen under CT imaging, even when those images are available.

The 15-year-old male was voxelized with high resolution using 2 mm x 2 mm x 2 mm

voxel size (302 voxel x 139 voxel x 836 voxel) as a case study. This geometry data was then

collapsed into a 1 cm x 1 cm x 1 cm voxel size (60 x 27 x 167 voxels), for total of 270540

voxel-model using GHOST-3-D. Subsequently, using PENMSH-XP, the collapsed model was

cast into a 3-D spatial distribution, and sub-divided into 10 coarse mesh z-levels containing 18

coarse meshes in x-y domain, and these contained a number of fine mesh cells (Figure 5-11). An

8 MV flat weighted, forward peaked volumetric (20 cm x 1 cm x 17 cm) X-ray source was

placed above the chest area of the phantom.













Fiur 51. iml tio mehoolg fo D- opttos sn ETA-Pcd
sytm et F nhomrhc hnomdle orsodntPNRNipt






Figre -1 ighlt:o EDK-Swg 3- borbe dose distriutinatrradations withg Pa x-ray soure










Although the model has relatively small fine mesh sizes, using Legendre-Chebychev S32

quadrature, the DTW differencing scheme, and CEPXS cross-sections with anisotropy level P3,

we were able to eliminate any unphysical oscillations that may result in the deterministic

solution. Using 16 energy groups, the SN model was executed on 16 processors and completed in

2.0 hours.

With the general purpose radiation transport code MCNP5, we simulated an equivalent

model to the one simulated using PENTRAN. The collapsed 3-D binary files were converted to

ASCII formatted matrix files and then incorporated into an MCNP5 lattice defined input; this

matrix was also generated by our down-sampling code GHOST-3-D.

5.2.1 EDK-SN Absorbed Dose Calculations

Calculations were performed to calculate the SN results for the scalar fluxes for all energy

groups. Using EDK-SN, 3-D absorbed dose distributions were calculated to each voxel in the

entire 15 year old male phantom model for a broad-beam external photon field. Figure 5-11

shows the stages of calculation in the PENTRANP-MP code system that resulted in 3-D

Absorbed dose distributions for the entire phantom. Figure 5-12 shows absorbed dose

contributions from selected energy groups, and the total absorbed dose resulting from adding all

energy group-absorbed dose contributions, as prescribed by the EDK-SN procedure.

Organ absorbed doses using EDK-SN, for a selected number of organs, were calculated and

compared to Monte Carlo predictions, and are presented in Table 5-1. The total execution time

for the Monte Carlo was ~4 hrs (wall clock time) on 16 processors, while the total wall clock

execution time for the EDK-SN was ~ 2hrs on the same number of processors. As anticipated, the

absorbed doses for the organs located directly in the x-ray field (right+ left lung) or partially in

the x-ray field (liver) were in good agreement for both the EDK-SN and MCNP5 pulse height

tally data.





































Figure 5-12. Absorbed dose contribution of each energy group is calculated via the EDK-SN
procedure, then folded to calculate the total absorbed dose delivered to the 15 year
old male phantom from 8 MeV flat weighted source. The original phantom was
collapsed into 270K voxels

Table 5-1. Comparison of selected organ absorbed dose rate (MeV/g. s) calculated using
MCNP5 pulse height tally with (photon, electron mode) and EDK-SN for the UF
hybrid 15-year-old male phantoms for a chest 8 MV X-ray flat weighted source


Si.


EG1


EG4 EG7


EG16


EG10


EG13


TotalI-Dose


Organ


M( F8)
(MeV/g.Sec)
1.35E-01
9.47E-05
3.56E-05

1.86E-04
1.41E-05


2a-MC
Uncertainty
6.80%
4.14%
4.00%
3.00%
4.60%
6.82% ~


EDK-Su
(MreV/g.Sec)
1.41E-01
9.85E-05
3.75E-05
1.18E-04
1.94E-04
1.38E-05


(MCI-EDK-
SN)/EDK-SN
4.56%
4.02%
5.43%
6.58%
4.09%
2.08%


Right+ Left Lung
Pancreas
SI W
Spleen
Stomach W
Thyroid












On the other hand, Monte Carlo calculations for organs outside the x-ray field (such as the

prostate) showed significant statistical uncertainty (44%). This poor absorbed dose tally result is

due to the fact that the Monte Carlo did not have enough histories for these results to converge to

low statistical uncertainty in an acceptable time.

Alternatively, as shown in Table 5-2,when running the Monte Carlo for extended running

times (~40 hours on our parallel cluster), the Monte Carlo calculated absorbed dose delivered to

the prostate began to show convergence to the absorbed dose calculated using EDK-SN. This

illustrates that the EDK-SN methodology, executed in 2 hours (SN execution time = 1.5 hr, EDK-

SN execution time = 0.5 hr), is capable of providing an accurate solution and a significant speed

up for a full physics absorbed dose calculation in human phantoms for high energy external

beams.

Table 5-2. Prostate absorbed dose rate (MeV/g.s) calculated for the test problem using EDK-SN
and MCNP5 using pulse height tally (*F8) with (photon, electron mode) at different
MCNP5 Monte Carlo running times

Dose 2o--MC (EDK-SN- Running
(M~ev/gm.Sec) Uncertainty MC)/EDK-SN Time
ED)K-SN 2.29E-08 ----- ----- 2hr
MC1t 2.21E-08 44.00% 3.62% 3.9hr
MC2 2.34E-08 39.20% 2.14% 4.7hr
MC3 2.18E-08 34.40% 5.05% 6;.1hr
MC4 2.23E-08 22.20% 2.69%/ 14.5hr
MC5 2.46E-08 16.86;% 6.91% 28.15hr
MC6 2.406E-08 14.14% 4.58% 38.85hr


5.2.2 Ray-Effects

When considering a forward peaked volumetric (10 cm x 1 cm x 17 cm) X-ray source

placed on top of the left lung area of the phantom, we noted significant differences between the

absorbed dose calculated by EDK-SN and the absorbed dose calculated by full physics Monte









Carlo calculations to the right lung. This behavior can be explained by the ray-effects in the low

density lung tissue (see Chapter 2). Since the X-ray source is highly directional, and the organ

(right lung) is of a low density material, and little primary radiation and very low scattered

radiation is streaming through the right lung, making conditions ideal for inducing ray effects in

SN computations for the photon component.

This is mitigated by using high order quadratures, and can reduce absolute differences;

these were reduced to 4% (Table 5-2). However, the increase of the quadrature level costs

additional running time, with an increase from 2 hours to 9 hours. This increase in quadrature

level was necessary for large low density volumes that experience low levels of scattered

radiation. Another method, ordinate splitting, allows for a biased quadrature set to focus on the

"important" directions, although the technique was not used here.

Table 5-2. Absorbed dose rate (MeV/g. s) calculated for the test problem using EDK-SN and
MCNP5 using pulse height tallies (*F8) with (photon, electron mode)
Organ MC(*F8) 2o--MC EDK-SH (MC-EDK-
(p, e mode) Uncertainty SN)/EDK-SN
Right Lung-(Sn32) 5.62E-06 2.0% 6.96E-06 19.33%
Right Lung-(Sn62) 5.62E-06 2.0%/ 5.39E-06 4.03%
Left Lung 2.70E-04 0.40% 2.79E-04 3.0%
Liver 1.00E-05 3.0% 1.02E-05 1.54%
Pancreas 1.11E-06 4.0% 1.17E-06 5.1%


5.2.3 MCNP5 Energy Deposited Tally (F6) and Pulse Height Tally (*F8) Results

When comparing MCNP5 organ absorbed dose results, for the 15-year old phantom,

calculated using energy deposited tally (F6) and pulse height tally (*F8), we noted

inconsistencies in the results of the two tallies (Table 5-3). This is in keeping with our earlier

investigation of the two tallies earlier in Chapter 4. This inconsistency in the F6 tally suggests

that full physics Monte Carlo calculations are essential for accurate absorbed dose estimation in

high energy external beam absorbed dose calculations.









Table 5-3. Absorbed dose rate (MeV/g. s) calculated for the test problem using MCNP5 using
pulse height tally *F8 with (photon, electron mode) and MCNP5 energy deposited
tallies (F6)
Organ MC(*F8) 2o--MC MC: (F6) (*F8-F6t)/F6t
(p, e mode) Uncertainty
Right Lung 5.62E-06 2.0% 4.28E-06 31.3%
Left Lung 2.70E-04 0.40% 2.74E-04 1.54%~
Liver 1.00E-05 3.0% 9.14E-06 9.41%
Pancreas 1.11E-06 4.0% 1.03E-06 7.84%

Wang et al. conducted a similar study using their VIP Man organ MCNP for absorbed dose

comparison for the three different tally methods including cell flux tally (F4), energy deposited

tally (F6) and pulse height tally (*F8). In their comparison they found that the F4 tallies agree

within 10% with the F6 tallies, while the differences between the tallies F8 and F6 are relatively

large (28%) 14. These results agree with our findings, and provide a strong proof for the need of

applying full physics transport when calculating whole body absorbed dose.

5.3 Discussion and Findings

To properly treat 3-D electron transport physics deterministically, yet still to achieve

reasonably fast and accurate whole body computation times using high energy photons, we used

angular and energy dependent electron transport "absorbed dose kernels." In a methodology

called 'EDK-SN' we have demonstrated that by using different electron kernels for different

materials (cortical bone, soft tissue and lung), the EDK-SN methodology incorporating a discrete

ordinates (SN) approach is an effective alternative to the Monte Carlo method for accurately

estimating the absorbed dose for high energy photon beams over an entire human phantom. In

doing so, we simulated three phantoms using MCNP5 and PENTRAN-MP. Each phantom had a

flat weighted 8 MV x-ray source impinging on 45x45x45 cc soft tissue phantom. Two of these

phantoms were slabs of 1-cm-thick discs representative of bone, or lung tissue. Comparisons of

MCNP5 Monte Carlo based dose results and those computed using PENTRAN-MP

implementing the EDK-SN methodology showed good agreement between the two approaches.









However, the Monte Carlo simulation took 100 minutes for the MC (20) uncertainty to be 20%

(on average) for tally points along the central axis of the phantom using an MCNP5 *F8 tally

(we note here that 20% uncertainty can't be viewed as converged). Additional runs totaling 500

minutes were made, reducing the Monte Carlo (20) uncertainty to 10%, on average. We

estimated the approximate time required to reduce Monte Carol (20) uncertainty to 2% for the

on-axis locations (relative to the source X-ray beam) to be 12,500 minutes. Undoubtedly, the

time required to reduce the Monte Carlo uncertainties at locations off-axis from the source beam

to 2% will be much greater.

Based on the UF hybrid phantoms developed by the University of Florida ALRADS group,

we simulated a complete phantom discretized using 270,540 voxels, and demonstrated that; in

this case, the EDK-SN method readily provides accurate whole body 3-D absorbed dose

distributions. We were able to accurately yield doses analogous to the MCNP5 Monte Carlo

calculations. We further noted that ray-effects were evident in regions of the tissue-lung

interface, which indicate higher angular quadratures or ordinatee splitting' treatments are

necessary in lung tissue and low density tissue interfaces. Also, tally results from the 16 energy

groups, used to represent the radiation spectrum from the CEPXS multi-group cross-section

library were compared to continuous energy MCNP5 Monte Carlo results using the ENDF/B-VI

data libraries.

A major advantage of the EDK-SN method is that it provides a detailed, accurate 3-D

absorbed dose distribution at thousands of cells throughout the system, while the Monte Carlo

method only provides highly accurate values for selected points near the source. Though the use

of volumetric "mesh tallies", Monte Carlo calculations can provide global absorbed doses across

a grid spanning the Monte Carlo geometry. However, at present, absorbed doses for this data are










based on assuming charged particle equilibrium conditions, not suitable for high energy beams.

Moreover, the statistical uncertainty associated with each mesh tally site for such computations

would require extended periods of running time, especially off-axis from the source and outside

the region covered by the primary beam to converge to acceptable levels. These factors limit the

Monte Carlo calculations from being fuuly adequate in practical applications where whole body

and organ-specific absorbed dose information is needed. Notably, an equivalent Monte Carlo

calculation to obtain the flux distribution with statistically reliable results over the entire problem

space may require orders of magnitude longer running times compared to parallel deterministic

calculations.









CHAPTER 6
SUMMARY, CONCLUSIONS AND FUTURE WORK

This work began with the primary goal of developing a new parallel 3-D Dose calculation

tool based on deterministically rendered high energy photon transport solutions in radiotherapy

settings.

6.1 Conclusions and Findings

In this dissertation, the EDK-SN method is presented for generating detailed absorbed dose

information using high energy external beam source terms. Because detailed whole body and

organ-specific absorbed doses can be accurately obtained throughout a voxelized phantom with

proper electron interaction physics projected onto the phantom voxels, based on SN-computed

photon fluxes and net current vectors, this highlights the strength of the EDK-SN transport based

methodology. Overall, based on the problems examined, with an adequate discretization

implemented using the EDK-SN method, it is possible to produce absorbed dose results

consistent within the statistical uncertainty of MCNP5 Monte Carlo calculations.

Overall, our studies examined the capabilities of the discrete ordinates method compared

with the Monte Carlo method with application to medical physics problems. Since radiation

transporting in such applications encounters few interactions in low density material, such as air,

attention was given to minimum quadrature levels and mesh size distributions necessary for

solution accuracy in air and in a water phantom. Also, it is evident that multi-group cross-section

libraries must also be carefully tested for problem applicability. Of particular importance was the

flexibility of the adaptive SN differencing scheme in PENTRAN, permitting more accurate

spatial differencing schemes in cells with a range of optical thicknesses. In addition,

deterministic models based on the UF series B pediatric phantoms and with the PENTRAN-MP









methodology, we were able to accurately yield a solution analogous to the MCNP5 Monte Carlo

calculations using the CPE principle for low energy beams. .

We have demonstrated; that by using electron kernels for the different materials (cortical

bone, soft tissue and lung) the EDK-SN methodology incorporating a discrete ordinates (SN)

approach is an effective alternative to the Monte Carlo method for accurately estimating the

absorbed dose for high energy photon beams over an entire human phantom. In doing so, we

simulated three phantoms using MCNP5 and PENTRAN. Each phantom had a flat weighted 8

MV x-ray source impinging on a 45x45x45 cc soft tissue phantom. Two of these phantoms have

slabs of 1-cm-thick discs of materials representative of bone, or lung tissue. The comparison of

the MCNP5 results and the EDK-SN showed good agreement between the two methodologies.

We also simulated the complete human phantom using 270,540 voxels, and demonstrated

that the EDK-SN method readily provides accurate whole body 3-D absorbed dose distributions.

The angular data provided by the SN method was used to guide the absorbed dose mapping

process for secondary electrons to the surrounding voxels; this makes our approach truly novel.

We were able to accurately yield a solution analogous to the MCNP5 Monte Carlo calculations

in the voxelized phantom. We note that that tally results from the 16 energy groups, used to

represent the radiation spectrum, from the CEPXS multi-group cross-section library, were

compared to continuous energy MCNP5 Monte Carlo results using the ENDF/B-VI data

libraries.

Parallel deterministic PENTRAN-MP results were obtained within comparable running

times to parallel Monte Carlo calculations for tally sites adjacent to the source. An equivalent

Monte Carlo calculation to obtain the absorbed dose distributions using photon, electron mode

with pulse height tallies converged to reasonable statistical uncertainty over the entire problem










space may require orders of magnitude longer running times compared to the time needed for

complete, whole body parallel deterministic calculations. The work presented in this dissertation

has extended the capabilities of the discrete ordinates method to accurately calculate radiation

absorbed dose contributions in medical physics procedures using the PENTRAN-MP code

system with the EDK-SN methodology.

6.2 Future Work

The new EDK-SN, integrated into the PENTRAN-MP Code System, represents a leap

forward in computational medical physics; large 3-D radiation transport calculations for high

energy photons and absorbed dose contributions can now be performed accurately in a fraction

of the computation time required in the Monte Carlo methodology. The methods described in

this dissertation can be further enhanced and developed by studying the following issues:

* Additional parallel performance assessments.
* Examine performance for solving additional real-world problems.
* Optimization study on the EDK-SN. This should be undertaken, in order to identify the
necessary level of accuracy which yields the best performance in terms of speed-up.
* Test alternative method for applying correction factors in tissue inhomogeneities.
* Conduct timing and performance studies with the new hybrid PENTRAN code implementing
Discrete Ordinates and Method of Characteristics.
* Apply for an NIH R0 1 grant.


Long-term plans are to implement PENTRAN-MP as a secondary treatment planning

software for documentation of both in-field and out-of-field organ absorbed doses to normal

tissues. This information will be extremely valuable to radiation epidemiology studies which are

currently being conducted using far more simplistic dosimetry schemes, particularly to out of

fields organs. This data will also be of use to future optimization studies where present goals of

tumor absorbed dose conformality and near-field absorbed dose avoidance to normal tissues will

be supplemented by the additional goal of secondary cancer risk minimization in out-of-field

tissues and organs.



























































Table A-1. Ratios of photon interactions in water for range of energies
Photon Photon Energy (MeV)
Inercton 2.0 3.0 6.0 10.0 14.0 18.0 20.0
% of Incoherent
99.21 % 97.15 % 88.59 % 77.06 % 67.32 % 59.44 % 56.04 %
scattering
% of Nuclear
0.79 % 2.82 % 10.78 % 21.18 % 29.76 % 36.69 % 39.64 %
P.P
% of electron
0.0 % 0.03 % 0.61 % 1.76 % 2.9 % 3.90 % 4.35 %
P.P


APPENDIX A
HIGH ENERGY PHOTON TREATMENT (IN CONDENSED ELECTRON TRANSPORT)

The photons energy used in radiotherapy treatment planning are 20 MeV or less. Photon

interactions that take place at these high energies predominantly Compton Effect (C.E.) or/and

Pair Production interactions by different percentages, as shown in Figure A-1 and Table A-1.


Att. Coefficient WATER (cm2/g)


+ Scatt. Coh.(cm2/g)
+e Scatt. Incoh.(cm2/g)
P.E.A(cm2/g)
P.P Nuc. Field (cm2/g)
+r P.P Elec. Field (cm2/g)
+ Total w/ coh
+e Total w/o Coh.


1.00E+04


1.00E+02


S1.00E+00

E


o
1.00E-04


1.00E-06


1.00E-08
1.00E-03


1.00E-02 1.00E-01 1.00E+00

Energy(MeV)


1.00E+01 1.00E+02


Figure A-1. Total mass attenuation coefficient for photons in water, indicating the contributions
associated with phoelectric absorption, Compton scattering and electron-positron pair
production









Compton Scattering:

When passing through matter, photons can interact with the electrons of the matter. The

Compton scattering describes an inelastic scattering of a photon with a free electron. If the

electron' s initial binding energy is small compared to the photon energy (En << hV ) then the

stuck electron can be considered initially free and unbound. Figure 10 shows this inelastic

scattering process. The kinematics of this scattering (conservation of momentum and energy)

lead to the energy of the scattered photon hv and to the energy of the scattered electron Es:







E = hv Y p~~


hv=c, v
PY hy'l',Y




Figure A-2. Sketch of Compton scattering, photon of energy hv collide with an electron. The
photon's incident forward momentum is hvl/c, where c is the speed of light in
vacuum



Based on the unbound and stationary condition of the electron, the Compton relation can

be obtained by solving energy conservation equations

T = hv hv (A-1)



And conservation of momentum










hv hv
-= -cs+ p cos 8e (A-2)
c c

hv sin B = pc sin 8e (A-3)

pc can be written in terms of T by invoking the "law of invariance":


p~c= T(T+moc2 (A-4)

In which mo is the electron's rest mass.

As a result of solving these equations algebraically, one can derive the following

Compton's relations

hv
hv = (A-5)
1+(hvlm0C2)(1-COS 8)


co~ 1hymC)a (A-6)
21


T, = hv hv = hv 1-+2 i2 Bi2 (A-7)


To obtain the fraction of the scattered photons in a given direction, Klein and Nishina

(reference) have carried out a quantum-mechanical treatment of the problem using the Dirac

equation for the electron and have obtained the equation


derKN 02 hy'hv hv .
~ -v +-si2 (A-8)
da 2 hv hv h


Equations similar to those for the Compton photon distribution can be obtained for the

Compton electron distribution. Since the probability that an electron will be scattered into the

solid angle daZ situated in the direction 8e is the same as the probability that a primary quantum









will be scattered into the solid angle daZ in the direction 6, B and being related by Eq. (A-

8), we have that

(A-9)
daZ daZ daZ

daZ d~ sin0 d8
(A-10)
d~Ze de sin 8e dBe


cot 8e = (1 +a) tan( ) (A-1 1)

-1 1
= (1+ a) dB (A-12)
sin2 8e COS2 8/2


sin 8e = 1l+(1+a)2 tan208/2 1/2 (A-13)

e = 27r (A-14)


de = -dj > = -1 (A-15)


daZ 1 (1 +cos8) sin 8
(A-16)
d~Ze (1+ a) sin3 8e

for 0.511 MeV photons, a = 1

S= 2 sin(8/ 2) 1+3 sin2 (6 /32) (A-17)
dae,



d~Ze 2 hv hv h v-8



























e*ink *10""o cmPEleckton


Figure A-3. Compton effects differential cross section per unite angle for the number of
electrons scattered in the direction 8e

Example

Considering a Compton scattering of 1.022 MeV (a = 2 ), the ratio of recoil electrons with

energies above 0.511 MeV to recoil electron with energies below 0.511 will be:

T hv hv a(1 pu)
S0.5 (A-19)
hv hv 1+ a(1- pu)


pu = 0.5 B = 600 (A-20)


cot 8e = (1 +a) tan (A-21)


(8e) = 300



0=180
de" 2sin Bd6
R = 0=60 (A-22)
de" 2sin 0 dB
0= doda











-IK~~ rO22 (+ 2 -a 2)/e+(1/a2 2)+(1 +2a*)e2/) (A-23)


E = (A-24)
1 + a(1- pu)

pu= cos9 ->~dpu= sin Bd6 (A-25)

dpu = dE (A-26)

hv
E = (A-27)
hv

From this example one can note that; the produced electron in a certain direction has a

unique energy for a giving initial photon. So for an angle aGZ aroundiscretized range of angles

we can calculate the average energy of the electrons produced in th

Pair production

Figure illustrates schematically a pair production event in the nucleus electric field. The

incident photon gives up all of its energy in the creation of an electron positron pair with kinetic

energies Tand T The energy conservation equation, ignoring the vanishing small kinetic

energy giving to the nucleus is simply

hv = 2moc2 + + T
(A-28)
= 1.022MIeVy + T- + T+

The average electron/positron average energy is

-v h-1.022M2eVy
T = (A-29)

















e"



Positron (e+)



Figure A-4. Pair Production in the Coulomb force field of the nucleus



For hv values well above the threshold energy 2moc2, the electron and positron are

strongly forward directed. Their average angle of departure relative to the original photon

direction is roughly:

moc2
B (radians) (A-30)


Pair production in the electron field:

In the kinematics of Pair Production in the electron filed (triple production), the photon

divides its energy between the positron-electron pair produced and the host electron. The energy

conservation equation becomes:

hv = 2moc2 + T T- T,-
(A-31)
= 1.022M~eV + T + T + T

The average kinetic energy of the three particles is

-v h-1.022M2eVy
T = (A-32)









In triple production, the threshold for this process is 4moc2 =2.044 MeV. The higher

threshold is required by conservation of momentum as shown in Table A-1 and also derived by

Perrin (1933). Triple production becomes significant in radiotherapy treatment planning for

energies greater than 14 MeV were it represents ~3% of the total photon interaction.

Straggling and Multiple Scattering

There is typically a distribution of angle and energy for charged particle passing the same

medium. This is a result of multiple scattering and angular straggling which the spread in angular

distribution observed in a population of initial identical charged particle after they have traverse

equal path lengths. It will be somewhat exaggerated if the particle have passed through a layer of

material, since multiple scattering then cause individual differences in scattering angle as well

causing them to propagate in a conical angular distribution in accordance with Moliere's theory.

However, for a discretized solid angle, one can consider an average angle for these

scattered electrons on which approximate transport calculations can be made. The spread of the

solid angle can be giving by Fermi angle for straggling especially when considering restricted

stopping power including soft collisions, that yield small spread in the angle, and hard collisions

resulting in 6 rays with energies less than the cutoff value A .

Restricted stopping power (dT/ pdx~)

Is the fraction of the collisions stopping power that include the soft collisions plus those

hard collisions resulting in 6 rays with energies less than the cutoff value A .


d T =kr I 2 (r + 2)2C
=~~ k In+ rI- (A-33)









Radiative losses:

(dT /pdx~)rad ZT
(A-34)
(dT / odx),,,, 820

Angular straggling (Fermi 1950)

Electrons suffer multiple scattering events preclude straight transport in the material.

Further complication are caused by the statistical fluctuation




















Figure A-5. Pair Production in the Coulomb force field of the nucleus




2= [N] Z,,,r24/ 0 (A-35)



ao = 2 (A-36)
moc










































$Ears
$Esophagus
$External nose
$Eye balls
$Gall Bladder W
$Gall Bladder C
$Heart W
$Heart C
$Kidney-cortex (L)
$Kidney-cortex (R)
$Kidney-medulla (L)
$Kidney-medulla (R)
$Kidney-pelvis (L)
$Kidney-pelvis (R)
$Larynx
$Lens
$Liver
$Lung (L)
$Lung (R)
$Nasal layer (anterior)
$Nasal layer (posterior)
$Oral cavity layer
$Pancreas
$Penis
$Pharynx
$Pituitary Gland
$Prostate
$Rectosigmoid W
$Rectosigmoid C


UF_15yr voxel MCNP model whole body
Matrix size [60,27,167]
Voxel resolution=1.0* 1.0 1.0
2008-7
Ahmad Al-Basheer
The University of Florida


WWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWWW


APPENDIX B
MCNP5 PHNTOM INPUT DECK


cells Card


Body composition and density

1 -0.001205 -999 imp:p,e=1 u=1


vol=78406.867 $ Air


Residual Soft Tissue
Adrenal (L)
Adrenal (R)
Brain
Breast
Bronchi
Right Colon W
Right Colon C


-1.03
-1.03
-1.03
-1.04
-0.94
-1.07
-1.03
-0.83
-1.10
-1.03
-1.05
-1.07
-1.03
-1.03
-1.04
-1.06
-1.05
-1.05
-1.04
-1.04
-1.04
-1.04
-1.07
-1.07
-1.06
-0.24C
-0.24C
-1.02
-1.02
-1.02
-1.02
-1.05
-1.03
-1.03
-1.03
-1.03
-0.49


i
i
i


-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
0 -99!
0 -99!
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999


mp:p,e=1 u=2
mp:p,e=1 u=3
mp:p,e=1 u=4


I

I
I
I


vol=3.94E+04 $
vol=2.00E+00 $
vol=3.00E+00 $
vol=1.39E+03 $
vol=1.70E+01 $
vol=4.00E+00 $
vol=9.70E+01 $
vol=1.45E+02 $
Svol=3.00E+00
vol=2.40E+01
vol=2.00E+00
Svol=1.10E+01
Svol=0.00E+00
Svol=4.50E+01
Svol=1.86E+02
vol=4.23E+02
Svol=8.00E+01
Svol=8.40E+01
Svol=3.10E+01
vol=3.20E+01
vol=5.00E+00
Svol=6.00E+00
Svol=2.20E+01
Svol=0.00E+00
Svol=1.23E+03
7 vol=1.72E+03
8 vol=2.04E+03
Svol=0.00E+00
Svol=0.00E+00
vol=0.00E+00
Svol=1.08E+02
Svol=2.30E+01
Svol=2.00E+00
Svol=0.00E+00
Svol=4.00E+00
Svol=0.00E+00
Svol=1.39E+02


imp:p,e=1 u
imp:p,e=1 u
imp:p,e=1 u
imp:p,e=1 u
imp:p,e=1 u
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
9 imp:p,e=1
9 imp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1
Simp:p,e=1


1=5
1=6
1=7
1=8
1=9
u=10
u=11
u=12
u=13
u=14
u=15
u=16
u=17
u=18
u=19
u=20
u=21
u=22
u=23
u=24
u=25
u=26
u=2
u=2
u=29
u=30
u=31
u=32
u=33
u=34
u=35
u=36
u=37
u=38











-1.03
-1.03
-1.03
-0.41
-1.09
-1.04
-1.06
-1.03
-1.02
-1.04
-1.03
-1.05
-1.05
-1.02
-1.03
-1.04
-1.00
-0.001
-1.03
-0.34
-1.03
-1.03
-1.10
-1.10
-1.10
-1.10
-1.519
-1.519
-1.387
-1.351
-1.219
-1.351
-1.196
-1.196
-1.196
-1.219
-1.351
-1.319
-1.319
-1.387
-1.319
-1.352
-1.352
-1.352
-1.352
-1.273
-1.273
-1.273
-1.273
-1.35
-1.415
-1.319
-1.319
-1.378
-1.378
-1.374


-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999
-999


imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1


u=39
u=40
u=41
u=42
u=43
u=44
u=45
u=46
u=47
u=48
u=49
u=50
u=51
u=52
u=53
u=54
u=55
u=56
u=57
u=58
u=59
u=60
u=61
u=62
u=63
u=64
u=65
u=66
u=67
u=68
u=69
u=70
u=71
u=72
u=73
u=74
u=75
u=76
u=77
u=78
u=79
u=80
u=81
u=82
u=83
u=84
u=85
u=86
u=87
u=88
u=89
u=90
u=91
u=92
u=93
u=94


vol=3.40E+01 $Salivary Glands parotidd)
vol=1.00E+01 $Scrotum
vol=1.40E+02 $SI W
vol=7.70E+02 $SI C
vol=1.72E+02 $Skin
vol=4.10E+01 $Spinal Cord
vol=1.21E+02 $Spleen
vol=3.30E+01 $Stomach W
vol=2.08E+02 $Stomach C
vol=1.80E+01 $Testes
vol=3.20E+01 $Thymus
vol=8.00E+00 $Thyroid
vol=5.50E+01 $Tongue
vol=2.00E+00 $Tonsil
vol=6.00E+00 $Trachea
vol=3.00E+00 $Urinary bladder W
vol=1.61E+02 $Urinary bladder C
vol=4.23E+02 $Air (in body)
vol=7.00E+01 $Left Colon W
vol=1.95E+02 $Left Colon C
vol=2.00E+01 $Salivary Glands submaxillaryy)
vol=7.00E+00 $Salivary Glands (sublingual)


vol=2.20E+01 $Coastal cartilage of ribs
vol=0.00E+00 $Cervical Discs
vol=2.80E+01 $Thoracic Discs
vol=1.20E+01 $Lumbar Discs
vol=7.48E+02 $Cranium
vol=3.30E+01 $Mandible
vol=1.06E+02 $Scapulae
vol=3.00E+01 $Clavicles
vol=4.40E+01 $Sternum
vol=2.90E+01 $Ribs
vol=5.10E+01 $Vertebrae-C
vol=2.28E+02 $Vertebrae-T
vol=1.95E+02 $Vertebrae-L
vol=1.29E+02 $Sacrum
vol=4.67E+02 $Os Coxae
vol=2.20E+02 $Femur-proximal
vol=1.63E+02 $Femur-upper shaft
vol=1.33E+02 $Femur-lower shaft
vol=2.59E+02 $Femur-distal
vol=2.07E+02 $Tibiae-proximal
vol=1.01E+02 $Tibiae-upper shaft
vol=1.01E+02 $Tibiae-lower shaft
vol=7.30E+01 $Tibiae-distal
vol=1.40E+01 $Fibulae-proximal
vol=4.00E+00 $Fibulae-upper shaft
vol=1.40E+01 $Fibulae-lower shaft
vol=1.70E+01 $Fibulae-distal
vol=2.40E+01 $Patellae
vol=4.30E+02 $Ankle+Feet
vol=1.50E+02 $Humerus-proximal
vol=7.40E+01 $Humerus-upper shaft
vol=6.70E+01 $Humerus-lower shaft
vol=9.30E+01 $Humerus-distal
vol=1.60E+01 $Radii-proximal










106 -1.374
106 -1.374
106 -1.374
106 -1.374
106 -1.374
106 -1.374
106 -1.374
105 -1.415
105 -3.00
1 -0.001205
28 -0.2995


-999
-999
-999
-999
-999
-999
-999
-999
-999


imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1
imp:p,e=1


u=95
u=96
u=97
u=98
u=99
u=100
u=101
u=102
u=103


vol=1.60E+01 $Radii-upper shaft
vol=1.70E+01 $Radii-lower shaft
vol=2.70E+01 $Radii-distal
vol=4.90E+01 $Ulnae-proximal
vol=2.70E+01 $Ulnae-upper shaft
vol=1.80E+01 $Ulnae-lower shaft
vol=7.00E+00 $Ulnae-distal
vol=8.60E+01 $Hand
vol=1.30E+01 $Teeth
vol= 78406.867 $ Air
vol= 10210.44 $ Lung


-999 imp:p,e=1 u=301
-999 imp:p,e=1 u=302


303 2 -1.00 -999 imp:p,e=1 u=303 vol= 27482.027 $ Tissue
304 101 -1.32 -999 imp:p,e=1 u=304 vol= 2304.37 $ Bone
C ******************* Lattice definition ************************************
1001 0 -100 fill=10000 imp:p,e=1 $ surrounding box
555 0 -200 lat=1 u=10000 imp:p,e=1 fill = 0:59 0:26 0:166
C *************** Image data start from here *******************************
Phantom Input from Ghoast-3-D
C *************************************************************

c Surface Cards

C *********************************************************
100 rpp 0 60 0 27 0 167 $(-29:30 -13:13 -83:83)
200 rpp 0 1.0 01.0 0 1.0
999 rpp -25 85 -25 52 -25 192

c ***************************************************************
Material Cards
C***************************************************************
mode pe
sdef x=d1 y=d2 z=d3 erg=d4 AXS= 0 1 0 dir= 0.99763 vec= 0 1 0
sil H 30 40
spl D 0 1
si2 H 0.0 1.0
sp2 D 0 1
si3 H 119 136
sp3 D 0 1
si4 H 0.01 8.0
sp4 D 0.0 1.0


c tallies cards


phys:p 4j 1
ctme 500
totnu
print









LIST OF REFERENCES


1R. Mohan, in XIlth International Conference on the Use of Computers in RadiRRR~~~~~RRRRR~~~~atio Therapy,
edited by (Medical Physics Publishing, Madison, WI, 1995, Salt Lack City, 1997), pp. 16-18.35

2T. R. Mackle, P. Reckwerdt, and N. Papanikolaou, "3-D Radiation Treatment Planning and
Conformal Therapy," M~edical Physics Publishing Corporation Madison WI, 1995), pp. 34

3B. A. Fraass, J. Smathers, and J. Deye, Summary and recommendations of a National Cancer
Institute workshop on issues limiting the clinical use of Monte Carlo dose calculation algorithms
for megavoltage external beam radiation therapy," Med Phys 30, 3206-3216 (2003).29

4F. M. Khan, The Physics ofRRRRRRRRRRRRRRRRRRRadito Therapy, 3 (Lippincott Williams & Wilkins, 2003).36

5P. R. Almond, A. E. Wright, and M. L. Boone, "High-energy electron dose perturbations in
regions of tissue heterogeneity. II. Physical models of tissue heterogeneities," Radiology 88,
1146-1153 (1967).37

6A. Ahnesjo, M. Saxner, and A. Trepp, "A pencil beam model for photon dose calculation," Med
Phys 19, 263-273 (1992).73

7A. L. EDWARDS, J. A. RATHKOPF, and R. K. SMIDT, "Extending the Alias Monte Carlo
sampling method to general distributions,," in the International Conference on] Aubeinal'ltliL and
Computations and Reactor Physics, edited by (American Nuclear Society, Inc., La Grange Park,
IL, 1991), pp. 38

sA. Ahnesj o, "Collapsed cone convolution of radiant energy for photon dose calculation in
heterogeneous media," Med Phys 16, 577-592 (1989).52

9T. Knoos, A. Ahnesj o, P. Nilsson, and L. Weber, "Limitations of a pencil beam approach to
photon dose calculations in lung tissue," Phys Med Biol 40, 1411-1420 (1995).50

10B. Dionne, "Automated variance reduction technique for 3-D Monte Carlo coupled electron-
photon-positron simulation using deterministic importance functions," Nuclear Engineering
Sciences (University of Florida, Gainesville, Fl, 2007), pp. 75

11E. E. Lewis, and W. F. Miller, Computational M~ethods of Neutron Transport ( American
Nuclear Society, La Grange Park, Ill., USA, 1993).18

12L. J. Shukovsky, "Dose, time, volume relationships in squamous cell carcinoma of the
supraglottic larynx," Am J Roentgenol Radium Ther Nucl Med 108, 27-29 (1970).31

13J. J. Fischer, and J. E. Moulder, "The steepness of the dose-response curve in radiation therapy.
Theoretical considerations and experimental results," Radiology 117, 179-184 (1975).30









14B. Wang, G. Xu, J. T. Goorley, and A. Bozkurt, "ISSUES RELATED TO THE USE OF
MCNP CODE FOR AN EXTREMELY LARGE VOXEL MODEL VIP-MAN," in The Monte
Calrlo M~ethod: Versatility Unbounded In A Dynamic Computing World, edited by (American
Nuclear Society, Chattanooga, Tennessee, 2005), pp. 32

1G. I. Bell, and S. Glasstone, Nuclear Reactor Theory, (Robert E. Krieger Publishing CO. Inc,
Malabar, FL, USA, 1985).39

16B. G. Carl son, and K. D. Lathrop, Discrete Ordinates Angular Quadrature of the Neutron
Transport Equation, (Los Alamos Scientific Laboratory Report, LA-3186, 1965).40

17D. Shedlock, and A. Haghighat, Neutron Analysis of Spent Fuel Storage Installation Using
Parallel Computing and Advance Discrete Ordinates and Monte Carlo Techniques," Radiat Prot
Dosimetry 116 662-666 (2005).45

IsK. D. Lathrop, "Remedies for Ray Effects," Nuclear Science and Engineering 45, 255-268
(1971).41

19G. Sjoden, and A. Haghighat, "PENTRAN A 3-D Cartesian Parallel SN Code with Angular,
Energy, and Spatial Decomposition," in Int 'l. Conf: on Math. M~eth & Supercomp. for Nuclear
App., edited by Saratoga Springs, New York, 1997), pp. 553.14

20A. Al-Basheer, M. Ghita, G. Sjoden, W. Bolch, and C. Lee, "Whole Body Dosimetry
Simulations using the PENTRAN-MP Sn Code System," in American Nuclear Socity 2007
Annual2~eeting, edited by Bosten, MA, 2007), pp. 66

21G. M. Daskalov, R. S. Baker, R. C. Little, D. W. O. Rogers, and J. F. Williamson, "Two-
Dimensional Discrete Ordinates Photon Transport Calculations for Brachytherapy Dosimetry
Applications," Nucl. Sci. Eng. 134, 121-134 (2000).19

22G. M. Daskalov, R. S. Baker, D. W. Rogers, and J. F. Williamson, "Multigroup discrete
ordinates modeling of 125I 6702 seed dose distributions using a broad energy-group cross
section representation," Med Phys 29, 113-124 (2002).20

23M. L. Williams, D. Ilas, E. Sajo, D. B. Jones, and K. E. Watkins, "Deterministic photon
transport calculations in general geometry for external beam radiation therapy," Med Phys 30,
3183-3195 (2003).5

24A. Haghighat, International Training Course/Workshop on M~ethodologies for Particle
Transport Simulation ofNuclear Systems, 76

25(ICRU) 1980, RadiRRR~~~~~RRRRR~~~~atio quantities and units, (International Commission on Radiation Units
and Measurements, Bethesda, MD, 1980).42

26F. H. Attix, "The partition of kerma to account for bremsstrahlung," Health Phys 36, 347-3 54
(1979).43









27F. H. Attix, "Energy imparted, energy transferred and net energy transferred," Phys Med Biol
28, 1385-1390 (1983).44

28N. I. o. S. a. T. NIST, ESTAR(Stopping Power and Range Tables for Electrons ),
http://physics.nist.gov/PhysRefData/Star/Tx/SARhm.1

29T. R. Mackle, A. F. Bielajew, D. W. Rogers, and J. J. Battista, "Generation of photon energy
deposition kernels using the EGS Monte Carlo code," Phys Med Biol 33, 1-20 (1988).54

30A. Ahnesjo, P. Andreo, and A. Brahme, "Calculation and application of point spread functions
for treatment planning with high energy photon beams," Acta Oncol 26, 49-56 (1987).53

31L. A. N. L. X-5 Monte Carlo Team, M~CNP5- A General Monte Calrlo N-Particle transpotr~r~r~t~rt~tr~rt~
code, Version 5, 10

32J. E. White, J. D. Intgersoll, C. Slater, and R. Roussin, BUGLE-96 : A Revised2\~ulti-group
Cross Section Library for L WR Applications Based on ENDF B- y7, 46

33L. J. Lorence, J. E. Morel, and G. D. Valdez, Physics Guide to CEPXS: A naultigroup coupled
electron-photon cross section generating code, 11

34G. SJODEN, and A. HAGHIGHAT, PENTRAN: Parallel Environment Neutral-Particle
7RA szort Sn in 3-D Cartesian Geonsetry


35C. Yi, and A. Haghighat, PENM~SHExpress, 62

36G. Longoni, "Advanced quadrature sets, acceleration and preconditioning techniques for the
discrete ordinates method in parallel computing environments," (University of Florida,
Gainesville, FL, 2004), pp. 47

37B. Petrovic, and A. Haghighat, "New directional theta-weighted (DTW) differencing scheme
and reduction of estimated pressure vessel fluence uncertainty," in the 9th hIternational
Synaposiunt on Reactor Dosintetry, edited by Prague, Czech Republic, 1996), pp. 746-753.48

38G. Sjoden, and A. Haghighat, "The exponential directional weighted (EDW) differencing
scheme in 3-D cartesian geometry," in The Joint hIternational Conference on Ma thema'lltliL and
Superconsputing for Nuclear Applications, edited by Saratoga Springs, New York, 1997 ), pp.
21267-21276.49

39G. Sj oden, "An Efficient Exponential Directional Iterative Differencing Scheme for Three-
Dimensional Sn Computations in XYZ Geometry," Nuc. Science and Eng 155, 179-189
(2007). 12

40G. Sj oden, and A. Haghighat, "Taylor Proj section Mesh Coupling Between 3-D Discontinuous
Grids for Sn," Transactions of the American Nuclear Society 74, 178-179 (1996).77









41T. Tabata, R. Ito, and S. Okabe, "Generalized Semiempirical Equations For The Extrapolated
Range of Electrons," Nucl. Instrum. Methods;,103: No. 1, 85-91(1972). (1972).65

42A. Al-Basheer, M. Ghita, and G. Sjoden, "Investigation of deterministic simulation parameters
for a 3-D radiotherapy Co-60 device model," in ANS Winter M~eeting and Nuclear Technology
Expo, edited by Albuquerque, NM, 2006), pp. 511-513.13

43C. Lee, C. Lee, J. L. Williams, and W. E. Bolch, "Whole-body voxel phantoms of paediatric
patients--UF Series B," Phys Med Biol 51, 4649-4661 (2006).7

44ICRU 1992 Photon, electron, proton and neutron interaction datafor body tissue,
(International Commission on Radiation Units and Measurements, Bethesda, MD, 1992). 15

45M. GHITA, A. Al-BASHEER, and G. SJODEN, "Validation of DXS: A Diagnostic X-Ray
Source Generator For X-Ray Transport Simulations," in The American Nuclear Society's 14th
Biennial Topical Meeting of the RadiRRR~~~~~RRRRR~~~~atio Protection and .1/ne/Jing~ Division edited by Carlsbad,
New Mexico, 2006), pp. 9

46D. M. Tucker, G. T. Barnes, and D. P. Chakraborty, "Semiempirical model for generating
tungsten target x-ray spectra," Med Phys 18, 211-218 (1991).6

47M. GHITA, DXS: A Diagnostic X-ray Souce Generator, 74

48C. M. Mal, E. Mok, A. Kapur, T. Pawlicki, D. Findley, S. Brain, K. Forster, and A. L. Boyer,
"Clinical implementation of a Monte Carlo treatment planning system," Med Phys 26, 2133-
2143 (1999).3

49W. Gropp, E. Lusk, and A. Skj ellum, Using MPI: Portable Parallel Pi "5l1g,1iannig \I ith the
M\~essage-Pa;ssing Interface, (The MIT Press, Cambridge, Massachussetts, 1994).72

soA. K. Jones, F. D. Pazik, D. E. Hintenlang, and W. E. Bolch, "MOSFET dosimeter depth-dose
measurements in heterogeneous tissue-equivalent phantoms at diagnostic x-ray energies," Med
Phys 32, 3209-3213 (2005).69

5C. Lee, D. Lodwick, D. Hasenauer, J. L. Williams, C. Lee, and W. E. Bolch, "Hybrid
computational phantoms of the male and female newborn patient: N~URBS-based whole-body
models," Phys Med Biol 52, 3309-3333 (2007).70









BIOGRAPHICAL SKETCH

Ahmad Al-Basheer was born in 1978 in Ramtha City, Jordan, in the heart of the Middle

East. Generations of his family have made Ramtha their hometown for centuries. This city,

which borders Syria and Jordan, is the gateway to the desert region of the Middle East.

Ahmad is the oldest of eight children, and his parents, Khaled and Fatima, are both

involved in educational careers; both are school principals. Ahmad was involved not only in

traditional educational venues, but also in chess, soccer, reading and poetry. Ahmad competed in

several reading and poetry contests throughout his schooling; he won the National Award for

high school students for Best Poet.

Ahmad graduated from Jordan University for Science and Technology with a Bachelor of

Science degree in applied physics in 2000. Upon graduation, he spent one year as a physics

teacher at the Yarmouk University Model School. In January, 2002, Ahmad was granted

admission to the graduate program in the Department of Nuclear and Radiological Engineering

at the University of Florida. He performed his research work under the supervision of Dr. Samim

Anghaie, where he investigated a new approach for reducing scattered photons and electron

contamination in Cobalt-60 therapy beam.

In fall 2004 he j oined the University of Florida Transport Theory Group (UFTTG) to

pursue his Ph.D. with Dr. Glenn Sj oden. During his Ph.D., He won several honors, including

best papers presented to number of conferences, a number of invited speeches at national

conferences, and several recognition form number of organization for his achievements. Ahmad

married Ghadeer Al-Basheer, on June 16th 2007 and has one daughter, Jude Al-Basheer who was

born in June 14th 2008.





PAGE 1

1 3D DETERMINISTIC RADIAT ION TRANSPOR T FOR DOSE COMPUTATIONS IN CL INICAL PROCEDURES By AHMAD AL-BASHEER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Ahmad Al-Basheer

PAGE 3

3

PAGE 4

4 ACKNOWLEDGMENTS Any project, no matter how individualized, will almost certainly require input, assistance or encouragement from others; my project is no exception. I extend my sincerest gratitude to Dr. Glenn Sjoden. He has been a constant source of thoughtful guidance in pursuing this project. Because of his input, advice and challenge, I have matured as a researcher and as a student. I would also like to acknowledge the contri bution of my supervisory committee: Dr. Ali Haghighat, Dr. James Dempsey, Dr. Wesley Bolch, Dr. Ayman Hawari a nd Dr. William Ditto. I would also like to take this opportunity to express my sincer e gratitude to my classmates and the faculty members in the Nuclear and Radiological Engineering and Radiation Oncology Departments for their constructiv e suggestions and helpful comment s. During my research, I had the pleasure of working with the University of Florida Transport Theo ry Group (UFTTG). With the help of coworkers and friends, even the tedium became bearable. Special thanks go to my family, and mostly my wife Ghadeer, Mom and Dad; they have definitely been encouraging me for the longest! They will always be dear to my heart. I thank God for all He has done for me.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4LIST OF TABLES................................................................................................................. ..........8LIST OF FIGURES.........................................................................................................................9ABSTRACT...................................................................................................................................14CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW..............................................................161.1Overview.......................................................................................................................161.2Initial Clinical Dose Evaluations.................................................................................. 171.3Pencil Beam Transport.................................................................................................. 191.4Solving the Boltzmann Equation.................................................................................. 211.4.1Monte Carlo Method.........................................................................................231.4.2Deterministic Methods...................................................................................... 251.4.2.1Discrete ordinates method (SN method)..............................................251.4.2.2Monte carlo versus deterministic SN methods.................................... 271.5Dose and Charged Particle Equilibrium........................................................................ 281.5.1Deterministic Dose Calculations for Diagnostic Application........................... 301.5.2High Energy Beam Dose Calculations.............................................................. 311.5.3Limits of Charged Particle Equilibrium............................................................ 311.6Outline of This Research............................................................................................... 352 CRITICAL DISCRETIZATION ISSUES IN 3-D DISCRET E ORDINATES MEDICAL PHYSICS SIMULATION................................................................................... 362.1Code Systems................................................................................................................372.2Simulation Geometry.................................................................................................... 382.2.1PENTRAN Model............................................................................................. 382.2.2Monte Carlo Model...........................................................................................392.3Comparison of Deterministic and Monte Carlo Results............................................... 392.3.1Investigation of Angular Quadrature................................................................392.3.2Investigation of Spatial Discretization (Meshing)............................................ 452.3.3Investigation of Cross-Section Libraries:..........................................................492.4Absorbed Dose Calculations.........................................................................................522.5Parallel Decomposition................................................................................................. 542.6Discussions and Findings.............................................................................................. 55

PAGE 6

6 3 WHOLE BODY DOSIMETRY SIMULATIONS WITH THE UF-B SERIES PHANTOM USING THE PENTRAN-MP CODE SYSTEM FOR LOW ENERGY PHOTONS........................................................................................................................ ......573.1Methodology.................................................................................................................583.2UF Series B Pediatric Phantoms...................................................................................593.3PENTRAN-MP Code System....................................................................................... 613.3.1Pre-processing Codes........................................................................................613.3.1.1The collapsing code (GHOST-3-D)....................................................613.3.1.2The mesh generator code (PENMSH-XP)..........................................613.3.1.3Source spectra and cross-sections generation..................................... 623.3.2Transport Calculation........................................................................................ 623.3.3Post-Processing Codes......................................................................................633.3.3.1Data management utility (PENDATA).............................................. 633.3.3.2Dose calculation code (3-D-DOSE)...................................................633.4Computational Example................................................................................................ 643.5Computational Results.................................................................................................. 673.5.1Flux Comparison of Deterministic and Monte Carlo Results........................... 673.5.2Dose Calculations and Dose Volume Histograms (DVH)................................ 703.5.3Parallel Computation.........................................................................................753.6Discussion and Findings...............................................................................................764 ELECTRON ABSORBED DOSE KERNELS TO ACCOUNT FOR SECONDARY PARTICLE TRANSPORT IN DE TERMINISTIC SIMULATIONS .................................... 794.1Introduction to The EDK-SN Method........................................................................... 794.2Monte Carlo Based Dose Kernels.................................................................................804.3EDK Determinations Based on The MCNP5 Monte Carlo Code................................. 824.3.1Computational Model........................................................................................ 834.3.2MCNP5 Tallies Comparison............................................................................. 834.4EDK -SN Code System Methodology........................................................................... 864.4.1Net Current........................................................................................................ 864.4.2MCNP Based Electron Kernels.........................................................................874.4.3Euler Angles......................................................................................................894.4.4Ordinate Projection........................................................................................... 924.4.5Final Dose Calculations.................................................................................... 934.5External Source Benchmark..........................................................................................954.6Comparison of Deterministic and Monte Carlo Results............................................... 964.7Parallelization of EDK-SN Code...................................................................................994.8Discussion and Findings.............................................................................................1015 APPLICATION OF EDK-SN TO MODELS WITH HETEROGENITIES......................... 1035.1Heterogeneous Tissue-Equivalent Phantoms.............................................................. 1035.1.1Monte Carlo Simulations................................................................................1035.1.2PENTRAN Model........................................................................................... 1075.1.3EDK-SN Absorbed dose Comparison.............................................................. 108

PAGE 7

7 5.1.4Correction Factor Approach............................................................................ 1095.2Computational Phantom of 15-Year Old Male........................................................... 1115.2.1EDK-SN Absorbed Dose Calculations............................................................ 1145.2.2Ray-Effects......................................................................................................1165.2.3MCNP5 Energy Deposited Tally (F6) and Pulse Height Tally (*F8) Results............................................................................................................. 1175.3Discussion and Findings.............................................................................................1186 SUMMARY, CONCLUSIONS AND FUTURE W ORK.................................................... 1216.1Conclusions and Findings...........................................................................................1216.2Future Work................................................................................................................123APPENDIX A HIGH ENERGY PHOTON TREATMEN T (IN CONDENSED ELECTRON TRANSPORT) ......................................................................................................................124B MCNP5 PHNTOM INPUT DECK......................................................................................133LIST OF REFERENCES.............................................................................................................136BIOGRAPHICAL SKETCH.......................................................................................................140

PAGE 8

8 LIST OF TABLES Table page 1-1 Comparison between Monte Carlo me thods and determ inistic techniques....................... 28 1-2 Approximate attenuation of photons with a layer of water equal to the maxim um range of secondary charged particles28..............................................................................31 1-3 Approximate thickness of water need ed to achieve various attenuation amounts com pared to maximum energy electron ranges generated by the same beam................... 32 2-1 Total number of directions, cumulative problem time** required for a single energy group, P3 scattering, 126,000 fine m esh cel ls using 12 processors on a parallel cluster.................................................................................................................................54 3-1 Computed tomography image sources fo r the developm ent of the UF pediatric phantom series................................................................................................................. ..60 3-2 Total number of directions, Cumulative Problem Time required for different energy groups, P3 scattering, and 189k fine mesh cells................................................................. 76 5-1 Comparison of selected organ absorbed dose rate (MeV/g. s) calculated using MCNP5 pulse height tally with (photon, electron mode) and EDK-SN for the UF hybrid 15-year-old male phantoms for a ch est 8 MV X-ray flat weighted source.......... 115 5-2 Prostate absorbed dose rate (MeV/g.s) calculated for the test problem using EDK-SN and MCNP5 using pulse height tally (*F8) with (photon, electron mode) at different MCNP5 Monte Carlo running times................................................................................ 116 5-2 Absorbed dose rate (MeV/g. s) calcu lated for the test problem using EDK-SN and MCNP5 using pulse height tallies (* F8) with (photon, electron mode).......................... 117 5-3 Absorbed dose rate (MeV/g. s) calculate d for the test problem using MCNP5 using pulse height tally *F8 wi th (photon, electron mode) and MCNP5 energy deposited tallies (F6)................................................................................................................... .....118 A-1 Ratios of photon interactions in water for range of energies........................................... 124

PAGE 9

9 LIST OF FIGURES Figure page 1-1 Comparison between thicknesses required to attenuate different percentages of prim ary photons and CSDAR range for maximum ener gy secondary electrons produced by the same beam............................................................................................... 331-2 Comparison between thicknesses required to attenuate different percentage of primary photons and average range for maximum energy secondary electrons produced by the same beam............................................................................................... 332-1 60Co unit (left: Mesh distribution genera ted by PENTRAN package; Right: MCNP5 Cell/Surface schematic input)............................................................................................ 382-2 60Co -rays 3-D Groups 1,2 3 and 4 scalar flux computed by PENTRAN with the CEPXS gamma cross section library................................................................................. 402-3 S12 (Level Symmetric) and S42 (Legendre-Chebychev) quadr ature first octants (with 21 and 231 directions/ oc tant respectively)....................................................................... 412-4 60Co -ray 3-D Group 1 scalar flux distribution for x-level depths at 0 cm, 5 cm, 10 cm and 15 cm, computed by PENTRAN with the CEPXS gamma cross section library........................................................................................................................ .........422-5 60Co -ray 3-D Group 1 scalar flux distribution for x-level depths at 0 cm, 5 cm, 10 cm and 15 cm, computed by PENTRAN with the CEPXS gamma cross section library; an S42 angular quadrature (1 848 directions) with P3 scattering anisotropy.......... 432-6 Scalar flux dist ributions using S12, S22, S32, and S42 with P3 scattering anisotropy in PENTRAN using the BUGLE-96 gamma library............................................................. 442-7 60Co model spatial meshing, axial level............................................................................. 472-8 Scalar flux comparison of variation in the interval of 3-D mesh cells containing air; fine meshes of 1 cm size (along each axis) in air proved to be best in providing high detail flux distribution....................................................................................................... .492-9 Scalar flux distributions generated by PENTRAN for a single energy group of each library using the MCNP5 solution as the reference case................................................... 512-10 Relative deviation of energy group 1 disc rete ordinates results with Monte Carlo results.................................................................................................................................522-11 Absorbed dose rate comparison between SN method using kerma approximation and MC techniques using F6 tally, MC statis tical uncertainty were less than 3% on average...............................................................................................................................53

PAGE 10

10 3-1 PENTRAN-MP Code System Structure Flowchart...........................................................583-2 The UF Series B pediatric phantoms series....................................................................... 603-3 Values of the mass energy-absorption coefficient, and the fitting function as a function of photon energy, for Lung (ICRU-44) soft tissue (ICRU-44) and cortical bone....................................................................................................................................643-4 UF-Series B phantom (11 years old male) with (398 242 252 voxels); the corresponding PENTRAN back-thinned mo dels with (132 80 84 voxels) and 72 materials (upper right), and another with (79 48 50 voxels) and 66 materials (lower right).......................................................................................................................663-5 Transversal view of the 11-year-old male Series B original phantomcollapsed into 189k voxels and 4 materials, collapsed into 900k voxels and 4 materials (lower left) and collapsed into 900k voxels and 72 materials............................................................... 673-6 x-ray 3-D Group 1, 3, 5 and 7 scalar flux computed by PENTRAN with the CEPXS cross section library, S32 angular quadrature (1 088 directions) with P3 scattering anisotropy...........................................................................................................................683-7 Flux distributions using S32 with P3 scattering anisotropy in PENTRAN using the CEPXS library vs. MCNP5 Monte Carlo results............................................................... 683-8 Angular flux distributions using S32 with P3 scattering anisotropy in PENTRAN using the CEPXS library vs............................................................................................... 693-9 Angular flux distributions using S32 with P3 scattering anisotropy in PENTRAN using the CEPXS library vs. MCNP5 Monte Carlo results for energy groups 1, 3, 5 and 7 along the z axis at differe nt depths in the phantom..................................................693-10 Simulation methodology for 3-D absorbed dose computations using PENTRAN-MP code system.................................................................................................................... ....713-11 CT view of the 11-year-old male Seri es B original phantom collapsed into 189 k voxels and the correspondent absorbed dose distributions 900 k voxels and the correspondent absorbed dose distributions........................................................................ 723-12 CT view of the 11-year-old male Seri es B original phantom collapsed into 189 k voxels and the correspondent absorbed dose distributions 900 k voxels and the correspondent absorbed dose distributions........................................................................ 733-13 CT view of the 11-year-old male Seri es B original phantom collapsed into 189 k voxels and the correspondent absorbed dose distributions 900 k voxels and the correspondent absorbed dose distributions........................................................................ 743-14 Simulation methodology for 3-D organ absorbed dose computations using PENTRAN-MP code system.............................................................................................74

PAGE 11

11 3-15 Dose Volume Histograms (DVH) comput ed based on PENT RAN (S32,P3) transport results using the CEPXS library vs. D VHs computed using MCNP5 Monte Carlo results.................................................................................................................................754-1 Procedure to produce EDK base d on Monte Carlo Calculations....................................... 824-2 Monte Carlo model used for comparisons between MCNP 5 pulse height tally, energy deposited tally and cell flux tally empl oying dose conversion factors (DE, DF).............. 844-3 Comparison between F6 (p, e mode), F6 (p mode), and F4 ( DE, DF) tallies in MCNP5, statistical uncertainty was less than 1%..............................................................844-4 Comparison between F6 (p, e mode) and *F 8 (p mode) tallies in MCNP5, statistical uncertainty was less than 1%............................................................................................. 854-5 Comparison between *F8 (p, e mode) and *F 8 (p mode) tallies in MCNP5, statistical uncertainty was less than 1%............................................................................................. 854-6 Deterministic calculation will provid e the angular information explicitly........................ 874-7 For each energy group, using Monte Carlo ra diation transport, a pencil beam of mono-energetic photons interacts at the center of a water phantom.................................. 884-8 Euler angles, zyz convention. Similar conventions can be obtained by following of the following order in rotation ( zyz, xyx, xzx, yzy, yxy )......................................................894-9 EDK-matrix after rotation to direct ion cosines A: (-0.5, 0.35, 0.8) and B: (0.25,0.40,-0.88) relative to the referenc e coordinate system (0, 0, 1)....................................... 914-10 First octant of S42 equal weight angles used for projecting the EDK-SN........................934-11 A flow-chart for the EDK-SN Code System...................................................................... 944-12 External Source impinging on block of water................................................................... 954-13 External Source 3-D absorbed dose di stribution for 2 Energy groups computed by EDK-SN based on PENTRAN-MP A. EG(7.5-8.0Mev) and B. EG(0.5-1.0Mev)............ 974-14 Energy group absorbed dose rate distri butions (in Mev/ g.sec) along the X-axis using the EDK-SN method based on PENTRAN-MP....................................................... 974-15 Energy group absorbed dose rate distri butions (Mev/ g.sec) along the y-axis using the EDK-SN method based on PENTRAN-MP................................................................ 984-16 EDK-SN Total absorbed dose rate distri butions along the X-axis compared to MCNP *F8 tally derived absorbed doses.......................................................................................984-17 Total absorbed dose rate distributions of EDK-SN along the Y-axis compared to MCNP *F8 tally................................................................................................................. 99

PAGE 12

12 5-1 Three Monte Carlo phantoms used in the absorbed dose study....................................... 1045-2 Monte Carlo derived total absorbed do se rate distributions along the X-axis computed using *F8 MCNP5 tally for three different phantoms.................................... 1055-3 Monte Carlo derived total absorbed do se rate distributions along the X-axis computed using *F8 MCNP5 tally for three different phantoms.................................... 1065-4 Monte Carlo derived total absorbed do se rate distributions along the X-axis computed using *F8 MCNP5 tally for the _SBLLBSS_ phantom.................................. 1065-5 Three SN phantoms equivalent to the one us ed in the Monte Carlo simulation..............1075-6 EDK-SN heterogeneous SBLLBSS phantom to tal absorbed dose rate distributions along the X-axis compared to MCNP *F8 tally............................................................... 1085-7 EDK-SN Homogenous phantom total abso rbed dose rate dist ributions along the Xaxis compared to MCNP *F8 tally................................................................................... 1095-8 Continuous slowing down approximation ra nge (Rcsda) for electrons in Cortical Bone, Soft Tissue and Lung as a function of energy....................................................... 1105-9 Continuous slowing down approximation range (RCSDA) ratio comparison for electrons in Cortical B one, Soft Tissue and Lung as a function of energy...................... 1115-10 Frontal views of 3-D rendering of th e 50th percentile UFH-NURBS 15-year male and female........................................................................................................................1125-11 Simulation methodology for EDK-SN computations using PENTRAN-MP code system..............................................................................................................................1135-12 Absorbed dose contribution of each en ergy group is calculated via the EDK-SN procedure, then folded to calculate the to tal absorbed dose deliv ered to the 15 year old male phantom from 8 Me V flat weighted source......................................................115A-1 Total mass attenuation coefficient for photons in water, indicating the contributions associated with phoelectric absorption, Compton scattering and electron-positron pair production.................................................................................................................124A-2 Sketch of Compton scattering, photon of energy h collide with an electron. The photons incident forward momentum is hc where c is the speed of light in vacuum.............................................................................................................................125A-3 Compton effects differential cross section per unite angle for the number of electrons scattered in the direction e..............................................................................................128A-4 Pair Production in the Coulom b force field of the nucleus.............................................. 130

PAGE 13

13 A-5 Pair Production in the Coulom b force field of the nucleus .............................................. 132

PAGE 14

14 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy 3D DETERMINISTIC RADIATION TRANSPORT FOR DOSE COMPUTATIONS IN CLINICAL PROCEDURES By Ahmad Al-Basheer August 2008 Chair: Glenn Sjoden Major: Nuclear Engineering Sciences The main goal of this dissertation was to es tablish the feasibility of basing megavoltage external photon beam abso rbed dose calculations in voxelized phantoms on SN deterministic calculations and pre-calculated electron absorb ed dose kernels derived from full-physics Monte Carlo. The SN derived electron absorbed dose kernel method EDK-SN, developed as part of this research, achieves total execution times that ar e on the order of several times to orders of magnitude faster than conventional full-phys ics Monte Carlo electron transport methods considering equivalently detailed models and data fidelity. With the rapid movement toward intensity mo dulated radiation therapy (IMRT), radiation beam intensities have increased dramatically over the past decade, thus heightening the need for further characterization of out-o f-field organ absorbed doses, along with their associated biological risks. Assessment of these tissue absorbed doses is complicated by two fundamental limitations. First, anatomic information on the patie nt is generally restrict ed to a partial body CT image acquired for treatment planning; conseque ntly, whole-body computational phantoms must be employed to provide the out-of-field anatomy model structure for absorbed dose evaluation. Second, existing methods based on Monte Carlo radiation transport, even with the application significant variance reduction, are quite computati onally inefficient at large distances from the

PAGE 15

15 primary beam, and point-kernel methods do not properly handle tissue inhomogeneities. Moreover, since absorbed dose ar e generally tracked in all majo r organs in th e body, variance reduction schemes for Monte Carlo are not all effective in this regard. The outcome of this dissertation is to dem onstrate that absorbed dose from high-energy external beams radiation can be accurately computed for whole body and organ-specific absorbed doses. The EDKSN method implements voxelized phantoms with discrete ordinates (SN) transport computations coupled with directional influences and (pre-computed) full-physics Monte Carlo based electron absorbed dose kernel s to yield total body abso rbed dose information. This research shows that the deterministic t echniques coupled with Monte Carlo based electron absorbed dose-kernels has significant potential fo r organ absorbed dose eval uation in the clinical management of radiation therapy patients.

PAGE 16

16 CHAPTER 1 INTRODUCTION AND LITERATURE REVIEW 1.1 Overview W ith the growing implementation of radiat ion applied in differe nt therapeutic and diagnostic medical procedures, comprehensive and accurate ab sorbed dose computations are fundamental to precisely understand the impact and outcome of such procedures. In practice, the clinical outcome in radiothera py can be significantly improved w ith enhancements provided by a treatment planning systems (TPS)1. Current TPS have been based on augmenting clinically measured absorbed dose distributions. Such me thods yield only approximate solutions, and are limited in rendering a precise solution. Detaile d solutions are achiev able by solving the Boltzmann transport equation, which describe the beha vior of neutral partic le radiation transport throughout the phase space of a problem. Often absorbed dose calculation systems employ relatively simplistic yet fast algorithms to pred ict the patient absorbed dose distribution from first principles, using simp lified radiation transport models2. Overall, these techniques are not capable of accurately describing high fidelity whole body detailed radiation transport through the human body, particularly as new procedur es in medical physics are developed. In recent years, Monte Carlo techniques have proven to be ca pable of representing very complex geometries in a rigor ous manner. This gave the Monte Carlo technique a great advantage over previous methods in rendering accurate estimates of tissue absorbed doses, especially when simulating human anatomy and yielded accurate prediction of absorbed dose delivered in various medical physics procedures. However, this accuracy is achieved at a high cost: extremely long running times are typically necessary to achieve a statistically precise solution. This limits the practical use of Monte Carlo simulations for clinical procedures even with High Performance/Parallel Computing (HPC) 3.

PAGE 17

17 Solving the Boltzmann equation directly using deterministic techniques can be significantly more efficient than using the Monte Carlo methods, a nd therefore it can be used to improve the quality of radiotherapy planning. Re lative to the amount of information generated, and with the use of HPC resources, such tech niques usually require significantly lower running times (relatively to Monte Carlo) to provide deta iled and precise solutions. Moreover, the rapid proliferation of Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) enable human anatomy to be represented using regular images consisting of rectangularly (hexahedral) shaped voxels in a regular matrix. These new developments make Cartesian geometry deterministic techniques ideal for medical physics applications, especial ly given the demand for high resolution absorbed dose distributions. This research is focused on the development of EDK-SN, a first-of-a-kind capability to assess organ absorbed doses incurred in radiati on therapy (RT) applica tions using anatomical patient models coupled with 3-D deterministic discrete ordinates (SN) radiation transport methods. Absorbed doses delivered as a result of secondary charged particles will be linked using electron kernels to account for secondary electron effects. This methodology is proposed for regions in the phase space where Charged Particle Equilibrium (CPE) is not satisfied, and subsequent absorbed dose estimation us ing a collisional kerma is not valid. 1.2 Initial Clinical Dose Evaluations In the field of radiation therapy, absorbed dose distributions were historically first approxim ated without doing any explicit radiation tr ansport at all, but rather by extrapolating from measured results taken for a particular en ergy, field size, and sour ce-to-surface distance (SSD) on a given machine4. An extrapolation was performed by modifying the absorbed dose that was experimentally determined under the re ference parameters by ratios representing the change between specific paramete rs and the parameters of interest. For example, if absorbed

PAGE 18

18 dose values were measured in water, a table of the percentage of the maximum dose as a function of depth (on-axis) in the water could be deduced. A similar table could be generated for off-axis ratios (OARS) by measuring latera l profiles at a variet y of depths. The absorbed dose at a given point P in the patient could then be estimated by knowledge of the maximum dose for the treatment, and applying a percent depth dose (PDD) correction for the depth of P in the patient, and an off-axis correction for the distance to P from the axis at that depth4. Tables of PDDs were made available for vari ous sources (incident en ergies), field sizes, depths in the phantom, and SSDs. To modify th e dose for an SSD that was different from the reference SSD, the ratio of inverse distances s quared would be applied. For a change in field size, tables of field-size-factor s would be subsequently applied. To compensate for the curved surfaces of a patient, as opposed to the flat surf ace of the phantom, isodose lines could be shifted along rays projected from the source4. It is important to point out that these techniques are often quite successful for predicting absorbed dose in a homogeneous volume. In f act, in some cases they may be superior to alternative simplified methods (for homogene ous volumes) because they are based on measurements from a particular machine; thus, th ey circumvent the proble m of how to model the photon distributions emerging from the source and in cident on the patient/phantom, which can be a significant obstacle in accomplishing comput ational radiation transport based models. The real challenge for these empirical-bas ed methods, then, lies in correcting the homogeneous water absorbed dos e distribution to account for th e many inhomogeneities present in the human body, including air cavities, bone, de nse muscle mass, and sparse lung tissue. One way of addressing this problem is to a pply equivalent-thickness corrections to the water measurements. These corrections were ba sed on compressing or stre tching isodose lines by

PAGE 19

19 factors that were derived from the relative phot on/electron density in the path through tissue as compared to that in water4,5. These methods, at their best, were useful for obtaining depth dose information in the center of a broad beam, for a medium consisting of homogeneous layers. Even then, these approaches do not a ccount for the fact that photons in teract not only wi th electrons, but also with the nucleus. For example, the elasti c scattering cross section depends on the atomic number of the material. The clinical implication of this is that not all materials behave like scaled-density water; real bone, for example, s catters photons to greater extent than water of equivalent bone density due to the presence of Ca and other high Z elements not found in liquid water. The effect of this was bette r addressed in the Pencil Beam method6. 1.3 Pencil Beam Transport. The Pencil Beam method sums the dose distri bution from individual small diameter rays called pencil beams. A pencil beam is made up of particles which pass through a differential cross sectional area, x y The off-axis dependence of the dose distribution for each pencil beam is described by thick-target multiple scattering7. The on-axis dose information is obtained from measured depth-dose data. The calculation of absorbed dose assumes that the patient or phantom can be represented by a stack of slabs of different material types. E ach individual slab is homogeneous and infinite in lateral extent. Assuming a single pencil be am has normal incidence on such geometry, the absorbed dose at a position (x, y, z) can be expressed as (,,)(,,)() Dxyzfxyzgz (1-1) where the first factor is the off-axis term, and th e second factor is the ce ntral-axis term. The offaxis term may be represented by a Gaussian of the form

PAGE 20

20 22 221() (,,)exp()zz x y fxyz (1-2) where z is the angular spread about the axis incl uding the penumbra. The on-axis term may be expressed in terms of its relations hip to the depth dose in matter, 0()effgz 2 0()()eff effSSDz gzgz SSDz (1-3) where SSD is the source-to-surface distance, and effz is the effective depth in water, assuming the linear stopping power of the material at depth z is relatively independe nt of electron energy for normal body tissue. The quantity 0()effgz can be related directly to a depth dose curve measured in a water phantom, assuming a uniform incident beam, by the Eq. 1-4 '2'2 '' 0 0 22() ()() (,,)expzz xygz xxyy Dxyz dxdy (1-4) where the limits of the integral depend on the dispersion of the field at depth z. An example of pencil method used for clinical photon radiotherapy is an application of the Ahnesjo pencil beam model6. In this method, combination of measured data and Monte Carlo calculated convolution kernels are used to char acterize clinical photon beams and eventually calculate the absorbed dose. The strength of the Ahnesjo pencil be am method is that it is fast enough for clinical use, on the order of minutes. It also uses measured data for input that may be specific to the treatment unit. However, because this method applies simple corrections using pencil beams in homogenous matter, it is not accurate as a comple te coupled charged-particlephoton Monte Carlo simulation in he terogeneous media. Moreover, it does not fully represent the directional component of the scattered radiation.

PAGE 21

21 Aside from the geometric assumptions of in finite homogeneous slabs, the pencil beam model does not handle changes in scatter from late ral heterogeneities; devi ation of up to 5% can be expected for low energy photon beam s, and 14% for high energy photon beams8,9. 1.4 Solving the Boltzmann Equation At the m ost fundamental level, radiation transport simulation involves determining the distribution of the particles (in physical space, direction, time, and energy) after they have traversed some distance into a medium. Formal ly, a way to begin this discussion is by writing the general equation for particle transport from kinetic theory of gases. This theory does not explicitly treat individu al particles, but rather describes th e aggregate behavior of a continuous particle distribution in a statistical manner. Let n be the expected number of particles within a differential volume element 3drat position r with a velocity (which is directly related to its kinetic energy) that is within a differential velocity3dv, at time t. Then, if a general solution is found for the particle density function n(r, v, t) in phase space, then the transport problem is solved. When transporting the radiati on through an arbitrary volume V, bounded with a surface area S, the rate of change of number of particle s is due to flow out through the surface of V, collision events within V, or sources (scattering and independe nt sources) of particles located inside of V. (all the terms of rate of change; commonly the first two terms are losses and the last term is production) R ateofChangedueChangedueChangedue changeintoflowtocollisionstosources nattimetthroughSinVinV

PAGE 22

22 This balance of neutral particles is cast as the linear Boltzmann Transport Equation (BTE). The BTE is an integro-differential equation which describes the behavior of neutral particles in terms of spatial, angular, and energy domains as they interact in a system; the steady-state transport equation is given in Equation (1-5): (,,)(,)(,,) rErErE 04 (,',') (,',')(,,)sdEdrEErEqrE (1-5) where (,,) rEdrdEd is the angular flux with energy between E and E+dE at position r within the volume element dr, and in direction within the solid angle d. The scattering term, defined with the double integr al term, represents the sum of the partic les scattered into dEd from all the '' dEd after scattering collisions repr esented by the double differential cross-section (,',')srEE (,,) qrE represent the external source. The left side of Equation (1-5) represents streaming and collision terms (loss), and the right side represents scattering and independent sources (gain). Since it describes the flow of radiation in a 3-D geometry with angular and energy depe ndence, this is one of the most challenging equations to solve in terms of complexity and model size. For rende ring a deterministic computational solution for a large problem in a reasonable time, it is necessary to utilize a robust parallel transport algorithm and a high perf ormance computing system. All the methods described below are different approaches to solvin g this equation. For cases with charged particle transport, the Boltzmann and Fokker-P lank formalisms (Boltzmann-FP) 10are generally used. A simplified form of this equation is usually used, which is:

PAGE 23

23 (,,)(,)(,,) rErErE 04 (,',') (,',') [(,',)(,',')] (,,)sdEdrEErE SrErE qrE E (1-6) However, the full Boltzmann-FP equation is extremely challenging to solve deterministically. 1.4.1 Monte Carlo Method Monte Carlo techniques are a widely used class of computational algorithms for the purpose of statistically modeling various physical and mathem atical phenomena. In general, individual probabilistic events are simulated sequentially, where probability distributions governing particle interactions ar e statistically sampled to desc ribe the total phenomenon. In principle, for each case sampled, there are four required inputs; namely, a random number generator, source characteristics, a simulatio n phase space, and probability distributions governing the possible outcomes. Results of indi vidual particle historie s are generated using pseudo-random numbers to sample the probability distributions. This process is repeated many times, until the statistical precision required for the solution is achieved, based on the law of Large Numbers and the Central Limit Theorem11. Therefore, the Monte Carlo technique is an effective method for indirectly solving the Boltzm ann radiation transport equation via statistical methods, and can readily treat th e full physics of coupled photon-el ectron transport. Simulations for particle histories are repeat ed many times for large number of histories, where the computer code tracks the desired physical qu antities, means, and uncertainti es (variances) for all of the simulated histories.

PAGE 24

24 One of the most important issues limiting th e use of Monte Carlo techniques for clinical absorbed dose calculations is the long execution time required to reduce th e statistical variance associated with the calculated absorbed dose at each differential volume. At the same time, the outcome of any radiotherapy procedure depends solely on the accuracy of the radiotherapy treatment planning system, so that a small cha nge in the absorbed dose delivered ( 5%) can result in a dramatic change in local response of the tissue ( 20%)12,13. In order to reduce the statistical variance 1 N associated with Monte Carlo, large numbers of histories N are required, and consequently this l eads to higher running time to yield acceptable results. Improved variance reduction techniques may reduce the running time. However, such techniques are to be used with caution, as improper application of statistical variance reduction, method applied un-conserva tively may cause wrong answers. Also, variance reduction is often tied to one objective; this can be problematic when faced with determining a dose at all locations in the phase space. The increasing use of voxelized human body ge ometries has created challenges in many aspects of applying Monte Carlo simulations14. With increasing use of modern technologies such as CT and MRI in vivo imaging, anatomical de finition has shifted from using simplified surface equations to segmented tomographic images in order to truly reflect the anatomy of a specific human body. The segmented images are represented in the form of a large uniform 3-D matrix of hexahedral voxels, where each vox el represents a constant mate rial with a tag number (for example, CT number). One digital image ma y contain billions of voxels, making it very challenging be fully represented explicitly using Monte Carlo, particularly considering large models size and number of tally sites, addi ng significant overhead to the computation. Some special data management treatments are ne eded for Monte Carlo simulations to make

PAGE 25

25 computation times more tractable, includi ng voxel down-sampling and parallel computing resources. 1.4.2 Deterministic Methods Several different methods have been deve loped for numerically solving the Boltzmann equation11,15. In general these methods depend on disc retizing the energy, angular, and spatial domains of the Boltzmann transport equation. 1.4.2.1 Discrete ordinates method (SN method) The most common deterministic approach cu rrently used for photon transport is the discrete ordinates method (SN method) developed in detail by Carlson over 50 years ago16. The discrete ordinates equations are derived by discretizing the BTE over phase space elements (i.e., spatial-energy-direction variables) to construct discrete-variable cell par ticle balances. Equation (1-7) is the standard Legendre expanded Cartes ian multigroup integro-differential formulation of the BTE, and z y x ,, are the spatial variables along each ax is in the Cartesia n coordinates. G g L l l k k l lgl lggs g g gP kl kl zyxPzyx l zyxzyx zyx zyx1 '0 1 ,' ,',! 2,, ,,12 ,,,,,,,,,, ', ',,,cos ,,sin ,,,,kk CglSglgxyzkxyzkqxyz (1-7) Other variables in Equation (1-7) are: correspond to the direction cosines along each axis for each ordinate g is the energy group angular flux g is the total group macroscopic cross section lggs ,', is the l th moment macroscopic scattering cross section for group g to g

PAGE 26

26 lg ,' is the l th flux moment for group g k lCg,' and k lSg,' are l th, kth associated cosine, sine flux moments for group g ) ( lP and )(k lP are the Legendre and Associ ated Legendre polynomials. ,,,,gqxyz is independent source for group g The SN methodology is a widely accepted determin istic scheme in yielding numerical solution to neutral particle radiation transport problems, and can be applied to medical physics problems, provided that discretizat ion and truncation errors can be reduced to an acceptable level over the problem space. Depending on the amount of data needed per floating point operation, the SN method is more efficient than the Mont e Carlo method for many problems, and with proper discretization produ ces an accurate solution17. The SN method is widely used in nuclear engineering to obtain a numerical solution of the Boltzmann transport equation. The method solves the BTE for a discrete number of ordinates (directi ons) with associated weights16. The combination of discrete directions and weights is referred to as quadrature set11. A drawback of the SN method is that with the limited number of directions involved, in certain situations, this may lead to so called ray-effects, which appear as unphysical flux oscillations. In general, this behavior occurs for problems with highly angular dependent/localized sources and/or fluxes, or when the source is localized in a small region of space, in low density or highly absorbent media. This can make medical physics applications, whic h often require small mesh sizes, relatively large models, and low density materials, suffer severe ray-effects. The ray-effects can be alleviated naturally by the presence of a distri buted isotropic source, which tends to flatten the flux in the angular domain. However, in medical phys ics, the photon transport problems may exhibit significant ray e ffects. This effect becomes worsen in a low

PAGE 27

27 scattering medium such as air, or in a strongl y absorbing medium such lead, where the photon flux usually is typically strongly angular dependent. In such situations, either a higher order quadrature set (many directions) is required or some ray-effects remedies must be applied11,18 (such as Taylor Projection techniques and adaptive differencing schemes19). More discussion on discretization and ra y-effects are presented in Chapter 2. 1.4.2.2 Monte carlo versus deterministic SN methods The most fundamental difference between the Monte Carlo methods and the deterministic SN transport method is that the SN method solves the transport equa tion for the average particle behavior throughout the problem phase space, while the Monte Carlo method, the terms of the transport equation are solved statistically by si mulating the individual pa rticles and recording some aspects (tallies) of their average behavior but only at specific tall y locations requested by the user. In the discrete ordinates method, the phase space is divided into many small meshes, in which the cross section is assumed to be consta nt. In the medical physics applications, the model geometry, in this case, the pati ent is represented using regular voxel-based grid generated from a segmented CT scan data or other digital imaging technique. The types of grids can then be easily represented in an input deck for the SN calculations, making both Monte Carlo methods and deterministic techniques capable of representing the model of study accurately. Another advantage of the MC methods is that they can ea sily utilize continuous energy interaction data, whereas deterministic calcul ations use multigroup or energy group averaged data. However, photon attenuation coefficients in the energy range of interest for radiotherapy (RT) are quite smooth with little structure (in contrast to neutron cross sections); and are accurately represented by multigroup da ta. Recent research we conducted20 has shown that multi-group deterministic methods produce highly accurate absorbed dose distributions in photon

PAGE 28

28 transport calculations. Similar conclusions were drawn by Daskalov et. al and Williams et. al2123. Table (1-1) shows a comparison between Monte Carlo and Deterministic techniques24. Table 1-1. Comparison between Monte Ca rlo methods and deterministic techniques 1.5 Dose and Charged Particle Equilibrium Absorbed dose, D, can be defined as a non-st atistical quantity in te rms of a statistical quantity, energy imparted The absorbed dose to matter at a point P is given by / Dddm (1-8) where is the expectation value of the energy imparted by ioni zing radiation to matter of mass m in a finite volume V and inout R RQ (1-9) in which in R = the radiant energy incident on the volume V out R = the radiant energy emerging from V and Q= the net energy derived from rest mass within V (negative for the reverse process). where the radiant energy R can be defined as the energy of particles (excluding rest energy) emitted, transferred, or received25.

PAGE 29

29 Collisional kerma,cK can be defined in terms of th e related net energy transferred, n tr 26,27 and R25 as. /n trddm (1-10) The energy-transferred in a volume V is ()()n nonrr trinuoutuu R RRQ = r tru R (1-11) in which ()inuR = the radiant energy of unchar ged particles entering the volume V ()nonr outuR= the radiant energy of uncharged particles leaving V except that which originated from radiative losses of kinetic energy by charged particles while in V r u R = radiant energy emitted, where u indicates uncharged particles. Charge particle equilibrium (CPE) can be es tablished in a volume when the energy carried into the volume V by charged particles is balanced by the energy carried out of the volume V by the charged particles. Note that, CPE is not established for thin materials irradiated by photons. The range of secondary electrons produced is larger than the material dimensions, so that much of the electron energy is lost simply by electrons leaking out of the material. Under the conditions of CPE, Attix demonstrat ed that when the energy imparted equals energy transferred, tr collisional kerma cK is equal to the total dose D26,27 Eq(11). ,.CPE en c EZDK (1-12)

PAGE 30

30 where is the energy fluence, and ,en EZ is the mass energy absorption coefficient for ray energy E in medium Z. This result is very im portant, since it allows us to calculate the measurable dose D based on the calc ulable quantity cK 1.5.1 Deterministic Dose Calculatio ns for Diagnostic Application For low energy beams, such as the one used in diagnostic applications, the concept of charged particle equilibrium is of a great im portance as a mean of equating absorbed dose D to collisional kerma cK For a volume V, containing a smaller volume v, irradiated with an indirectly ionizing radiation field (i.e., Photons and Neutron fields), there are four basic causes for CPE failure in volume v: In-homogeneity of the atomic composition within volume V In-homogeneity of density in V Non-uniformity of the field on indi rectly ionizing radiation in V Presence of a nonhomogeneous el ectric or magnetic field in V These four situations will disrupt CPE by prev enting charged particles of a given type and energy leaving the volume v to be replaced by an identical particle of the same energy entering27. In practice, CPE conditions are met if the attenu ation of the primary beam is negligible in the maximum range of the secondary electrons ge nerated by the beam. As indicated by Table (12) for a layer of water of 5 cm thickness, this scenario is plausible in the limit of diagnostic medical physics applications, where the energies fall within 10 to 200 keV, and CPE conditions are met for water/tissue volumes less than 55 cm3, and consequently D is equal to cK under such conditions.

PAGE 31

31 Table 1-2. Approximate attenua tion of photons with a layer of water equal to the maximum range of secondary charged particles28 1.5.2 High Energy Beam Dose Calculations Based on the previous discussion, as the ener gy of the ionizing radiation increases, the penetration power of the secondary charged particles increases more rapidly than the penetration power of the primary radiation in a given volume. This leads to CPE failure, caused by the small number of charged particles produced at deeper depths in volume V relative to the number of charged particle produced at initial points along the primary radiation field direction27. This is indicative of significant atte nuation of the primary indi rectly ionizing radiation. As a result of this phenomenon, higher numbers of charged particles will be generated close to the entrance compared to the rest of the volume, leadi ng to CPE failure considering dose in that volume. The degree of this failure becomes progressively large for hi gher energies, so that D no longer equals tocK A more detailed discussion on this topic is presented in chapter 4. 1.5.3 Limits of Charged Particle Equilibrium Assuming that a maximum energy is transferred from a photon to an electron, we can use this assumption to investigate a CPE requirement for a water phantom region, as defined earlier. Considering the continuous slowing dow n approximation range of the electron (CSDAR), and average range of the maximum energy electrons as function of energy. 01 0 T CSDAdT R dT dx (1-13)

PAGE 32

32 0 0 0 01 .() ()ftdt ttttdt N dNt dt dt (1-14) where dT dx is the electron stopping power (MeV cm2/g) T0 is the starting energy of the electron (MeV) () ()fdNt tt dt is the differential distribution of farthest depth of penetration ()Ntis the number of particles penetrating a slab of thickness t Through inspection of these relationships de scribed in Figures 1-1 and 1-2, one can decide what kind of tolerance is acceptable in applying a collisional kerma approximation for a dose calculation based on a particular voxel size i.e. to determine CPE approximation limits. This also has been demonstrated in Table 13; for example at 2 MeV the CPE computed dose will be accurate to approximately 5% using a 1 cm voxel, assuming the CSDA range applies. Table 1-3. Approximate thickness of water needed to achieve various attenuation amounts compared to maximum energy electron ranges generated by the same beam

PAGE 33

33 Figure 1-1. Comparison between thicknesses required to attenua te different percentages of primary photons and CSDAR range for maximum ener gy secondary electrons produced by the same beam Figure 1-2. Comparison between thicknesses required to attenua te different percentage of primary photons and average range for maximum energy secondary electrons produced by the same beam With proper model discretization of the angleenergy-spatial transpor t phase space, and at low photon energies yielding primary and secondary electron interactions with small electron

PAGE 34

34 transport paths relative to a spatially discretized mesh grid interval, 3-D SN methods can be implemented to directly yield ve ry accurate dose distributions. Th erefore, the deterministically derived dose using a kerma approximation will be accurate up to the range of electron path lengths comparable to ranges smaller than the photon spatial SN mesh grid interval in anatomical models (typically representative of the anatomical data voxel size). As mentioned earlier, at higher photon ener gies, where electron ranges exceed the SN mesh grid interval, this is not the case, and interacti ons yielding energetic el ectrons must account for the mechanism of electron tr ansport through the problem phase space to deliver an absorbed dose after representative electron transport dist ances. This work will set the foundation for state of the art 3-D absorbed dose computations for high energy external beams to account for interactions yielding ener getic electrons. The EDK-SN methodology will account for electron transport in the phantom by applying a pre-comput ed electron absorbed dose kernel calculated to a very low variance It is worth comparing the EDK-SN methodology, proposed here with the traditional convolutionsuperposition techni que, which applies a database of energy deposition kernels, calculated using Monte Carlo techni ques to determine a absorbed dose29,30. We note that these kernels can be used to describe the spread of energy about the primary photons point of interaction in the phantom, however, these have limited use since the co nvolutionsuperposition technique is performed without regard to the incident photon angular component. While, in contrast, the EDK-SN procedure is based on a detailed, full physics Boltzmann transport solution with direction and flux information directly available for coupling with the EDKs along a propagated photon dire ction of travel.

PAGE 35

35 1.6 Outline of This Research The method developed in this dissertation, EDK-SN, includes implementing voxelized phantoms with discrete ordinates (SN) transport computations coupled with multi-directional and (pre-computed) electron absorbed dose kernels (EDK). Particular care was taken when generating the EDKs using full-physics Monte Carlo calculations (with the MCNP5 code) and tallying the results using pulse he ight tally (*F8) tallies in MCNP5 to extremely low statistical uncertainties. The application of the pre-com puted EDKs was based on the photon flux and net current vector for the dose driving voxel (DDV) calculated via the disc rete ordinates parallel code PENTRAN with a voxelized phantom. Chapter 2 introduces code systems, and presents is a study of critical discretization issues in 3-D discrete ordina tes medical physics simulations benchmarked with the Monte Carlo. The discre tization of the phase space variables in the transport equation will be disc ussed, along with the absorbed dose calculation comparisons. Chapter 3 discusses whole body dosimetry simulations with the UF-B series phantoms. It also presents the creation of the PENTRAN-MP code system. Chapter 4 elaborates further on CPE, and compares absorbed dose calculations using different absorbed dose tallies in the MCNP5 code. It also introduces the EDK-SN methodology and the parallel st rategies. In Chapter 5, we apply the new EDK-SN methodology to heterogeneous phantom s, including the whole body 15 year old phantom and compare the results to the full physics (Monte Carlo MCNP5 *F8) tally. Chapter 6 summarizes the research and draws conclusions on the objectives accomplished; aspects for future development are also highlighted.

PAGE 36

36 CHAPTER 2 CRITICAL DISCRETIZATION ISSUES IN 3-D DISCRETE ORDINATES MEDICAL PHYSICS SIMULATION Discretization of the BTE m ust be well unders tood in solving this equation for medical physics problems. As discussed, the fundamental difference between the SN and Monte Carlo methods is that the SN method directly solves the BTE an d requires phase space discretization, while the Monte Carlo method applies a statistical approach to represent the radiation transport. It is important to explore discreti zation issues associated with the SN method. To account for discretization effects, 3-D flux distributions were computed in a water phantom exposed to 60Co -rays using the PENTRAN parallel SN code and compared to analogous models implemented using the continuous energy MCNP531 Monte Carlo results with E NDF/B-VI data libraries. A major issue affecting the accuracy of the SN method is selecting an ap propriate discrete set of directions, or the angular quadrature set for which the transport equation is to be solved. Choosing a quadrature set that does not include enough discrete directions to represent the physics can lead to ray-effects. The ray-effects are numerical artif acts sometimes exhibit in the SN computations when there is high radiation streaming and little scattering. With a finite number of directions considered in each fine mesh cell for the radiation emanating at points distal from the source; it often re sults in the spines at points distal from the source. In this work, a number of studies have been conduc ted through comparison with the Monte Carlo simulations to determine suitable quadrature sets required to reasonably li mit the ray-effects in a proposed radiotherapy model that includes radiation transport in ai r to still yield high accuracy. Another set of simulations were accomplished to address anothe r difficulty in performing the SN simulations with regard to spatial discretization. Selection of an adequate 3-D spatial mesh size to represent the different geometric features of the model is also criti cal; at the same time, selection of a differencing scheme capable of representing the physics w ithin tractable memory

PAGE 37

37 storage and time limitations is also a must. This i ssue has been investigated particularly in the low density portions of the model (mostly in air). Finally, we compare the performance achieved with three cross section libraries availabl e from the nuclear engineering and physics communities. These multi-group gamma-ray libraries were derived from the BUGLE9632library, the CEPXS package33, and the SCALE5 package. 2.1 Code Systems PENTRAN (Parallel Environment Neutral particle Transport) is a 3-D discrete ordinates code that solves the Boltzmann transport equation on distributed parallel co mputers. It is capable of complete, hybrid phase space decomposition with up to a 97% parallel fraction19,34. In executing a parallel transport calculation on each distributed node, PENTRAN performs discrete ordinates sweeps only on the local portion of th e phase space assigned. As problem sizes increase, more machines may be needed to s cale up to accommodate the total parallel memory demand. The inherent parallel scalability of PENTRAN with advanced numerical treatments (including block adaptive mesh structures and ad aptive differencing schemes) makes it ideal to apply to challenging problems in medical physics. The PENTRAN c ode is part of a code system that includes preand post-proce ssors, and other utilities to manage large parallel datasets. For the purpose of comparison to the M onte Carlo methods, the well known MCNP5 Monte Carlo code developed by Los Alamos National Laboratory is used to simulate a 60Co beam for an equivalent unit. MC NP5 is used to calcula te the scalar flux along the central axis of the model. For these studies, cell flux mesh-ta lly (F4) and energy deposit ion volume tallies (F6) are defined in the Monte Carlo model in a way that is analogous to the Cartesian fine mesh cells used in the PENTRAN.

PAGE 38

38 2.2 Simulation Geometry Because the purpose of our investigations was to identify critical discretization issues, the models we considered were simplified to enable us to focus on proper numerical discretization. 2.2.1 PENTRAN Model We considered a box of lead containing a 60Co source and a water tank for use in multiple benchmarking applications (Figure 2-1). As we mentioned earlier, this model consisted of four key components; a 60Co source (2 cm3), a lead source housing 4 cm thick, an air layer, and a homogeneous water phantom (30 cm3) at a source-to-surface distance (SSD) of 34 cm. A 3-D spatial distribution was generated by the PENMSHXP35 mesh generating code in the PENTRAN system. The model was divided into 5 z-levels at 0 cm, 30 cm, 60 cm, 64 cm, and 66 cm; each z-level was divided equally into 9 coa rse meshes containing various numbers of fine spatial mesh cells; a total of 126,000 fine mesh cel ls were used in the SN problem. Figure 2-1. 60Co unit (left: Mesh distribution genera ted by PENTRAN package; Right: MCNP5 Cell/Surface schematic input)

PAGE 39

39 2.2.2 Monte Carlo Model The MCNP5 Monte Carlo model geometry was define d to be equivalent to that used in the PENTRAN model. The 60Co source was distributed isotropically over multiple equi-probable sources cells in one energy group. Source spectra were defined in a consistent manner with the various multi-group libraries cons idered. Volumetric cell flux F4 mesh-tallies and energy deposited F6 tallies were used in the Monte Carl o geometry description that was equivalent to the discretized SN volumes. The MCNP5 F6 tallies were used along the central axis of the model across the lead box, air region, and water phantom. All simulations were performed using a PC cluster composed of 16 Dual Intel Xeon processors at 2.4 GHz with 2 Gb/processor, for a total memory of 32 Gb. MCNP5 tallies along the centra l axis were obtained using variance reduction techniques. The cell fluxes c onverged, on average, to within 3 % for the reference case. 2.3 Comparison of Deterministic and Monte Carlo Results Calculations were performed to compare the SN and Monte Carlo results for the scalar fluxes along the model central axis. Figure 2-2 de picts a 3-D scalar flux distribution computed by PENTRAN for the reference case for different energy groups. The highest energy group (group 1) is forward peaked with some ray-effects al ong the central axis. However, as the radiation down scattered to lower energy groups, it demonstrates more isotropic behavior evident in energy groups 2 and 3. This simulation was based on the use of an S42 angular LegendreChebychev quadrature (1848 directions/mesh) with P3 scattering anisotropy using the CEPXS library. This highlights the fidelity of the information available in a converged deterministic SN computation. 2.3.1 Investigation of Angular Quadrature Medical physics applica tions represented in SN models often require small mesh sizes, relatively large models, and low density materials. These problems can lead to severe ray-effects,

PAGE 40

40 which appear as unphysical flux oscillations in the scalar flux distribution unless treated properly. In general, this be havior occurs for problems with highly angular dependent sources and/or fluxes, or when the source/sink is localiz ed in a small region of space, in low density or highly absorbent media. Group 1 Group 2 Group 3 Group 4 Figure 2-2. 60Co -rays 3-D Groups 1,2 3 and 4 scalar flux computed by PENTRAN with the CEPXS gamma cross section library, (S42 angular quadrature (1848 directions) with P3 scattering anisotropy where used) To mitigate this problem, high angular quadrature orders are often needed. Figure 2-3 depicts a schematic of the ordinate de nsity distribution associated with S12 level-symmetric (rotationally level-symmetric) and S42 Legendre-Chebychev quadrature sets on 1/8th of a unit sphere (other octants are symmetri c). Note that ordinates (vectors ) in Figure 2-3 would begin at the origin and project to e nd at the square point plot ted in the Figure. Beyond S20, LegendreChebychev quadrature sets must be used due to negative weights that occur in standard levelsymmetric sets36. Particular attention must be made in deriving quadrature sets to enable preservation of Spherical harmonics (Legendre) scattering moments a nd numerical quadrature orthogonality relationships.

PAGE 41

41 S12 S42 Figure 2-3. S12 (Level Symmetric) and S42 (Legendre-Chebychev) quadr ature first octants (with 21 and 231 directions/ octant respectively) Using a uniform 1 cm mesh (for reasons di scussed later) in the air region, PENTRAN was used to calculate the gamma flux di stributions in the system for several different quadrature sets. PENTRAN provides the option of r eadily using any quadrature that could be prescribed by the user. The flux distributions produced by an initia l computation, using S12 level-symmetric quadrature with P3 scattering anisotropy and 126,000 fine mesh cells, is shown in Figure 2-4. The ray-effects in the S12 computation are evident when compar ed to scalar fluxes in Figure 2-5 using S42 quadrature, with a similar mesh structure and P3 scattering (the reference case). While there is evidence of small ray e ffects in the reference case far from the source, the differences between Figures 2-4 and 2-5 is dramatic. We al so found that there was a small difference (on the order of several percent, depending upon location) in flux solutions computed with S42 between P3 and P1 computations, but little difference noted in the problems tested between P3 and P5 anisotropy, and so P3 was implemented for all other computat ions. To further investigate the effects of raising the quadrat ure order, unbiased quadrature sets were tested using S22, S32, etc, up to S42 with P3 anisotropy. A comparison of the PENT RAN computed fluxes in Group 1 using selected quadratures and e quivalent MCNP5 calculations are given in Figure 2-6.

PAGE 42

42 0 cm 5cm 10 cm 15cm Figure 2-4. 60Co -ray 3-D Group 1 scalar flux distribution for x-level de pths at 0 cm, 5 cm, 10 cm and 15 cm, computed by PENTRAN with the CEPXS gamma cross section library; an S12 angular quadrature (1 68 directions) with P3 scattering anisotropy was used as a reference case

PAGE 43

43 0 cm 5 cm 10 cm 15 cm Figure 2-5. 60Co -ray 3-D Group 1 scalar flux distribution for x-level de pths at 0 cm, 5 cm, 10 cm and 15 cm, computed by PENTRAN with the CEPXS gamma cross section library; an S42 angular quadrature (1 848 directions) with P3 scattering anisotropy

PAGE 44

44 1.00E+00 1.00E+02 1.00E+04 1.00E+06 1.00E+08 1.00E+10 0.00E+003.00E+016.00E+019.00E+01 Distance (cm)Flux Sn12 Sn22 Sn32 Sn42 MCNP5 As shown in the Figure 2-6, the PENTRAN results with S42 Legendre-Chebychev quadratures are in good agreemen t with the MCNP5 resu lts, and differences between the two are within the statistical uncertainty of the Monte Carlo in air and the water phantom distal from the source. Overall, this figure shows that incr easing the angular quadrature order results in a progressive improvement in solution accuracy, and that a substantial number of directions are required for proper representation in air region between the source and the water phantom. We further note that while the S12 level-symmetric quadrature is quite adequate to represent neutron scattering in nuclear engineering neutronics a nd reactor physics applications, it does not produce accurate results here. For a general problem w ith a high degree of angular dependence, the rayeffects must be largely mitigated by employing a large number of directions, i.e., a high-order angular quadrature, From our analysis, a minimum of S32 Legendre-Chebychev quadratures should be considered. Investigat ion of alternative, less computationally in tensive methods for mitigating the ray-effects in problems of this t ype has shown that ordinate biasing methodologies can also be of benefit19. Figure 2-6. Scalar fl ux distributions using S12, S22, S32, and S42 with P3 scattering anisotropy in PENTRAN using the BUGLE-96 gamma library

PAGE 45

45 2.3.2 Investigation of Spatial Discretization (Meshing) In addition to proper resoluti on of the angular variable, the spatial variable must be appropriately discretized to represent the materi als and geometry of the system to yield the solution to the BTE. In PENTRAN, a block ad aptive Cartesian three-dimensional hexahedral mesh arrangement is used to define the spatial-ma terial distribution of the system, i.e. the system is partitioned into coarse mesh blocks containing fine mesh cells having a uniform cross-section in each fine mesh cell. The solutions to the SN equations are obtaine d by sweeping along each cell and discrete direction on the unit sphere; this process is usua lly referred to as a transport sweep 11. In discretizing a spatial mesh, it is of ten necessary to examine, using successive computations, the effect of the coupling between angle, mesh size, cross section, and the spatial differencing scheme used to solve SN equations. Here, we will briefly highlight how spatial differencing schemes are applied in the SN method. Equation (1-6), is integrated over the volume ,,ijkijkVxyz to yield volume and surface averaged flux terms. Rearranging, we can obtain the zeroth moment balance equation (omitting group subscripts for brevity): AA z in z out m y in y out m x in x out mq z y x (2-1) Equation (2-1) contains seven unknowns. Three of which, entrant fluxes (in) are assumed to be known. Also, the collectiv e cell averaged angular volumetric source A q (which includes scatter and fixed source s) is assumed to be known fr om previous source iteration. Therefore, the average angular flux A and the exiting ( out) surface angular fluxes are unknowns, where these values are derived using a set of auxiliary equations which enable

PAGE 46

46 coupling of the average and boundary angular fluxes. Linear weighted sc hemes use the auxiliary formulations in Equation (2-2): 1 1 a ax inA x out (2-2a) 1 1 b by inA y out (2-2b) 1 1 c cz inA z out (2-2c) The standard Diamond Differenci ng (DD) scheme results when a=, b=, and c= in equations (2-2a, 2-2b, and 2-2c ); the DD scheme is second or der accurate, but may lead to negative solutions11. Petrovic and Haghighat developed the Directional Theta-Weighted scheme DTW37 where a can be expressed as in in in in () 1 22 mmm A xmyz mm xq xyz a yz (2-3) Similar weights can be derived along the y and z -axes for b and c respectively. DTW can remedy the unphysical oscillations inherent in solutions rendered by the DD scheme, and it is always positive. In cases of high angular stre aming, exponential aux iliary equations may be needed. These are implemented into PENTRAN to represent streaming behavior of the angular flux across the spatial cell. The expo nential Directional Weighted (EDW)38,39 can be employed in this situation with the following auxiliary formulation: (2-4) 111(,,)exp(()/||)exp(()/||)exp(()/||)moimjmkmxyzaPx Py Pz

PAGE 47

47 Multiple differencing schemes in PENTRAN have an adaptive structure between successive schemes of increasing accuracy within a coarse mesh containing many fine meshes called adaptive differencing strategy (ADS): if one begins with the Diamond scheme (Diamond Zero (DZ)), the detection of a negative flux-fix-up causes select ion of the Directional Theta-Weighted (DTW) scheme, and if weights (as in Equation 2-3) are beyond certain values, selection of the Exponential Dire ctional Weighted (EDW) scheme is made. In recent PENTRAN versions selection of the Exponential Directi onal Iterative (EDI) method is made as an improvement over EDW. This ADS enables the code to automatically select the most appropriate differencing scheme within a particular coarse mesh We examined the effect of different sizes of mesh cells with various differencing schemes in our model. Figure 2-7 depi cts the fourth z-level mesh structure (plotted at hexahedral cell cente rs) relative to the full-size model; note that sparser meshing was used in the surrounding air region. Figure 2-7. 60Co model spatial meshing, axial level 4 The size of the fine mesh cells in the ai r region of the problem was of particular importance due to low density; smaller meshes yield high accuracy, but can lead to more pronounced ray effects; larger meshes mitigate th e ray-effects, but may be inaccurate. Figure 2-8

PAGE 48

48 shows a comparison between fine meshes side size of 1 cm, 2 cm and 5 cm in the air region with the DTW scheme fixed in air. These results demons trate that a 1-cm mesh in air is necessary to maintain accuracy. Based on these observations, we designed our PENTRAN model as follows; the model is 30x30x70 cm3, and consists of 45 total coarse mesh blocks arranged among 5 axial levels, containing a total of 126, 000 fine meshes. In recent PENTR AN versions, selection of the Exponential Directional Iterat ive (EDI) method is made as an improvement over EDW39. The geometry of the model has been represented using the block adap tive meshing capability of the PENTRAN code that allows the user to use a hi gher mesh density in the regions of interest. Axial level thicknesses were determined based on major components of our model, and the X-Y spatial dimensions of coarse meshes (contai ning various densities of fine meshes) were uniformly 10 cm on each side. The DTW differencing scheme was locked for the regions with low density material (air regi ons), and the adaptive differencing scheme (with adaptive upgrade potentially up to the EDW differencing scheme) was prescribed for higher material density regions. A mesh coupling scheme, Tayl or Projection Mesh Coupling (TPMC)40, was used to link discontinuous meshes using first order projectio ns as radiation sweeps through the problem. The Cartesian mesh sizes for the full size model were selected to preserve component volumes, also taking into account the optical thickness of each material. The fine-mesh intervals in different problem regions are: Source0.5 cm; lead0.5 cm; water1.0 cm; Air1.0 cm. Smaller mesh sizes better represen t curved boundaries and irre gular structure in the model, but require larger memory and may lead to longer running time (depending upon the spectral radius of the equations). Smaller mesh sizes were selected for the source region of the problem, where the particle flux has a strong gradient. Mesh sizes in the air can be made larger near the edges of the problem where a more detailed solution is not required.

PAGE 49

49 Figure 2-8. Scalar flux comparison of variation in the interval of 3-D mesh cells containing air; fine meshes of 1 cm size (along each axis) in air proved to be best in providing high detail flux distribution Using the ADS described earlier with different mesh sizes, we were able to achieve a converged solution. Figure 2-8 depicts the solution behavior versus mesh size in the air region (even though there is very little optical th ickness or attenuation by the ai r); this can be explained by the fact that larger mesh sizes in the air do not adequately propagate the radiation from the source, and therefore result in so lutions with significant truncation error, particularly near the source region, while using smaller mesh sizes in air can provide good solutions with fine detail over the problem domain. 2.3.3 Investigation of Cross-Section Libraries: After satisfying the requiremen ts for properly representing a ngular quadratures and spatial discretization, the use of an a ppropriate cross section library is also important in solving problems in medical physics. Th e scalar fluxes resulting from the deterministic calculations using different cross-section libra ries have been examined (Figur e 2-9) and were compared with

PAGE 50

50 similar values generated using MCNP5. To best present comparisons with the Monte Carlo results (since Monte Carlo statistical errors were lowest (<3% relative) along the central axis), we consider a slice through the central axis of the model. It is im portant to note that even with the use of volumetric mesh tallies that yield global fluxes across a grid spanning the Monte Carlo geometry, the statistical uncer tainty associated with each mesh tally site, particularly those distal or off-axis from the source region, can routinely exceed 40% or more, limiting their utility in practical applications. Figure 2-10 shows that for this particular prob lem, the relative differences in scalar fluxes between all deterministic solutions and the M onte Carlo reference increased adjacent to the gamma source. Also, for this problem, deterministic results using BUGLE-96 and CEPXS demonstrated good results in this water phantom and results were w ithin the statistical uncertainty of the Monte Carlo method. Subsequent investigation of these solutions could not attribute this behavior to grid density, quadrature, or anisot ropic scattering order. We did determine that the relative fluxes of the Monte Carlo results (converged to <3% relative error), particularly near the source, were strongly dependent upon the precise method by which the geometric volume tallies were defined, and this could easily account for the differences computed near the source. On the other hand, cross section libraries such as CEPXS provide flexibility in defining energy group width not available in BUGLE-96. In general, differences can be attributed to the energy group structure of the libraries. Result s here demonstrate the importance of library benchmarking exercises for specific problem applications. This may also suggest that in some cases the use of an equivalent surface source fluence prescribed on the entering phantom surface may be more practical to implement for dosimetry in the phantom, discussed in the next section.

PAGE 51

51 A B C Figures 2-9. Scalar flux dist ributions generated by PENTRAN for a single energy group of each library using the MCNP5 solution as the refe rence case. These fluxes span the central axis in the three layers of the model. In ge neral, the deterministic results yield a very good match to the Mon te Carlo results

PAGE 52

52 Figure 2-10. Relative deviation of energy group 1 discrete ordi nates results with Monte Carlo results. Results using PENTRAN with BUGLE-96 were overall best in the water phantom, and differences were within the Monte Carlo statistical uncertainty 2.4 Absorbed Dose Calculations For the energy beam used in this study, the maximum electron range produced by the primary photons is approximately 0.6 cm (as com puted by the Tabata-ItoOkabe (TIO) formula, estimate for electron range, 41), which is less than the size of th e smallest fine mesh used in the water phantom. This allows electrons produced to deposit their energy within the fine mesh, achieving CPE within the phantom volume. By applying this collisional kerma approximation for absorbed dose computation based on the fluxes calculated using the SN method with PENTRAN, we compared the SN based absorbed dose estimates with absorbed doses calculated based on MCNP5 track length (F6) tallies, which estimate the energy deposited averaged over a given cell volume and are equivalent to the co llisional kerma, this is shown in Figure 2-11.

PAGE 53

53 Dose Rate Along The X-axis (5 cm Depth)1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 0102030X-axis(cm)Dose rate Sn MC Dose Rate Along the X-axis (15cm Depth)1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 0102030X-axis(cm)Dose rate Sn MC Water Phantom Dose Rate Along The Z axis1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 051015202530Depth(cm)Dose rate MC Sn A B C Figure 2-11. Absorbed dose rate comparison between SN method using kerma approximation and MC techniques using F6 tally, MC statistical uncerta inty were less than 3% on average For the energy range used in this problem, we demonstrate that the absorbed dose rate calculated using the kerma approximation for the SN method using the reference case (P3 anisotropy, CEPXS gamma cross section library, and S42 angular quadrature) is comparable to that calculated using F6 tall y in the MCNP5 code. However, for high energy photons, typically used in radiotherapy, the secondary electrons ge nerated can have large ranges exceeding the fine mesh interval, consequently vi olating the requirements for CPE, and subsequently invalidating

PAGE 54

54 the collisional kerma approximation as a basis fo r absorbed dose computation. In order to use deterministic techniques in high energy photon dosim etry applications, some special treatment for secondary electron transport is therefore re quired; this will be solved using the EDK-SN method. 2.5 Parallel Decomposition For solving the 60Co model problem in a parallel environment, different domain decomposition strategies were cons idered. In Table 2-1, we pr esent the CPU times and memory requirements for the PENTRAN calculations us ing different quadrat ure sets. Parallel decomposition results for solution of the 60Co problem model are also presented in Table 2-1 based on a constant number of meshes of 126,000 with P3 anisotropy for a single energy group and different angular quadrature orders. From the table data, it can be observed that increasing the number of directions scales up running times to the processor workload. More work on each machine lowers the relative impact of message passing synchronization and latency, relative to additional computational work, as evidenced wh ere eleven times added workload only required 9.3 times as much computati onal effort (referring to the S42 calculation). This is a useful illustration that emphasizes that w ith parallel computation, added model fidelity may not actually cost additional wall clock time to yield a more robust solution. Table 2-1. Total number of directions, cumula tive problem time** required for a single energy group, P3 scattering, 126,000 fine mesh cells using 12 processors on a parallel cluster Directions and time ratio are referred to the S12 quadrature set. ** Non dedicated running time

PAGE 55

55 2.6 Discussions and Findings In this chapter, we highlight ed recent studies examining th e discretization requirements of the deterministic SN method compared with the Monte Ca rlo method with application to dosimetry problems. In doing so, we chose to simulate an idealized 3-D 60Co model using the PENTRAN code. Since radiation transporting in such applications encounters little interaction in low density materials, such as air, attention was given to the quadratur e level and mesh size distribution needed for solution accuracy through the air and in a water phantom. As expected, radiation transport in this gamma ray application is highly directional, and that a traditional SN quadrature, such as an S12 level-symmetric set (168 total directions), considered robust for many neutron reactor physics applications, is in sufficient for modeling gamma rays in the 60Co model. This limitation was overcome by the use of the higher levels of quadrature, up to S42 (1848 total directions per cell) with Legendre-Chebychev quadrature sets and P3 anisotropy. We also noted that at least an S32 angular quadrature leads to the elimin ation of most ray effects and scalar fluxes within the statistical uncertainty of Mon te Carlo. Moreover, model mesh sizes were selected to preserve system volumes while taking into account the optical thickness of each material. Smaller Cartesian meshes do better re present the curvature of the model, but they require higher quadrature orders, and therefore larger memory and longer running times. Even in optically thin air regions, a mesh cell interval of on the order of 1 cm wa s necessary to yield an accurate solution, and more cells were necessary where the particle flux encountered strong gradients; cell intervals may be relaxed near boundaries where deta il is not required. Also of importance was the flexibility of the ADS in PENTRAN, permitting more appropriate spatial differencing schemes. Scalar fl ux distributions obtained from PENTRAN agreed with the Monte Carlo prediction when simulating equi valent geometry models of the 60Co model with three different multigroup cross-section libraries. Still, realistic representations of the energy spectra

PAGE 56

56 for dosimetry applications may require fine ener gy group structures; it is clear the applicability of the cross section library shoul d be investigated for each problem considered. Development of a standardized multi-group cross-section library fo r these kinds of applications would be of benefit to minimize potential inconsistencies in th e application of curren tly available libraries. The results of parallel deterministic PENTRAN c ode were obtained within comparable running times to parallel MCNP5 Monte Carlo calculations for tally sites adjacent to th e source. Overall, we demonstrated that with an adequate quadrat ure set, cell mesh interval, and cross-section library, it is possible to produce results with th e PENTRAN discrete ordinates code that agree with the Monte Carlo prediction. A major advantage of the SN method is that it provides a detailed, accurate flux distribu tion at thousands of cells throughout the system, while the Monte Carlo method is best for providing accurate values at a few selected points near the source. Both methods can be advantageous in these types of applications, prov ided the level of SN discretization is adequate. This chapter laid out the founda tions necessary for accurate SN discretizations in general, although only a homogenous phantom was consider ed. The need to implement heterogeneous media in the SN simulation is essential for any medical phys ics applications. In the next chapter, we set the foundation for the new PENTRAN-MP c ode system. In this system, we implement a state of the art series B phantom into our methodology by do wn-sampling these phantoms into accurate phantoms to generate an equivalent PENTRAN input. Moreover, we calculate the absorbed dose by applying a kerma approximation for diagnostic procedures. Then, the EDK-SN method is developed, presented, and applie d in subsequent chapters.

PAGE 57

57 CHAPTER 3 WHOLE BODY DOSIMETRY SIMULATIONS WITH THE UF-B SERIES PHANTOM USI NG THE PENTRAN-MP CODE SYSTEM FOR LOW ENERGY PHOTONS This work describes our 3-D SN Medical Physics code syst em PENTRAN-MP in support of ongoing development research in larg e scale 3-D absorbed dose computations42. In the PENTRAN-MP code system, Figure 3-1, an initi al high-resolution data phantom is obtained from state of the art 3-D whole-body voxeli zed human phantoms based on Computed Tomography (CT) images43. A pediatric patient (an 11-year-old male) 3-D CT image is then down sampled (collapsed) into a coarser one. Th e collapsed model was cast into a 3-D spatial distribution and sub-divided into coarse meshes c ontaining fine meshes with constant densities. SN calculations were performed using PENTRAN a nd Monte Carlo results were generated with MCNP5. To accurately compare our deterministic results with the Monte Carlo, we also generated an equivalent MCNP5 31 model implementing conti nuous energy ENDF/B-VI crosssection data libraries. Post pr ocessing in the PENTRAN-MP code system includes seamless parallel data extraction using the PENDATA code, followed by th e application of the 3-D-DOSE code to determine absorbed dose in each voxel of the 3-D phantom based on fitted mass-energy absorption coefficients 44, hence yielding a whole-body absorbed dose distribution. This chapter is presented as follows: Sect ions 3.1, 3.2 and 3.3 highlight the methodology for accomplishing our deterministic calculations. Section 3.4 pres ents a computational human phantom example implementing the methodology of PENTRAN-MP a nd an equivalent MCNP 5 model. Section 3.5 compares the Monte Carlo and deterministic soluti ons and analyzes the parallel performance of the PENTRAN. Finally, Section 3.6 discusses the findings of this Chapter.

PAGE 58

58 Figure 3-1. PENTRAN-MP Code System Structure Flowchart 3.1 Methodology In this work, we must determine the radi ation transport flux and absorbed dose in a computational tomographic (or voxelized) phantom using deterministic techniques. Central to this is the application of deta iled anatomical models in whic h we can render accurate absorbed doses. Therefore, coupling of these anatomical models with robust 3-D deterministic transport simulations can yield critical or gan and whole body absorbed dose pr ofiles from patient-specific radiotherapy treatments, includi ng absorbed dose delivered to colla teral tissue located away from the exposure site. To properly attribute absorb ed doses in our models, we must set up a discretized phase-space based on a detailed phantom for the material spatial distribution. Due to

PAGE 59

59 interest in fully characteriz ing diagnostic absorbed doses to younger patients, for this presentation, we chose a pediatri c phantom. Also, to ultimately render absorbed doses in the phantom wherever we desire, there are a number of codes that constitute the PENTRAN-MP code system that include preprocessing to establish an x-ray source, appropriate interaction cross sections, phantom spatial mesh, followed by the radiation transport calculation, and postprocessing computation data to yi eld absorbed doses and absorbed dose volume histograms, etc. The details of this methodology are pr esented in the paragraphs below. 3.2 UF Series B Pediatric Phantoms The UF Advanced Laboratory for Radiatio n Dosimetry Studies (ALRADS) has a well established background in work w ith detailed human phantoms and phantom data. Following the development of their CT-Contours segmentation so ftware, and the construction of their first tomographic dosimetry model (UF Newborn) the ALRADS research team focused on constructing an expanded series of pediatric tomographic phantoms (Series B phantoms), this time using live patient CT images from the Sha nds image archive. A total of five additional phantoms have been constructed from CAP (chest abdomen pelvis) and head CT scans of pediatric patients examined at the University of Florida (UF) Shands Childrens Hospital. Patient inclusion criteria included those patients displaying normal or near-normal anatomy at their respective age and gender (e.g ., avoidance of regions of tumo r, edema, organ enlargement) Table 3-1. While the previous phantoms preserved the body dimensions and organ masses as seen in the original patients who were scanne d, comprehensive adjustments were made for the Series B phantoms to better ma tch International Commission on Radiological Protection (ICRP) age-interpolated reference body masses, body heights, sitting heights, and internal organ masses43.

PAGE 60

60 Table 3-1. Computed tomogr aphy image sources for the development of the UF pediatric phantom series Figure 3-2. The UF Series B pe diatric phantoms series (Newborn, 9-months, 4-years, 8-years, 11-years and 14-years old patients)43 For this work, a pediatric patient (an 11-year-old male, depicted in Figure 3-2) is voxelized with high resolution using 0.94 mm 0.94 mm 6. 00 mm voxels as a case study. This geometry data was then collapsed into a 4.7 mm 4.7 mm 30.0 mm voxelized model using GHOST-3D, described in section 3.3. Subsequently, using PENMSH-XP, the co llapsed model was cast

PAGE 61

61 into a 3-D spatial grid distribution and sub-divi ded into coarse mesh grids, each containing a corresponding number of fine mesh cells. 3.3 PENTRAN-MP Code System Currently, the PENTRAN-MP code system is composed of 3 levels of calculation, including pre-processing algor ithms that include the GHOST -3-D phantom voxel collapsing (back-thinning) code, the DXS x-ray source generation code 45, the PENMSH-XP 3-D mesh distribution and input generato r code, and the CEPXS code (a product of Sandia National Laboratories) for generating multigroup cross sections 33. The radiation transport calculations are carried out in PENTRAN, and post-processed using PENDATA to gather parallel distributed data, and 3-D-DOSE to render user specified absorbed dose data. 3.3.1 Pre-processing Codes 3.3.1.1 The collapsing code (GHOST-3-D) The GHOST-3-D, developed by Ghita and Al-Bas heer, is a code used to input CT-scan binary data obtained from a high resolution voxelized human phantom. There are several constraints in simulating mega-voxel models, including balancing limited computer memory (even with the benefit of parallel memory storag e, as in PENTRAN) with proper discretization, etc. To optimize accuracy and computational wo rk in representing the voxelized phantom, we developed a code, GHOST-3-D, that provides an equivalent model with a reduced number of meshes (subject to user-derived accuracy targ ets), overcoming the mentioned difficulties. 3.3.1.2 The mesh generator code (PENMSH-XP) PENMSH-XP35 by Yi and Haghighat is a 3-D Cartesian-based mesh generator that prepares discretized material and source distribu tions for the PENTRAN particle transport code. PENMSH-XP is capable of performing two di fferent methodologies in generating PENTRAN input files. In the first methodol ogy, a physical model is partitione d into slices al ong the z-axis,

PAGE 62

62 referred to coarse z-levels, and then each z-level is partitioned into single coarse levels of 3-D x-y-z fine mesh grids. Predefined shapes can also be used to ge nerate the desired shapes in a particular model. A second met hodology is based on two binary input files, one for the phantom geometry, which can be established from real CT data, and the other containing the source density for each voxel/fine mesh, which can be used to specify a fixed-source distribution in the model. By combining the two files, PENMSH-X P directly generates an input deck for the PENTRAN transport code. 3.3.1.3 Source spectra and cross-sections generation Based on the bremsstrahlung and ch aracteristic parametric functi ons of the model proposed in 1991 by Tucker, Barnes, and Chakraborty 46 the DXS code, principally developed by Monica Ghita, generates x-ray spectra in the diagnos tic range (50-140 keV), according to the input parameters (tube potential, anode angle, type and amount of filtration), and accommodates them to any energy group structure. Optionally, as part of the PENTRAN-MP package, DXS also generates the input files for CEPXS (Sandia Na tional Lab), the x-ray cross-section generator code, corresponding to a desire d energy group structure and PN order. Group-dependent crosssections rendered by CEPXS are then stripped fr om the output file and adapted to a format suitable for the PENTRAN deterministic calculati ons using another code developed by Ghita, GREPXS47. 3.3.2 Transport Calculation Transport calculations were performed using the PENTRAN SN and MCNP5 Monte Carlo as discussed in Chapter 2.

PAGE 63

63 3.3.3 Post-Processing Codes 3.3.3.1 Data management utility (PENDATA) Immediately following a transport calculation, a data management utility, PENDATA, seamlessly gathers parallel stored data automatically following a parallel PENTRAN run, and provides several options for the user in stripping results from parall el output files, including data extractions from binary file storage which is an essential utility for a c ode with parallel I/O. 3.3.3.2 Dose calculation code (3-D-DOSE) The 3-D-DOSE code, developed by Al-Basheer is used to re nder absorbed doses based on photon fluence results derived fr om the PENTRAN transport calculations. In doing so, a methodology for determining absorbed dose in the phantom based on photon fluence was developed for situations where electron transport effects do not require specific treatment. CPE is assumed to be established in a volume when the energy carried into the volume by charged particles is balanced by the energy carried ou t of the volume by the charged particles. In the example simulation in Section 3.4, the maximum energy used is 90 kV, for which the range of the secondary electrons produced is smaller than smal lest voxel dimensions used in the model. Moreover, the atomic number of the material Z is relatively low, excluding major production of bremsstrahlung radia tion, thus depositing all th e photon energy in the voxel, achieving CPE. Consequently kerma is equal to absorbed dose in this case. We note this is not true for regions in close proximity to the su rface of the model using higher energies due to absorbed dose buildup, where corrections fo r secondary electron tr ansport may be quite significant, especially for high energy beams. After extracting 3-D scalar fluxes from the phantom model we apply the 3-D-DOSE c ode. Based on fitted mass-energy absorption coefficients 44, shown in Figure 3-3, and the scalar fluxes from the PENTRAN calculations, 3-D-

PAGE 64

64 Lung(ICRU-44)1.00E-02 1.00E-01 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E-031.00E-021.00E-011.00E+001.00E+011.00E+02Energy(Mev)Mass-absor p( cm2/ g) ICRU-44 Fit Cortical Bone (ICRU-44) 1.00E-02 1.00E-01 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+041.00E-031.00E-021.00E-011.00E+001.00E+011.00E+02Energy(Mev)Mass-absor p( cm2/ g) ICRU-44 FIT Soft tissue (ICRU-44)0.01 0.1 1 10 100 1000 10000 1.00E-031.00E-021.00E-011.00E+001.00E+011.00E+02Energy(Mev)Men/roh ( cm2/ g) ICRU-44 FIT DOSE calculates absorbed dose in each voxel yi elding a whole-body absorbed dose distribution and dose-volume histograms for cr itical organs of interest. Figure 3-3. Values of the mass energy-absorption coefficient, and the fitting function as a function of photon energy, for Lung (ICRU44), soft tissue (ICRU -44) and cortical bone (ICRU-44) 3.4 Computational Example For this work, a pediatric patient (an 11-year -old male) was voxelized with high resolution using 0.94 mm 0.94 mm 6.00 mm voxel size (398 voxel 242 voxel 252 voxel) as a case

PAGE 65

65 study. This geometry data was then collapsed into a 4.7 mm 4.7 mm 30.0 mm voxel size (79 48 50 voxels) model using GHOST-3-D. Subs equently, using PENMSH-XP, the collapsed model was cast into a 3-D spatial distribution a nd sub-divided into six coarse mesh z-levels containing five coarse meshes in x-y domain, and these containe d 6320 fine mesh cells (Figure 3-4, 3-5). An isotropic volumetric (17.8 cm x 3.65 cm x 30 cm) x-ray source was placed above the upper left side of the phantom. Although the model has relatively sm all fine mesh sizes, using S32 Legendre-Chebychev quadrature, the adaptive differencing, and CEPXS cross-sections with P3 anisotropy, we were able to eliminate any unphysical oscillations that may result in the deterministic solution. Using 4, 6, and 8 energy groups, the SN model was executed on 16 Opteron processors and was completed in 1.8, 2.1 and 3.8 hours, respectively. With the general purpose radiati on transport code MCNP5, we simu lated an equivalent model to the one simulated by PENTRAN. The collapsed 3-D binary files composing the UF voxel phantom was converted to ASCII formatted matrix files and then incorp orated into a MCNP5 lattice defined input. An input d eck example for the MCNP5 model is presented in Appendix B. Volumetric flux (F4) mesh ta llies were used in the Monte Carlo geometry and were equivalent to the discretized SN volumes all over the model. All simulations were performed using the parallel cluster Einstein composed of 16 processors at 2.4 GH z with 4 Gb/processor, for a total of 64 Gb of parallel memory used fo r general purposes. Parall el MCNP5 results in the source region converged with a st atistical error less than 4% in the source region on average.

PAGE 66

66 Figure 3-4. UF-Series B phantom (11 year s old male) with (398 242 252 voxels); the corresponding PENTRAN back-thinned mo dels with (132 80 84 voxels) and 72 materials (upper right), and another with (79 48 50 voxels) and 66 materials (lower right)

PAGE 67

67 Figure 3-5. Transversal view of the 11-year-old male Series B original phantom (upper left), collapsed into 189k voxels and 4 materials ( upper right), collapsed into 900k voxels and 4 materials (lower left) and collapse d into 900k voxels and 72 materials (lower right) 3.5 Computational Results 3.5.1 Flux Comparison of Deterministic and Monte Carlo Results Comparison of the SN and Monte Carlo results for the scalar fluxes along the source central axis and in the source region along the z axis at different depths for different energy groups were made. Figure 3-6 depicts a 3-D scal ar flux distribution computed by PENTRAN for the reference case. This simulation was based on the use of an S32 angular Legendre-Chebychev quadrature (1088 directions/mesh) with P3 scattering anisotropy using th e CEPXS with 8 energy groups.

PAGE 68

68 Figure 3-6. x-ray 3-D Group 1, 3, 5 and 7 scalar flux computed by PENTRAN with the CEPXS cross section library, S32 angular quadrature (1 088 directions) with P3 scattering anisotropy Figure 3-7. Flux distributions using S32 with P3 scattering an isotropy in PENTRAN using the CEPXS library vs. MCNP5 Monte Carlo result s for energy groups 1, 2, 3 and 4 along the source central axis. Mont e Carlo uncertainties were less than 5%, 4%, 4%, and 3% respectively

PAGE 69

69 Figure 3-8. Angular fl ux distributions using S32 with P3 scattering anisotropy in PENTRAN using the CEPXS library vs. MCNP5 Monte Ca rlo results for energy groups 5, 6, 7 and 8 along the source central axis. Monte Carlo uncertainties were less than 2%, 3%, 6%, and 30% respectively Figure 3-9. Angular fl ux distributions using S32 with P3 scattering anisotropy in PENTRAN using the CEPXS library vs. MCNP5 Monte Ca rlo results for energy groups 1, 3, 5 and 7 along the z axis at diffe rent depths in the phantom. Monte Carlo uncertainties were less than 5% at y=4.3 cm, 6% at y=13.5 cm

PAGE 70

70 As shown in (Figures 3-7, 3-8, and 3-9), the PENTRAN results with S32 LegendreChebychev quadrature are in very good agreemen t with the MCNP5 predictions in the source region; MCNP5 mesh tallies were converged to <6 % rela tive error, and typi cally < 2% for most of the results for the reference case (using 8 en ergy groups). Overall, this figures show that deterministic techniques are capable of producing accu rate results within th e statistical error of Monte Carlo methods. Since the uncertainty is pr oportional to 1/ (number of histories), to achieve an equivalent-uncer tainty Monte Carlo simulation with converged mesh tally information 70 cm away from the source in energy group 1, we esti mate over 200 hours are needed for MCNP5 executed on 16 processors to achieve fluxes to within 5% for this group; longer times would be require d for lower energy x-rays. 3.5.2 Dose Calculations and Dose Volume Histograms (DVH). Using 3-D-Dose, 3-D absorbed dose distributions were calc ulated in each voxel in the whole model of the 11-year old male phantom model for broad-beam ex ternal photon field. Figure 3-10 shows the stages of calculation in th e PENTRANP-MP code system that results in 3D Dose distributions for the entire phantom. Figures 3-11, 3-12, a nd 3-13 show selected slices of the phantom and the corresponding absorbed dose distribution, while Figure 3-14 depicts selected organs with correspondent 3-D Absorbed dose distributions.

PAGE 71

71 Figure 3-10. Simulation met hodology for 3-D absorbed dose computations using PENTRANMP code system. Left: UF series B v oxelized phantom, middle: correspondant PENTRAN input, right: 3-D abso rbed dose distribution after irradiation with a x-ray source a.

PAGE 72

72 b. Figure 3-11. CT view of the 11-year-old male Se ries B original phantom collapsed into (a) 189 k voxels and the correspondent absorbed dos e distributions (b) 900 k voxels and the correspondent absorbed dose distributions a.

PAGE 73

73 b. Figure 3-12. CT view of the 11-year-old male Se ries B original phantom collapsed into (a) 189 k voxels and the correspondent absorbed dose distributions (b) 900 k voxels and the correspondent absorbed dose distributions a.

PAGE 74

74 b. Figure 3-13. CT view of the 11-year-old male Se ries B original phantom collapsed into (a) 189 k voxels and the correspondent absorbed dose distributions (b) 900 k voxels and the correspondent absorbed dose distributions Figure 3-14. Simulation met hodology for 3-D organ absorbed dose computations using PENTRAN-MP code system

PAGE 75

75 Dose volume histograms (DVHs), for a selected nu mber of organs giving the percentage of tissue volume (organ) that receives an absorbed dose greater than the value indicated on the abscissa, were calculated and compared to thos e results from the equivalent Monte Carlo model. As anticipated, DVH results are in good agreement for the SN and Monte Carlo radiation transport simulations (Figure 3-15). However, Monte Carlo calculations can require substantial run times for low precision at o ut of field distan ces in the phantom, particularly as photons down-scatter to lower energies The Monte Carlo results comput ed values will eventually converge with increased run time as more pa rticles reach sites distal to the source. a. b. Figure 3-15. Dose Volume Histograms (D VH) computed based on PENTRAN (S32,P3) transport results using the CEPXS library vs. DVHs computed using MCNP5 Monte Carlo results. MCNP5 Monte Carlo results required 16 hour s 16 processors compared with the PENTRAN-MP SN solution, requiring <1.9 hours on 12 processors for the whole-body computation 3.5.3 Parallel Computation For the UF 11 year old male phantom problem, a decomposition strategy was used to assist the parallel performance. In Table 3-2, we present the cumulative problem time and memory

PAGE 76

76 requirements for deterministic calculations disc ussed in previous sections using different quadrature sets. Table 3-2. Total number of directions, Cumula tive Problem Time required for different energy groups, P3 scattering, and 189k fine mesh cells Parallel Decomposition Number of Energy groups SN (relative workload) Number of Processors A E S Cumulative Problem Time (hours) 4 42 (2) 12 4 1 3 4.1 6 24 (1) 12 8 1 2 2.1 8 32 (2.32) 12 4 1 3 3.69 8 32 ( 2.32) 16 8 1 2 3.06 Parallel decomposition results for solving the phantom are presented in Table 3-2 based on a constant problem phase space of 189,000 cells with P3 anisotropy and diffe rent numbers of energy groups numbers and several angular quadratures. A, E, S refers to the angular, energy group, or spatial decomposition applied to solve th e problem using parallel computing. From the table data, it can be observed th at increasing the number of dire ctions requires longer running times that essentially scale with processor workload. 3.6 Discussion and Findings We have demonstrated a methodology in corporating a discrete ordinates (SN) approach that is an effective alternative to the Monte Carlo method for accurately solving the transport equation over an entire human phantom. Models he re were based on the UF series B pediatric phantoms developed by the University of Florid a ALRADS group. We simulated the complete phantom discretized using 900k/186k voxe ls, and demonstrated that the SN method in the PENTRAN-MP code system, developed as a te am effort, readily provides accurate whole body fluxes. Radiation transport throughout the phantom encounters few interactions in low density

PAGE 77

77 materials, such as air. Attention was given to the quadrature level, using at least an S32 LegendreChebychev angular quadrature, yielding 1088 total di rections per mesh-cell; we were able to accurately obtain a solution analogous to the MCNP5 Monte Carlo calcula tions. Also, the tally results from the eight energy gr oupsrepresenting the spectrum a nd using the CEPXS multi-group cross-section library, were co mpared to the continuous en ergy MCNP5 Monte Carlo results using the ENDF/B-VI data libraries. Overall, with an adequate quadrature, mesh si ze, and cross-section libr ary, it is possible to produce results with the PENTRAN discrete ordinates code that agree within the statistical uncertainty of the on-axis MCNP5 Monte Carlo ca lculations for the proposed model. Parallel deterministic PENTRAN results were obtained within comparable running times to parallel MCNP5 Monte Carlo calculations for ta lly sites adjacent to the source. A major advantage of the SN method is that it provides a detailed, accurate flux distribution at thousands of cells throughout the sy stem, while the Mont e Carlo method only provides highly accurate values for selected points near the source. Though the use of volumetric mesh tallies can provide global fluxes across a grid spanning the Monte Carlo geometry, the statistical error asso ciated with each mesh tally site that would require, according to estimates, more than 200 hr for tallies 70 cm away from the source to converge to less than 5%., limits their utility in prac tical applications. An equivalent Monte Carlo calcula tion to obtain the flux distribution with reasonable statistical sampling error over the entire problem space may require orders of magnitude longer running times compared to parallel deterministic SN calculations. While more tests are needed, the recently developed Exponential Directional Iterative (EDI) SN differencing scheme was excellent in d ealing with fine inhomogeneous material

PAGE 78

78 densities inherent in medical physics problems, properly handling numerical radiation transport physics for diagnostic energy beams. Overall, for the energy ranges used in th e problem, the secondary electrons need not specifically be modeled using charged particle transport methods to accu rately determine global absorbed dose distributions. This is due to the fact that for an SN calculation with an appropriate numerical fidelity (evidenced by the excellent agreement with Monte Carlo), secondary electrons that result from photon interac tions in the phantom have equi valent ranges smaller than the minimum SN spatial grid interval used to represent the anatomical structure. Alternatively, at higher photon energies (not consider ed here), this is not the case, and charged particle transport and corresponding interactions yielding energetic electr ons must be considered over the problem phase space to yield an accurate absorbed dose distribution. In Chapter 4, a new EDK-SN approach is presented to account for higher energy beams and sec ondary electron absorbed dose.

PAGE 79

79 CHAPTER 4 ELECTRON ABSORBED DOSE KERNELS TO ACCOUNT FOR SECONDARY PARTICLE TRANSPORT IN DETERMINISTIC SIMULATIONS In previou s chapters, Charged Particle Equilibrium (CPE) and its limitations were presented, and that in cases with high energy photons, corre ctions for secondary electron transport may be significant. As mentioned, this is readily treated in Monte Carlo codes, yet quite difficult to treat explicitly in deterministic codes due to th e large electron ranges and added numerical complexities in reaching conve rgence in photon-electron transport problems10 The goal here is to produce a methodology to accurately estimate organ absorbed doses and attribute whole body absorbed doses incurred from irradiations characteristi c of external beam therapy using the CT-based anatomical patient geometri es in 3-D deterministic discrete ordinates (SN) radiation transport models. Accurate absorbed doses will be rendered using pre-computed, high resolution Monte Carlo based electron kernels. In doing so, the flux and the curren t provided by the SN method will be used to both scale and pr oject the absorbed dos e using the Monte Carlo based kernels, and map it to the surrounding voxels. Hence, this a pproach preserves high fidelity and accuracy, accounting for complete photon transpor t, with the final absorbed dose delivered from electron transport propagate d by coupling with 3-D determinis tic photon transport; this is what makes this approach novel. In this chapte r, water phantom results using a 0 to 8-MeV step uniform beam will demonstrate that absorbed doses can be accurately obtained within the uncertainties of a full physics Monte Carlo simulation. 4.1 Introduction to The EDK-SN Method Various modalities are applied in absorb ed dose calculations for high energy photon beams, including empirical models that are based on measurements made in the clinic2, or theoretically based models, such as the application of Monte Carlo techniques3,48. Alternatively, as mentioned earlier, deterministic methods have been implemented to assess absorbed dose

PAGE 80

80 calculations20,21, although experience has shown that 3-D deterministic electron transport is quite challenging in order to yield a converged solution. To properl y treat 3-D electron transport physics deterministically, yet still achieve reasonably fast and accurate whole body computation times for simulations with high energy photons, we have developed angular-energy dependent transport electron dose kernels, or EDKs that can be coupled to SN solutions. These kernels were derived via full physic s MCNP5 Monte Carlo el ectron transport simulations, and are applied based on deterministic photon soluti ons over the problem phase space, thereby accurately accounting for the abso rbed dose that is ultimately rendered by charged particle electron transport. This is possible because th e PENTRAN discrete ordi nates code preserves detailed photon flux and current data on a voxel -by-voxel basis, so that accurate whole body absorbed doses may be rapidly achieved for hi gh energy photon sources by performing a single deterministic SN multigroup photon calculation on a parallel cluster. This is followed by linking the SN-derived photon fluxes and net currents to our Monte-Carlo-based El ectron Dose Kernels to account for a full-p hysics absorbed dose; we call this the EDK-SN methodology for determining total body dosimetry. This total body dosimetry enables dose to be computed across the entire phantom phase space, wherever it is required, presented in the following sections. 4.2 Monte Carlo Based Dose Kernels Using Monte Carlo calculations, electron abso rbed dose kernels for pre-determined photon energy groups were generated; energy was deposited in a collective cloud of voxels ( i, j, k ) composed of tissue, where absorbed doses attributed to these voxels are directly caused from incident primary photons in a given energy gr oup that propagate electrons from voxel, ( i',j,k ) the dose-driving voxel or DDV When producing these electron absorbed dose kernels, the photons having energies lower than the limits of the energy group are cut off as they are

PAGE 81

81 generated to prevent a cascadi ng contribution to the EDK for lower energy groups. These kernels were then applied along the direction of the ne t current and scaled by the magnitude of the photon flux on a fine mesh basis pr ovided by the PENTRAN discrete ordinates code, as well be explained in later sections. Figure 4-1 illustrates the DDV, with an indicated direction of photon current. The Electron absorbed Dose Fractional kernel in voxel ( i, j, k ), given as the EDFg, due to a primary photon group, can be determined in term s of the initial photon energy for a particular energy group g By partitioning the energy deposited in voxel ( i,j,k ) into multiple energy bins that are aliased to the SN energy group structure, we can pred ict the fractional electron absorbed dose kernel contribution per unit photon flux per source particle: (,,)(,,)(',',')gEDFijkEDKijkijk gg MC (4-1) Therefore, the SN transport-guided absorbed dose rate can then be obtained for any voxel location ( i, j, k) by summing the product of the EDFg( i,j,k ), the particle normalization factor ( ), and the SN scalar flux (,,)NgSijk divided by the voxel mass M (i, j, k): (,,)[(,,)](,,)/(,,)Ng ggsS s D ijkEDFijkijkMijk (4-2) This procedure will enable the detailed info rmation to be obtained on the absorbed dose deposited in the model by scaling the EDKs by the photon flux and pr ojecting them by along photon net current vectors computed for each voxe l. This will directly account for secondary electron physics in each voxel to achieve an accurate total body attributed absorbed dose. The particle normalization factor in equation 4-2 correctly scales the number of voxels linked using the EDK, and is set by the deterministic grid spacing. We note that this factor is specific to the grid used to estimate the EDK-SN. Consequently, to determine this factor, the absorbed dose is calcu lated through the EDK-SN model and should be scaled to the absorbed

PAGE 82

82 dose calculated via a Monte Carlo full physics absorbed dose in an equivalent model. This factor is a kernel-specific f actor, and once determined, it can be used for any execution implementing the corresponding electron kernels, provided grid spacing is consistently used. Figure 4-1. Procedure to produce E DK based on Monte Carlo Calculations 4.3 EDK Determinations Based on The MCNP5 Monte Carlo Code In the MCNP531 Monte Carlo code, there are three different methods by which one can tally to yield absorbed doses to be used for inte grated organ absorbed dose calculations; these are achieved by using either a cell flux tally (F4), an energy deposited tally (F 6), or a pulse height tally (*F8), and each has a different basis for computing the absorbed radiation dose14,31. To illustrate this, we compare absorbed doses computed using the F4, F6 and *F8 tallies, respectively to highlight the differences in thes e computations, and establish a procedure that ensures that our electron kernel s are calculated accura tely, and the Monte Carlo transport and subsequent absorbed doses determined as a be nchmark are completely based on a detailed physics treatment.

PAGE 83

83 4.3.1 Computational Model To highlight the differences between the different tallies, used for absorbed dose estimation, in the MCNP5 code, we c onsidered a box of 11x11x11 cm3 composed of a water cube containing a 1x1x1 cm3 8 MeV forward peaked X-ray source placed at the center of the cube Figure (4-2). The size of this model allows complete deposition of secondary electrons produced from the Xray source. A flux mesh-tally, using DE absorbed dose energy and DF absorbed dose function cards, was generated to cover the entir e model. Moreover, since no mesh tally is available for *F8 and F6 tallies in the MCNP5 code, a 3-D tally distribution of cells, equivalent in purpose to the MCNP5 F4 mesh tally, were implemented in the model. This model, with the implemented tally sites, was used for highlighti ng the differences between the different tallies, and consequently was used for reporti ng a scaled full physics absorbed dose. 4.3.2 MCNP5 Tallies Comparison The source in the center of the model was app lied to generate a pencil beam impinging perpendicularly, along the z-axis, on the box of wate r. As shown in Figures 4-3, 4-4, and 4-5, the absorbed dose is highly forward peaked, with little dose resulting from back-scattered radiation. Absorbed doses resulting from the use of an F4 volume flux tally method, using a DE dose energy and DF dose function cards, compared to an F6 tally using bo th the photon-electron mode (p, e mode) and photon only (p mode) were a ll in agreement, since the two tallies depend on similar physics in predicting absorbed dose. Thes e results are depicted in Figure 4-3. On the other hand, when comparing absorbed doses com puted using a *F8 using a fully coupled photon electron transport mode (p, e mo de) and the F6 derived dose ta llies, the differences were significant, as depicted in Figure 4-4. This is expected, since F6 tallies are analogous to F4 tallies in MCNP, and can be directly aliased to assumptions equivalent to CPE, while *F8 tallies are determined based on full charged particle transport physics, as discussed below31.

PAGE 84

84 Figure 4-2. Monte Carlo model used for comparisons between MCNP5 pulse height tally, energy deposited tally and cell flux tally employing dose conversion factors (DE, DF). The source is located at the dose driv ing voxel DDV Figure 4-3. Comparison between F6 (p, e mode), F6 (p mode), and F4 ( DE, DF) tallies in MCNP5, statistical uncertainty was less than 1%

PAGE 85

85 Figure 4-4. Comparison between F6 (p, e mode) a nd *F8 (p mode) tallies in MCNP5, statistical uncertainty was less than 1% Figure 4-5. Comparison between *F8 (p, e mode) and *F8 (p mode) tallies in MCNP5, statistical uncertainty was less than 1%

PAGE 86

86 When comparing *F8 (p mode) transport and *F 8 (p, e mode) transport, (Figure 4-5) shows significant differences in the outcomes for the sa me simple model. This directly results from accessing different calculation modes in MCNP5 with regard to electron physics. If electron transport is turned on (mode p, e), then all ph oton collisions except coherent scatter lead to electrons that are banked for later transport. A lternatively, if electron transport is turned off (achieved by omitting the e on the MCNP5 Mode card), then a thick-target bremsstrahlung model (TTB) is used. This model generates electrons, but assumes that they are locally slowed to rest. The TTB production mode l contains many approximations co mpared to models used in actual electron transport. The choice of which tally is to be used and the detailed physics treatment by which the Monte Carlo code is execu ted is essential to th e success of the EDK-SN method. In conducting computations for generati on of the EDKs, we used the full detailed physics treatment, implementing a *F8 (p, e mode) tally, as opposed to the alternative methods that lead to less robust electron physics, es sentially employing a collisional kerma. A detail discussion on photons interactions in matter is presented in Appendix A. 4.4 EDK -SN Code System Methodology 4.4.1 Net Current For any SN calculation, the PENTRAN discrete ordina tes code preserves angular information explicitly in parallel data storage arrays. This allows one to calculate netJ and to apply the precomputed electron absorbed dose kernel along th e net current direction, Figure 4-6, where: (4-3) (4-4.a) nxnynzijk netJJJJ nxnxnx J JJ

PAGE 87

87 (4-4.b) (4-4.c) (4-5) Where u j is the photon net current unit vector. Figure 4-6. Deterministic calculation will provide the angular information explicitly 4.4.2 MCNP Based Electron Kernels When generating these electron kernels, a si ngle beamlet of mono-en ergetic photons, for a certain energy group, interacts at the center of a water phantom, as disc ussed in the previous section. For the purpose of this study, we generated the EDKs fo r 16 photon energy groups covering the energy spectrum from 0 to 8 MeV. Based on our previous study, as discussed in chapter 1, these energies will violate CPE for the model considered. Moreover, the width of the energy group is set to be 0.5 MeV, minimizing the effect of energy discretization in an acceptable SN running time (we note that energy group disc retizations can be optimized for better nynyny J JJnznznz J JJ net u net J j J

PAGE 88

88 time performance, although this is not considered here). Kernels are computed and then scaled by photon flux in the DDV in dose kernel voxels (pro jected from the DDV) along the net current direction of the deterministica lly computed beams of same en ergy group (Figure 4-7). It is important to note that for each energy group calcu lation, the Monte Carlo model was set to cutoff photons with energies lower than the lower lim it of the energy group being modeled. This is important to prevent any cas cading contribution to the EDK to lower energy photons. Figure 4-7. For each energy group, using Monte Carlo radiation transport, a pencil beam of mono-energetic photons interacts at the center of a water phantom These Electron Dose Kernels (EDKs) can th en be implemented to enable direct attribution of the final equivalent absorbed dose initiated by the photon beamlets. The EDKs can also be placed in various tissue types and ener gies relative to a spec ific direction of photon beam determined from solution of the Boltzma nn transport equation using PENTRAN. This will permit complete attribution of the absorbed dose ultimately due to electron straggling, where the EDK coupled with SN (EDK-SN) provides an equivalent elect ron absorbed dose look-up for each photon beamlet in a discrete ordinates computation. However, to achieve precise solutions using the EDK-SN, the SN calculation should be carried out with great attention to phantom model representation and meshing, proper quadrature sets, cross s ection libraries, and

PAGE 89

89 SN differencing schemes; these issu es were investigated and presen ted in Chapter 2 in assessing what critical discretizations are required to yield accura te photon transport soluations42. 4.4.3 Euler Angles Euler angles, described in (Figure 4-8), pr ovide a means of representing the spatial orientation of any frame of angular space as a composition of rotations from a given reference frame. The translations afforded by Euler angles were implemented to enable projections of the EDKs for a collection of cells aliased to a single reference pencil beam v ector, as modeled in the Monte Carlo simulations (as in Figure 4-2) to generate EDKs to an arbitrary net current vector specifying the direction of the photons derived from SN computations. In the following discussion, we denote a fixed system in lower case ( x,y,z ) and a corresponding ro tated system in upper case letters ( X,Y,Z ). We use the term "Euler Angle" for any representation of three dimensional rotations of the electron kernel absorb ed dose matrix that aligns with the direction of the net current for each DDV. Figure 4-8. Euler angles, zyz convention. Similar conventions can be obtained by following of the following order in rotation ( zyz, xyx, xzx, yzy, yxy ) Where :

PAGE 90

90 is the angle between the y-axis and the line of nodes is the angle between the z-axis and the Z -axis is the angle between the line of nodes and the Y -axis. The above description works for the zyz form. Similar conventions are obtained just renaming the axis ( zxz, xyx, xzx, yzy, yxy ). There are six possible combinations of this kind, and all of them behave in an identical way to the one described before. Just imagine three rotations over the set of axis form ing the rotation matrix R [][()][()][()]zyzRRRR (4-6) Where R is the rotational element matrix Depending on the convention used for the calculat ing these angles, they can be related to the angles for the net current obtained from th e PENTRAN run. Assuming that rotated matrix indices to be known, we generated the vector A for each point from the center of the matrix, where the dose values of this vector form th e matrix dose point to be projected onto the SN matrix: ['][][] ARA (4-7) 11[]['][][][] R ARRA (4-8) 1[][]['] A RA (4-9) When calculating the original un-rotated A vector for each point, we interpolate the vector location, in 3-D, for the corresponding dose valu e based on the original EDK matrix calculated with the Monte Carlo model. This generates a 3-D translated (by rotation) electron absorbed dose kernel matrix corresponding to a particular direc tion in space. For the current code system, this process was performed for 1848 directions, gene rating 1848 files for each energy group. An example of a rotation of the EDK from a referen ce (Monte Carlo) EDK is depicted in Figures 49.a and 4-9.b.

PAGE 91

91 A B Figure 4-9. EDK-matrix after rotation to direct ion cosines A: (-0.5, 0.35, 0.8) and B: (0.25,0.40,-0.88) relative to the reference coordi nate system (0, 0, 1). Dose Driving Voxel is located at (0,0,0)

PAGE 92

92 4.4.4 Ordinate Projection In the discrete angle approximations, it is assumed that the EDK-SN is applied on distinct equal weight angles (,)(,,) To generate these angles, the following relations should hold for any given octant: orderN (4-10) The orderN in any 3-D EDK-SN quadrature corresponds to the number of levels from each direction cosine on the unit sphere, and there are (2)dirorderorderNNN ordinates on the unit sphere, with _(2)/8diroctorderorderNNN in each octant, with orderN/2 distinct direction cosine values. Figure 4-10 shows orderN42 used rendering 231 direction per octant. (21) 2i (4-11) (21) () 22j i (4-12) where i = orderN For any unit angular direction vector the ordinate must lie on the unit sphere: 2221 (4-13) ()()SinCos (4-14.a.) ()()SinSin (4-14.b.) ()Cos (4-14.c.)

PAGE 93

93 Figure 4-10. First octant of S42 equal we ight angles used for projecting the EDK-SN 4.4.5 Final Dose Calculations The EDK-SN model specific portion, described in Figure 4-11, serves as the critical link in the EDK-SN system to accumulate the absorbed dose in each fine mesh. The algorithm determines the accumulated energy deposited in each voxel for each photon energy group, based on the SN net current vector and is scaled to the approp riate dose by the photon flux. Finally, the code accumulates all the absorbed dose contributions from all the energy groups in to a 3-D total absorbed dose matrix. Figure 4-11 shows a flow ch art representing the EDK-SN methodology. An energy dependent EDK was generated using high reso lution direction projection sets and a MC pre-computed EDK for a single direction. These ke rnels were generated over the uni t sphere of ordinates via the Euler Angles described in Section 4.4.3. Concurrently, when generating the EDK, a FORTRAN90 module will be generated for later use in the dose calcula tion portion of the

PAGE 94

94 Figure 4-11. A flow -chart for the EDK-SN Code System methodology. The function of th is module is to define a on which the net current vector will be discretized into. For each specific model, after executing the de terministic calculati on using PENTRAN, we generate 3-D fluxes, current vectors, and mate rial distributions for each energy group for a particular model. These 3-D da ta distributions, along with precomputed EDKs, will then be

PAGE 95

95 implemented to calculate the 3-D absorbed dose distribution, for each energy group, scaled by the flux along the current vector direction. Finally the EDK-SN will accumulate these individual energy group dose contributions into a 3D total absorbed dose distribution. In Section 4.7, we describe the parallel code algorithm. 4.5 External Source Benchmark An external source problem for benchmarking purposes is presented to demonstrate the EDK-SN methodology and provide a dire ct comparison with a full -physics Monte Carlo solution. The model considered was a homogenous block of water (45 45 453cm) with a 15 15 1 3cm source (defined for 16 energy groups spanning from 0.0 to 8.0 MeV cast as a flat-weighted energy distribution source term). A 3-D spat ial mesh distribution was generated by the PENMSH-XP mesh generating code in the PENTRAN system19. The model was then subdivided into 3 z-levels; each z-level was then divided equally into 9 coarse meshes, each coarse mesh contained 3375 fine mesh cells; therefore, a to tal of 91,125 fine mesh cells were used in the complete SN problem as shown in Figure 4-12.A. A. B Figure 4-12. External Source impinging on block of water (left: Mesh di stribution generated by PENTRAN package; Right: MCNP5 Cell/Surface schematic input)

PAGE 96

96 For the purposes of direct comparison to the Monte Carlo method, the MCNP5 Monte Carlo code is used to simulate an equivalent external sour ce problem as shown in Figure 4-12.B. For validation of the EDK-SN approach, the equivalent MCNP5 model is used to calculate the absorbed dose along the central axis and across the model for these studies using MCNP5 pulse height tallies (*F8), with full physics transport (p, e mode) to al low direct comparison of fluxes and absorbed dose values for the Cartesian fine mesh cells used in PENTRAN-MP and the EDKSN approach. 4.6 Comparison of Deterministic and Monte Carlo Results Using a uniform 1 3cm mesh, PENTRAN was used to calcu late the flux distributions and the net current for each fine mesh cell in th e system. Based on the PENTRAN results and the pre-calculated electron kernels, the EDK-SN code was used to produce 3-D absorbed dose rate distributions for all energy groups. The absorbed dose rates for energy groups (7.5-8.0 MeV) and (0.5-1.0 MeV) are depicted in Figures 4-13A and 4-13B, respectively. These figures provide a clear insight of the 3-D absorbed dose behavi or for these two different energy groups. For example, the high energy photons indicate more fo rward peaked absorbed dose distributions (as expected), and, while still forward peaked, the lower energy photons indicate more isotropic behavior, also as expected. Figures 4-14 and 4-15 depict the absorbed dose rate for each energy group (over a total of 16 energy groups) calculated by EDK-SN along the central axis, and laterally across the problem respectively. Alternatively, Figures 4-16 and 417 show the MCNP Monte Carlo total absorbed dose along the central axis and laterally ac ross the problem after 1000 minutes using 16 processors of same PC cluster. The statistical uncertainty 2 on average, of the *F8 MCNP5 Monte Carlo tally was 6.0% for the study along the central axis (x-axis), and 4.0% along the y-

PAGE 97

97 axis, 2 cm away from the source, in the source region. On the other hand, the average absolute relative differences between the EDK-SN absorbed dose calculation and the full physics MCNP derived absorbed doses were 3.7% along the cent ral axis (X-axis), and 2.5% along the Y-axis, 2cm away from the source. A B Figure 4-13. External Source 3-D absorbed dos e distribution for 2 Energy groups computed by EDK-SN based on PENTRAN-MP A. EG(7.5-8.0Mev) and B. EG(0.5-1.0Mev) Figure 4-14. Energy group absorbed dose rate di stributions (in Mev/ g. sec) along the X-axis using the EDK-SN method based on PENTRAN-MP

PAGE 98

98 Figure 4-15. Energy group absorbed dose rate distributions (Mev/ g.sec) along the y-axis using the EDK-SN method based on PENTRAN-MP Figure 4-16. EDK-SN Total absorbed dose rate distributions along the X-axis compared to MCNP *F8 tally derived absorbed doses. MCNP uncertainties (2 ) average (6.0%) in the Source region. The average absolute re lative difference betw een the two methods was 3.7%

PAGE 99

99 Figure 4-17. Total absorbed dose rate distribu tions of EDK-SN along the Y-axis compared to MCNP *F8 tally. MCNP uncertainty (2 ) average (4.0%) in the Source region. Average absolute relative difference (2.5%) Although small, less than 1%, the ripple apparent in lateral absorbed dose, shown in Figure 4-15 and Figure 4-17, can be explained by the mesh structure used in projecting the absorbed doses along the collective region of elec trons, and low level ray effect, in the SN calculation, causing absorbed dose to oscillate 4.7 Parallelization of EDK-SN Code EDK-SN is designed to run on distri buted memory architectures, where each processor is an independent unit. This system is a distribute d memory computer system, where the processing elements are connected by a network and tare highly scalable. The interconnection scheme among the processors is fundamental for distribute d memory architectures because it affects, in part, the performance of the system. For clustertype architectures, the cluster using identical processors (matched for speed) connected with an Ethernet local area network to enable data transfer among the processor units.

PAGE 100

100 To achieve maximum speedup with the EDK-SN methodology, para llel programming methods were implemented in the EDK-SN code using energy group decomposition via the Message Passing Interface (MPI)49 library. The MPI library is meant to provide an essential virtual topology, process sync hronization, and communication f unctionality between a set of processes (that have been mapped to nodes/server s/computers) in a language independent way, with library specific syntax (bindings), plus a few features that ar e library specific. The EDK-SN code has been designed in ANSI FORTRAN-90 (to take advantage of dynamic array allocation) and parallelized for operation on distributed memory, multiple instruction/ multiple data (MIMD) mach ine architectures using initial standard Message Passing Interface protocols (MPI-1). All Input/ Output (I /O) is performed by each processor in parallel (as required) to include input processing, initialization, processing, and file output. The EDK-SN code carries out the computations with electron dose kernels based on the scalar flux scaling for each energy group independe ntly Consequently, the decomposition used is energy domain decomposition. The basic philosophy of this approach is to decompose energy groups to each processor, and carry the total angular (net current vector) and spatial domain information on each processor. The main advantages of the EDK-SN parallel algorithm is that appreciable speed up can be obtained to determine the dose in any voxel throughout the spatial domain. Since little communication is required between the processors, n processors would solve the problem nearly n-times faster than a single unit. Energy group absorbed doses are partitioned for the entire problem, and partitioned energy group assignments are scaled to n processors for coupling photon transport information with EDKs to produce group absorbed doses over the spatial domain. An MPI_REDUCE call is made to collect all absorbed doses across processors. Since

PAGE 101

101 this work focused on the use of 16 energy groups for a flat weighted 8 MeV high energy source term, the parallel EDK-SN computation could be pa rsed to a maximum of 16 parallel processors. Near linear speedups were achieved. The parallel code follows the following steps in executing the EDK-SN methodology for total dose calculation: Initiate the code on all pro cessors for sets of energy groups, up to one group per processor Sort the locally parallel-a llocated fine mesh distribu tions into a global monotone increasing mesh distributions Create Energy group dose matrix aliased to the location of the DDVs Translate material tags into densities for later calculation Project the Group-dependent E DK based on the group current v ector and material at the DDV Scale group dose using group scalar flux, correc ting for densities at material interfaces Use MPI-REDUCE to sum all doses for the phase space across the parallel processing grid 4.8 Discussion and Findings We examined a new EDK-SN methodology of coupling precomputed Monte Carlo based electron absorbed dose kernels (EDKs ) with the discrete ordinates (SN) solutions to achieve accurate absorbed doses for megavolt energy photon scenarios, since charged particle equilibrium fails and full phot on-electron physics must be acc ounted for to fully attribute absorbed dose. The EDK-SN methodology was compared with the Monte Carlo method and shown to give results consistent within the statistical uncertainty of the absorbed dose computed using Monte Carlo calculations. By following the EDK-SN approach, the absorbed dose applied to tissue from any incident external high energy pho ton beam (pencil, fan, areal, etc) can therefore be readily determined based on a coupling of the deterministic SN photon transport solution from the PENTRAN code using the EDK-SN methodology. As a result, this research rend ers the first

PAGE 102

102 application of integrated electron transport kernels coupled to the SN photon transport methodology to properly account for the total abso rbed dose in high energy photon irradiation scenarios. A fundamental difference between the deterministica lly guided EDKSN technique and the existing convolutionsuperposition te chnique is the fact that our EDK-SN approach is based on a full physics Monte Carlo generated kernel projected using an accurate SN transport solution that provides a net curren t direction and flux scaling to sc ale the absorbed dose using the spectrally dependent EDKs. Overa ll, based on the water phantom problem examined here, with an adequate discreti zation using the SN method coupled to properly computed energy dependent electron kernels, it is po ssible to produce absorbed dose results with the EDK-SN method that are consistent with Monte Carlo calculations A major advantage of the EDK-SN method is that it provides a detailed, accurate absorbed dose distribution wherever required throughout the system, while incorporating the full physics electron transport, and is generally much faster than Monte Carlo. Since PENTRAN is based on a Cartesian geometry, CT voxel data can readily be mapped onto the problem geometry to define the tissue and organ geometry structure. Subsequent discussion will account for material density variations (encountered at tissue-bone interfaces) where we will apply the EDK-SN methodology to a whole body voxelized phantom to rapidly achieve a full physics whol e body, total body absorbed dose.

PAGE 103

103 CHAPTER 5 APPLICATION OF EDK-SN TO MODELS WITH HETEROGENITIES This chapter focuses the many aspects that make up the EDK-SN methodology in order to apply it to a voxelized phant om, validated side-by-side using full physics Monte Carlo calculations. The SN calculations are performed based on a full photon transport calculation using deterministic techniques in a heterogene ous phantom. This eliminates any need for any photon corrections for tissue he terogeneities, since the SN solution yields the complete solution to the Boltzmann transport equation. Moreover, because the EDK-SN calculation is performed over the full spectrum of the radiation for a discrete number of ener gy groups, we can thereby generate separate sets of elect ron absorbed dose kernels to account for the different densities present in the phantom model. In this chapter, we will address applying the EDK-SN method to a heterogeneous phantom, including the application to a new state of the art human phantom. 5.1 Heterogeneous Tissue-Equivalent Phantoms Three slab based phantoms were simulated, using the EDK-SN method, as a prelude to more complex calculations for full human phantoms. Each phantom had a flat weighted 0 to 8 MV x-ray source impinging on 45x45x45 cc soft ti ssue (S) phantom. Two of these phantoms have slabs of 1-cm-thick discs of materials representative of bone (B), or lung tissue (L) (Figure 5-1). 5.1.1 Monte Carlo Simulations In order to compare the accuracy of the EDK-SN methodology, a series of Monte Carlo simulations were also performed using the radi ation transport code MC NP5. An X-ray flat weighted spectrum 8 MV used as input to MCNP5 to guarantee no specific preference was giving to any particular energy gr oup. Cell estimates of tissue absorbed dose were calculated in MCNP5 by tallying the energy depos ited by the photons and their seco ndary electron within 1

PAGE 104

104 cm3 voxels, equivalent to the size of the EDK-SN voxel and the voxel size to be used in the SN calculation. Figure 5-1. Three Monte Carlo phantoms used in the absorbed dose study. Phantom SSSSSSS is a homogeneous soft-tissue (S) phantom; phantoms SSBBSSS and SBLLBSS are heterogeneous phantoms consisting of combin ations of soft and bone tissue (B), and soft (S), bone (B), and lung (L) tissue-equivalent materials, respectively50 After running the three phantoms for 100 mi nutes, Figure 5-2, using MCNP5 photon, electron mode and pulse height tallie s (*F8) the results uncertainties (2 ) were 20% on average. A continue run was executed for the three phantoms for 400 minutes more, for 500 minutes total, reducing the Mont e Carlo uncertainty (2 ) down to 10% on average (Figure 5-3). However, the Monte Carlo computed difference in the depos ited dose between the two tissues is within 2%, making it impossible to observe a ny difference between the differe nt tissues in the phantoms, especially for therapeutic photon beams; this is shown in (Figure 5-4). The estimated time required to reduce Monte Carlo uncertainty to (2 ) 2% on-axis was 12500 minutes; hence, the time required to reduce the Monte Carlo uncertainty off-axis to 2% will be much greater.

PAGE 105

105 Therefore, this is an example that highlights that tissue heterogeneities can be difficult to discern in these problems using the Monte Carlo method. As explained previously, one of the most im portant issues limiti ng the use of the Monte Carlo techniques for clinical absorbed dose calculat ions is achieving a statistically reliable result in a reasonable time. This problem is even gr eater when a photon-electron mode is required, when CPE is not applicable, causing computer running times needed to reduce the statistical error to increase rapidly. Improved variance re duction techniques may reduce the computation times. However, such techniques are to be us ed with caution, as improper application of statistical variance reduction may cause wr ong answers. Moreover, variance reduction techniques may not reduce the time required to reduce the variance to acceptable levels, especially when accurate absorbed dose esti mations are needed over the entire problem. Figure 5-2. Monte Carlo derived total absorbed dose rate distributions al ong the X-axis computed using *F8 MCNP5 tally for three different phantoms. MCNP5 uncertainty (2 ) average (20.0%), 100 Minute run

PAGE 106

106 Figure 5-3. Monte Carlo derived total absorbed dose rate distributions al ong the X-axis computed using *F8 MCNP5 tally for thre e different phantoms. MCNP uncertainty (2 ) average (10.0%), 500 Minutes run Figure 5-4. Monte Carlo derived total absorbed dose rate distributions al ong the X-axis computed using *F8 MCNP5 tally for th e _SBLLBSS_ phantom. MCNP uncertainty (2 ) average (7.0%), 1000 Minutes run

PAGE 107

107 5.1.2 PENTRAN Model As shown in Figure 5-5, we simulated equiva lent three phantoms to those simulated with Monte Carlo. The heterogeneous models consiste d of three key components; an X-ray source (15 cm3) with flat weighted 8 MV source, divided into 16 energy groups, a soft tissue phantom (45 cm3), equivalent layers of bone and lung tissue identical to that used in the Monte Carlo simulations. Models were divided into 3 z-levels; each z-level was divided equally into 9 coarse meshes containing (15) fine mesh cells for each coarse mesh; a total of 91,125 fine mesh cells were used in the SN problems. Although lower quadrature set (S32) are sufficient to produce accurate SN results for similar calculations (Chapter 1), to assure minimum ray-effects on the SN calculations, these simulations were based on the use of an S62 angular Legendre-Chebychev quadrature with P3 scattering anisotropy using the CEPXS library. Figure 5-5. Three SN phantoms equivalent to the one us ed in the Monte Carlo simulation. Phantom SSSSSSS is a homogeneous soft-tissue phantom; phantoms SSBBSSS and SBLLBSS are heterogeneous phantoms consisting of combinations of soft and bone tissue, and soft, bone, and lung tissueequivalent materials, respectively

PAGE 108

108 5.1.3 EDK-SN Absorbed dose Comparison The SN calculation was performed based on a full photon transport calculation using deterministic techniques in the heterogeneous phantom. In principle, this eliminates any need for any correction for tissue heterogeneities which affect the different sc attering components in differing ways. Moreover, the EDK-SN calculation was performed for material specific absorbed doses. This allowed the use of the pre-computed electron absorbed dose kernels for the different densities present in the phantom model. Figure 5-6. EDK-SN heterogene ous SBLLBSS phantom total absorb ed dose rate distributions along the X-axis compared to MCNP *F8 tally. MCNP uncertainty (2 ) average (8.0%) on-axis. Average absolute relative difference (4.1%) For the phantoms used in this study, calculations were performed to compare the EDK-SN and Monte Carlo pulse heig ht tally results for the absorbed dose deposited along the model central axis. As shown in the Figures 5-6 and 5-7, the EDK-SN results are in very good agreement with the MCNP5 results, and differen ces between the two are within the statistical uncertainty of the Monte Carlo ca lculation. The Monte Carlo uncerta inties, on the central axis of

PAGE 109

109 the source, were 4% on average while the aver age absolute difference at the same cells was 3.7%. Figure 5-7. EDK-SN Ho mogenous phantom total absorbed dos e rate distributions along the Xaxis compared to MCNP *F8 tally. MCNP uncertainty (2 ) average (6%) on-axis. Average absolute relative difference (3.7%) We note that at the interface l ung region, Figure 5-6, the EDK-SN derived absorbed dose indicated a lower absorbed dose th an Monte Carlo. This can be related to the application of the lung electron kernel over the bone and soft tissue regions. Th is phenomenon was mitigated by applying a correction factor for voxe ls located at the interface of different material densities. 5.1.4 Correction Factor Approach The range of a charged partic le of a given type and ener gy in a given medium is the expectation value of the path-length that it follow s until it comes to rest On the other hand, the RCSDA represents the range in the continuous slowing down approximation (CSDA). Figure 5-8 gives the CSDA range RCSDA for electron in cortical bone, soft tissue and lung tissue for electron energies between 0.0 MeV to 17.5 MeV. Figure 5-9 depicts a comparison between the ratio of

PAGE 110

110 the CSDA ranges for cortical bone to soft tissue, lung to soft tissue, and cortical bone to tissue. One can notice that the behavior of electrons in the different tissues is quite similar for the same electron energy. This implies that there is a possibility of applying a simple correction factor within different tissues to account for the different densities in the phantom for each energy group. Unlike other methodologies, this is possibl e due to the dependence of the EDK-SN on a full photon physics transport calculation accomplished by the SN computation to provide the scattered photon information and a directional (current) component. Figure 5-8. Continuous slowing down approximation range (Rcsda ) for electrons in Cortical Bone, Soft Tissue and Lung as a function of energy

PAGE 111

111 Figure 5-9. Continuous slowi ng down approximation range (RCSDA) ratio comparison for electrons in Cortical B one, Soft Tissue and Lung as a function of energy The key to this approach is that photon transport is carried with the 3-DSN code PENTRAN. The use of such code provided us with the directional component of the photon transport in the homogenous medium. By having this information we were able to project the 3D electron kernel along the net cu rrent direction. The assumption of a homogenous medium in photon transport in a heterogene ous problem, often used in other methods, may lead to an underestimation of the scattered component of the photon beam, and misrepresentation of its directional component, especially between the different organs of the human body (Figure 5-9), leading to misrepresentation of the absorbed dose. 5.2 Computational Phantom of 15-Year Old Male A new set of stylized phantoms have been developed by the UF Advanced Laboratory for Radiation Dosimetry Studies (A LRADS). These phantoms represent a totally new class of whole-body patient hybrid phantom s so named as they retain th e non-uniform scalability of

PAGE 112

112 stylized phantoms, while mainta ining the anatomic realism of segmented voxel phantoms with respect to organ shape, depth, and inter-organ positioning to great accuracy. One study showed that these phantoms match (to within 1%) to that of a newborn of the ICRP Publication 89 reference newborn child. In this study, hybrid computational phantoms represen ting reference 15-year male body anatomy and anthropometry is used, shown in Figure 5-10. This phantom was matched, by the ALRADS group, to anthropometric data and re ference organ mass data provided by the ICRP, where it matched to within 1% overall51. Figure 5-10. Frontal views of 3-D rendering of the 50th percen tile UFH-NURBS 15-year male and female The availability of hybrid phantoms as a re placement for voxel-based phantoms provides the needed platform for constructing phantoms representing the unique internal and external

PAGE 113

113 anatomy of individual pediatri c patients. However, this process would be complicated tremendously if each individual organ of the phantom had to be adjusted to match patientspecific organ volumes as seen under CT imagin g, even when those images are available. The 15-year-old male was voxelized with hi gh resolution using 2 mm 2 mm 2 mm voxel size (302 voxel 139 voxel 836 voxel) as a case study. This geometry data was then collapsed into a 1 cm 1 cm 1 cm voxe l size (60 27 167 voxels), for total of 270540 voxel-model using GHOST-3-D. Subsequently, using PENMSH-XP, the collapsed model was cast into a 3-D spatial distributi on, and sub-divided into 10 coarse mesh z-levels containing 18 coarse meshes in x-y domain, and these contained a number of fine mesh cells (Figure 5-11). An 8 MV flat weighted, forward peaked volumetric (20 cm x 1 cm x 17 cm) X-ray source was placed above the chest area of the phantom. Figure 5-11. Simulati on methodology for EDK-SN computations using PENTRAN-MP code system. Left: UF Anthropmorphic phantom, middle: corresponda nt PENTRAN input, right: EDK-SN 3-D absorbed dose distribution after irradiation with a x-ray source

PAGE 114

114 Although the model has relatively small fine mesh sizes, using Legendre-Chebychev S32 quadrature, the DTW differencing scheme, and CEPXS cross-sections with anisotropy level P3, we were able to eliminate any unphysical oscillations that may result in the deterministic solution. Using 16 energy groups, the SN model was executed on 16 processors and completed in 2.0 hours. With the general purpose radiation transport code MCNP5, we simulated an equivalent model to the one simulated using PENTRAN. The collapsed 3-D binary fi les were converted to ASCII formatted matrix files and then incorporat ed into an MCNP5 lat tice defined input; this matrix was also generated by our down-sampling code GHOST-3-D. 5.2.1 EDK-SN Absorbed Dose Calculations Calculations were performed to calculate the SN results for the scalar fluxes for all energy groups. Using EDK-SN, 3-D absorbed dose distributions we re calculated to each voxel in the entire 15 year old male phantom model for a broad-beam external photon field. Figure 5-11 shows the stages of calculation in the PENT RANP-MP code system that resulted in 3-D Absorbed dose distributions for the entire phantom. Figure 5-12 shows absorbed dose contributions from selected en ergy groups, and the total absorbed dose resulting from adding all energy group-absorbed dose contribu tions, as prescribed by the EDK-SN procedure. Organ absorbed doses using EDK-SN, for a selected number of or gans, were calculated and compared to Monte Carlo predictions, and are pr esented in Table 5-1. The total execution time for the Monte Carlo was ~4 hrs (wall clock time) on 16 processors, while the total wall clock execution time for the EDK-SN was ~ 2hrs on the same number of processors. As anticipated, the absorbed doses for the organs loca ted directly in the x-ray field (r ight+ left lung) or partially in the x-ray field (liver) were in good agreement for both the EDK-SN and MCNP5 pulse height tally data.

PAGE 115

115 Total-Dose EG1 EG13 EG10 EG7 EG4 EG16 Figure 5-12. Absorbed dose contribution of eac h energy group is calculated via the EDK-SN procedure, then folded to calculate the to tal absorbed dose deli vered to the 15 year old male phantom from 8 MeV flat wei ghted source. The original phantom was collapsed into 270K voxels Table 5-1. Comparison of sel ected organ absorbed dose rate (MeV/g. s) calculated using MCNP5 pulse height tally with (photon, electron mode) and EDK-SN for the UF hybrid 15-year-old male phantoms for a ch est 8 MV X-ray flat weighted source

PAGE 116

116 On the other hand, Monte Carlo calculations for organs outside the x-ra y field (such as the prostate) showed significant statistical uncertainty (44%). This poor absorbed dose tally result is due to the fact that the Monte Carlo did not have enough histories for these results to converge to low statistical uncertainty in an acceptable time. Alternatively, as shown in Table 5-2,when running the Monte Carl o for extended running times (~40 hours on our parallel cl uster), the Monte Carlo calculated absorbed dose delivered to the prostate began to show convergence to the absorbed dose calculated using EDK-SN. This illustrates that the EDK-SN methodology, executed in 2 hours (SN execution time = 1.5 hr, EDKSN execution time = 0.5 hr), is capable of providi ng an accurate solution and a significant speed up for a full physics absorbed dose calculation in human phantoms for high energy external beams. Table 5-2. Prostate absorbed dose rate (MeV/g .s) calculated for the test problem using EDK-SN and MCNP5 using pulse height tally (*F8) with (photon, electron mode) at different MCNP5 Monte Carlo running times 5.2.2 Ray-Effects When considering a forward peaked volumet ric (10 cm x 1 cm x 17 cm) X-ray source placed on top of the left lung area of the phant om, we noted significant differences between the absorbed dose calculated by EDK-SN and the absorbed dose ca lculated by full physics Monte

PAGE 117

117 Carlo calculations to the right lung. This behavior can be explained by the ray-effects in the low density lung tissue (see Chapter 2). Since the X -ray source is highly directional, and the organ (right lung) is of a low density material, and little primary radiation and very low scattered radiation is streaming through the right lung, making conditions ideal for inducing ray effects in SN computations for the photon component. This is mitigated by using high order quadratures, and can reduce absolute differences; these were reduced to 4% (Table 5-2). However, the increase of the quadrature level costs additional running time, with an increase from 2 hours to 9 hours. This increase in quadrature level was necessary for large low density volu mes that experience low levels of scattered radiation. Another method, ordinate splitting, a llows for a biased quadrature set to focus on the important directions, although the technique was not used here. Table 5-2. Absorbed dose rate (MeV/g. s) calculated for the test problem using EDK-SN and MCNP5 using pulse height tallies (* F8) with (photon, electron mode) 5.2.3 MCNP5 Energy Deposited Tally (F6) and Pulse Height Tally (*F8) Results When comparing MCNP5 organ absorbed dos e results, for the 15 -year old phantom, calculated using energy deposited tally (F6) and pulse heig ht tally (*F8), we noted inconsistencies in the results of the two tallies (T able 5-3). This is in keeping with our earlier investigation of the two tallies earlier in Chapter 4. This in consistency in the F6 tally suggests that full physics Monte Carlo ca lculations are essential for accu rate absorbed dose estimation in high energy external beam absorbed dose calculations.

PAGE 118

118 Table 5-3. Absorbed dose rate (MeV/g. s) cal culated for the test pr oblem using MCNP5 using pulse height tally *F8 wi th (photon, electron mode) and MCNP5 energy deposited tallies (F6) Wang et al. conducted a similar study using th eir VIP Man organ MCNP for absorbed dose comparison for the three different tally methods including cell flux tally (F4), energy deposited tally (F6) and pulse heig ht tally (*F8). In thei r comparison they found that the F4 tallies agree within 10% with the F6 tallies, while the differenc es between the tallies F8 and F6 are relatively large (28%) 14. These results agree with our findings, a nd provide a strong proof for the need of applying full physics transport when cal culating whole body absorbed dose. 5.3 Discussion and Findings To properly treat 3-D electron transport physic s deterministically, yet still to achieve reasonably fast and accurate whole body computa tion times using high energy photons, we used angular and energy dependent electron transport absorbed dos e kernels. In a methodology called EDK-SN we have demonstrated that by using different electron kernels for different materials (cortical bone, soft tissue and lung), the EDK-SN methodology incorporating a discrete ordinates (SN) approach is an effective alternativ e to the Monte Carlo method for accurately estimating the absorbed dose for high energy photon beams over an entire human phantom. In doing so, we simulated three phantoms using MCNP5 and PENTRAN-MP. Each phantom had a flat weighted 8 MV x-ray s ource impinging on 45x45x45 cc soft tissue phantom. Two of these phantoms were slabs of 1-cm-thick discs representative of bone, or lung tissue. Comparisons of MCNP5 Monte Carlo based dose results a nd those computed using PENTRAN-MP implementing the EDK-SN methodology showed good agreement between the two approaches.

PAGE 119

119 However, the Monte Carlo simulation took 100 minutes for the MC (2 ) uncertainty to be 20% (on average) for tally points along the central axis of the phantom using an MCNP5 *F8 tally (we note here that 20% uncertain ty cant be viewed as converged). Additional runs totaling 500 minutes were made, reducing the Monte Carlo (2 ) uncertainty to 10%, on average. We estimated the approximate time required to reduce Monte Carol (2 ) uncertainty to 2% for the on-axis locations (relative to the source X-ray beam) to be 12,500 minutes. Undoubtedly, the time required to reduce the Monte Carlo uncertainties at locations off-axis from the source beam to 2% will be much greater. Based on the UF hybrid phantoms developed by the University of Florida ALRADS group, we simulated a complete phantom discretized using 270,540 voxels, and demonstrated that; in this case, the EDK-SN method readily provides accurate whole body 3-D absorbed dose distributions. We were able to accurately yield doses analogous to the MCNP5 Monte Carlo calculations. We further noted that ray-effects were evident in regions of the tissue-lung interface, which indicate higher angular quadratures or ordinat e splitting treatments are necessary in lung tissue and low density tissue inte rfaces. Also, tally results from the 16 energy groups, used to represent the radiation spectru m from the CEPXS multi -group cross-section library were compared to continuous energy MCNP5 Monte Carlo results using the ENDF/B-VI data libraries. A major advantage of the EDK-SN method is that it provid es a detailed, accurate 3-D absorbed dose distribution at thousands of ce lls throughout the system, while the Monte Carlo method only provides highly accurate values for sele cted points near the source. Though the use of volumetric mesh tallies, Monte Carlo calcula tions can provide global absorbed doses across a grid spanning the Monte Carlo geometry. However, at present, absorbed doses for this data are

PAGE 120

120 based on assuming charged particle equilibrium conditions, not suitable for high energy beams. Moreover, the statistical uncertainty associated w ith each mesh tally site for such computations would require extended periods of running time, es pecially off-axis from the source and outside the region covered by the primary beam to converg e to acceptable levels. These factors limit the Monte Carlo calculations from bei ng fuuly adequate in practical applications where whole body and organ-specific absorbed dose information is needed. Notably, an equivalent Monte Carlo calculation to obtain the flux distri bution with statistically reliable results over the entire problem space may require orders of magnitude longer running times compared to parallel deterministic calculations.

PAGE 121

121 CHAPTER 6 SUMMARY, CONCLUSIONS AND FUTURE WORK This work began with the prim ary goal of developing a new parallel 3-D Dose calculation tool based on deterministically rendered high energy photon transpor t solutions in radiotherapy settings. 6.1 Conclusions and Findings In this dissertation, the EDK-SN method is presented for generating detailed absorbed dose information using high energy external beam so urce terms. Because detailed whole body and organ-specific absorbed doses can be accurate ly obtained throughout a voxelized phantom with proper electron interaction physics projected onto the phantom voxels, based on SN-computed photon fluxes and net current vectors, this highlights the strength of the EDK-SN transport based methodology. Overall, based on the problems ex amined, with an adequate discretization implemented using the EDK-SN method, it is possible to produce absorbed dose results consistent within the stat istical uncertainty of MCNP 5 Monte Carlo calculations. Overall, our studies examined the capabilitie s of the discrete ordi nates method compared with the Monte Carlo method with application to medical physic s problems. Since radiation transporting in such applications encounters few interacti ons in low density mate rial, such as air, attention was given to minimum quadrature levels and mesh size distributions necessary for solution accuracy in air and in a water phantom. Also, it is eviden t that multi-gr oup cross-section libraries must also be carefully tested for probl em applicability. Of particular importance was the flexibility of the adaptive SN differencing scheme in PENTRAN, permitting more accurate spatial differencing schemes in cells with a range of optical thicknesses. In addition, deterministic models based on the UF series B pediatric phantoms and with the PENTRAN-MP

PAGE 122

122 methodology, we were able to accurately yiel d a solution analogous to the MCNP5 Monte Carlo calculations using the CPE principle for low energy beams. We have demonstrated; that by using electron kernels for th e different materials (cortical bone, soft tissue and lung) the EDK-SN methodology incorporating a discrete ordinates (SN) approach is an effective alte rnative to the Monte Carlo met hod for accurately estimating the absorbed dose for high energy photon beams over an entire human phantom. In doing so, we simulated three phantoms using MCNP5 and PENT RAN. Each phantom ha d a flat weighted 8 MV x-ray source impinging on a 45x45x45 cc soft ti ssue phantom. Two of these phantoms have slabs of 1-cm-thick discs of materials represen tative of bone, or lung ti ssue. The comparison of the MCNP5 results and the EDK-SN showed good agreement between the two methodologies. We also simulated the complete human pha ntom using 270,540 voxels, and demonstrated that the EDK-SN method readily provides accurate whol e body 3-D absorbed dose distributions. The angular data provided by the SN method was used to guide the absorbed dose mapping process for secondary electrons to the surrounding voxels ; this makes our approach truly novel. We were able to accurately yield a solution analogous to the MCNP5 Monte Carlo calculations in the voxelized phantom. We note that that tally results from the 16 energy groups, used to represent the radiati on spectrum, from the CEPXS multi-g roup cross-section library, were compared to continuous energy MCNP5 Monte Carlo results using the ENDF/B-VI data libraries. Parallel deterministic PENTRAN-MP results were obtained within comparable running times to parallel Monte Carlo calculations for ta lly sites adjacent to the source. An equivalent Monte Carlo calculation to obtain the absorbed dose distributions using photon, electron mode with pulse height tallies converg ed to reasonable sta tistical uncertainty ove r the entire problem

PAGE 123

123 space may require orders of magnitude longer running times compared to the time needed for complete, whole body parallel deterministic calculat ions. The work presente d in this dissertation has extended the capabilities of the discrete ordinates method to accurately calculate radiation absorbed dose contributions in medical phys ics procedures using the PENTRAN-MP code system with the EDK-SN methodology. 6.2 Future Work The new EDK-SN, integrated into the PENTRAN-MP Code System, represents a leap forward in computational medical physics; large 3-D radiation transport calculations for high energy photons and absorbed dose contributions can now be performed accurately in a fraction of the computation time requi red in the Monte Carlo methodol ogy. The methods described in this dissertation can be further enhanced and developed by studying the following issues: Additional parallel performance assessments. Examine performance for solving additional real-world problems. Optimization study on the EDK-SN. This should be undertaken, in order to identify the necessary level of accuracy wh ich yields the best performance in terms of speed-up. Test alternative method for applying corre ction factors in tissu e inhomogeneities. Conduct timing and performance studies with the new hybrid PENTRAN code implementing Discrete Ordinates and Me thod of Characteristics. Apply for an NIH R01 grant. Long-term plans are to implement PENTRA N-MP as a secondary treatment planning software for documentation of both in-field an d out-of-field organ absorbed doses to normal tissues. This information will be extremely valuab le to radiation epidemiology studies which are currently being conducted using far more simplisti c dosimetry schemes, particularly to out of fields organs. This data will also be of use to future optimization studies where present goals of tumor absorbed dose conformality and near-field ab sorbed dose avoidance to normal tissues will be supplemented by the additiona l goal of secondary cancer risk minimization in out-of-field tissues and organs.

PAGE 124

124 Att. Coefficient WATER (cm2/g) 1.00E-08 1.00E-06 1.00E-04 1.00E-02 1.00E+00 1.00E+02 1.00E+04 1.00E-031.00E-021.00E-011.00E+001.00E+011.00E+02 Energy(MeV)Att. Coefficient(cm2/g) Scatt. Coh.(cm2/g) Scatt. Incoh.(cm2/g) P.E.A(cm2/g) P.P Nuc. Field (cm2/g) P.P Elec. Field (cm2/g) Total w/ coh Total w/o Coh. APPENDIX A HIGH ENERGY PHOTON TREATMENT (IN CONDENSED ELECTRON TRANSPORT) The photons energy used in radiotherapy tr eatm ent planning are 20 MeV or less. Photon interactions that take place at these high energies predomin antly Compton Effect (C.E.) or/and Pair Production interactions by different percenta ges, as shown in Figure A-1 and Table A-1. Figure A-1. Total mass attenuation coefficient for photons in wate r, indicating the contributions associated with phoelectric absorption, Compton scattering and electron-positron pair production Table A-1. Ratios of photon interacti ons in water for range of energies Photon Energy (MeV) Photon Interaction 2.0 3.0 6.0 10.0 14.0 18.0 20.0 % of Incoherent scattering 99.21 % 97.15 % 88.59 % 77.06 % 67.32 % 59.44 % 56.04 % % of Nuclear P.P 0.79 % 2.82 % 10.78 % 21.18 % 29.76 % 36.69 % 39.64 % % of electron P.P 0.0 % 0.03 % 0.61 % 1.76 % 2.9 % 3.90 % 4.35 %

PAGE 125

125 Compton Scattering: When passing through matter, photons can interact with the electrons of the matter. The Compton scattering describes an inelastic scattering of a photon w ith a free electron. If the electrons initial bindi ng energy is small compared to the photon energy (<< h ) then the stuck electron can be considered initially free and unbound. Figure 10 shows this inelastic scattering process. The kinematics of this sc attering (conservation of momentum and energy) lead to the energy of the scattered photon 'h and to the energy of the scattered electron Figure A-2. Sketch of Compt on scattering, photon of energy h collide with an electron. The photons incident forward momentum is hc where c is the speed of light in vacuum Based on the unbound and stationary condition of the electron, the Compton relation can be obtained by solving energy conservation equations 'Thh (A-1) And conservation of momentum

PAGE 126

126 'coscosehh p cc (A-2) 'sinsinehpc (A-3) p c can be written in terms of T by invoking the law of invariance: 2 0(2) p cTTmc (A-4) In which 0m is the electrons rest mass. As a result of solving these equations algebraically, one can derive the following Comptons relations 2 01()(1cos) h h hmc (A-5) 2 0cot(1)tan 2ehmc (A-6) 21 1 12sin/2eThhh (A-7) To obtain the fraction of the scattered photons in a given direction, Klein and Nishina (reference) have carried out a quantum-mechanical treatment of the problem using the Dirac equation for the electron and have obtained the equation 2 2 '' 2 0 'sin 2KNdr hhh dhh h (A-8) Equations similar to those for the Compt on photon distribution can be obtained for the Compton electron distribution. Sinc e the probability that an electron will be scattered into the solid angle 'd situated in the direction e is the same as the probability that a primary quantum

PAGE 127

127 will be scattered into the solid angle d in the direction and being related by Eq. (A8), we have that KNKN eedd d ddd (A-9) sin sineeeeddd ddd (A-10) cot(1)tan() 2e (A-11) 2211 (1) sincos2ed (A-12) 1/2 22sin1(1)tan2e (A-13) 2e (A-14) 1e ed dd d (A-15) 31(1cos)sin (1) sine ed d (A-16) for 0.511 MeV photons,1 3/2 22sin(/2)13sin(/2)ed d (A-17) 2 2 '' 3/2 22 0 '2sin(/2)13sin(/2) sin 2KN edr hhh dhh h (A-18)

PAGE 128

128 Figure A-3. Compton effects differential cross section per unite angle for the number of electrons scattered in the direction e Example Considering a Compton scattering of 1.022 MeV (2 ), the ratio of recoil electrons with energies above 0.511 MeV to recoil elect ron with energies below 0.511 will be: '(1) 0.5 1(1) Thh hh (A-19) 0.560o (A-20) cot(1)tan 2e (A-21) ()30o e 180 60 60 02sin 2sinKN KNd d d R d d d (A-22)

PAGE 129

129 2 22 222 0(22)/(1)(12) 2KNdr d (A-23) 1 1(1) (A-24) cossin dd (A-25) dd (A-26) 'h h (A-2 7) From this example one can note that; the prod uced electron in a cer tain direction has a unique energy for a giving initial photon. So for an angle aroundiscretized range of angles we can calculate the average energy of the electrons produced in th Pair production Figure illustrates schematically a pair production event in the nucleus electric field. The incident photon gives up all of it s energy in the creation of an el ectron positron pair with kinetic energies TandT. The energy conservation equation, ignoring the vanishing small kinetic energy giving to the nucleus is simply 2 02 1.022 hmcTT M eVTT (A-28) The average electron/positron average energy is 1.022 2 hMeV T (A-29)

PAGE 130

130 Figure A-4. Pair Production in the Coulomb force field of the nucleus For h values well above th e threshold energy 22 0mc, the electron and positron are strongly forward directed. Their average angle of departure relative to the original photon direction is roughly: 2 0mc T (radians) (A-30) Pair production in the electron field: In the kinematics of Pair Production in the electron filed (triple production), the photon divides its energy between the positron-electron pair produced and the host electron. The energy conservation equation becomes: 2 012 122 1.022 hmcTTT M eVTTT (A-31) The average kinetic energy of the three particles is 1.022 3 hMeV T (A-32)

PAGE 131

131 In triple production, the threshold for this process is 2 04 mc=2.044 MeV. The higher threshold is required by conservation of momentum as shown in Table A-1 and also derived by Perrin (1933). Triple production becomes signifi cant in radiotherapy treatment planning for energies greater than 14 MeV were it repres ents ~3% of the total photon interaction. Straggling and Multiple Scattering There is typically a distribution of angle and energy for charged particle passing the same medium. This is a result of multip le scattering and angular straggling which the spread in angular distribution observed in a population of initial identical ch arged particle after they have traverse equal path lengths. It will be some what exaggerated if the particle have passed through a layer of material, since multiple scattering then cause indi vidual differences in scattering angle as well causing them to propagate in a conical angular distribution in accordance with Molieres theory. However, for a discretized solid angle, one can consider an average angle for these scattered electrons on which approximate transport calculations can be made. The spread of the solid angle can be giving by Fermi angle for stra ggling especially when considering restricted stopping power including soft collisions, that yield small spread in the angle, and hard collisions resulting in rays with energies less than the cutoff value Restricted stopping power (/) dTdx Is the fraction of the collisions stopping power that include the soft collisions plus those hard collisions resulting in rays with energies less than the cutoff value 2 2 0(2) 2 ln (,) 2 dT C kG dx Z Imc (A-33)

PAGE 132

132 Radiative losses: (/) (/)820rad ionizingdTdx Z T dTdx (A-34) Angular straggling (Fermi 1950) Electrons suffer multiple scattering events preclude straight transport in the material. Further complication are caused by the statistical fluctuation Figure A-5. Pair Production in the Coulomb force field of the nucleus 22 22 0 24322() ln 2 Ta zee NZ Z e T (A-35) 2 0 2 0h a mc (A-36)

PAGE 133

133 APPENDIX B MCNP5 PHNTOM INPUT DECK c UF_15yr voxel MCNP model whole body c Matrix size [60,27,167] C Voxel resolution=1.0 1.0 1.0 c 2008-7 c Ahmad Al-Basheer c The University of Florida C ************************************************************************** c c cells Card c C ************************************************************************** c ---------------------------c Body composition and density c ---------------------------1 1 -0.001205 -999 imp:p,e=1 u=1 vol=78406.867 $ Air 2 2 -1.03 -999 imp:p,e=1 u=2 vol=3.94E+04 $Residual Soft Tissue 3 3 -1.03 -999 imp:p,e=1 u=3 vol=2.00E+00 $Adrenal (L) 4 4 -1.03 -999 imp:p,e=1 u=4 vol=3.00E+00 $Adrenal (R) 5 5 -1.04 -999 imp:p,e=1 u=5 vol=1.39E+03 $Brain 6 6 -0.94 -999 imp:p,e=1 u=6 vol=1.70E+01 $Breast 7 7 -1.07 -999 imp:p,e=1 u=7 vol=4.00E+00 $Bronchi 8 8 -1.03 -999 imp:p,e=1 u=8 vol=9.70E+01 $Right Colon W 9 9 -0.83 -999 imp:p,e=1 u=9 vol=1.45E+02 $Right Colon C 10 10 -1.10 -999 imp:p,e=1 u=10 vol=3.00E+00 $Ears 11 11 -1.03 -999 imp:p,e=1 u=11 vol=2.40E+01 $Esophagus 12 12 -1.05 -999 imp:p,e=1 u=12 vol=2.00E+00 $External nose 13 13 -1.07 -999 imp:p,e=1 u=13 vol=1.10E+01 $Eye balls 14 14 -1.03 -999 imp:p,e=1 u=14 vol=0.00E+00 $Gall Bladder W 15 15 -1.03 -999 imp:p,e=1 u=15 vol=4.50E+01 $Gall Bladder C 16 16 -1.04 -999 imp:p,e=1 u=16 vol=1.86E+02 $Heart W 17 17 -1.06 -999 imp:p,e=1 u=17 vol=4.23E+02 $Heart C 18 18 -1.05 -999 imp:p,e=1 u=18 vol=8.00E+01 $Kidney-cortex (L) 19 19 -1.05 -999 imp:p,e=1 u=19 vol=8.40E+01 $Kidney-cortex (R) 20 20 -1.04 -999 imp:p,e=1 u=20 vol=3.10E+01 $Kidney-medulla (L) 21 21 -1.04 -999 imp:p,e=1 u=21 vol=3.20E+01 $Kidney-medulla (R) 22 22 -1.04 -999 imp:p,e=1 u=22 vol=5.00E+00 $Kidney-pelvis (L) 23 23 -1.04 -999 imp:p,e=1 u=23 vol=6.00E+00 $Kidney-pelvis (R) 24 24 -1.07 -999 imp:p,e=1 u=24 vol=2.20E+01 $Larynx 25 25 -1.07 -999 imp:p,e=1 u=25 vol=0.00E+00 $Lens 26 26 -1.06 -999 imp:p,e=1 u=26 vol=1.23E+03 $Liver 27 27 -0.240 -999 imp:p,e=1 u=27 vol=1.72E+03 $Lung (L) 28 28 -0.240 -999 imp:p,e=1 u=28 vol=2.04E+03 $Lung (R) 29 29 -1.02 -999 imp:p,e=1 u=29 vol=0.00E+00 $Nasal layer (anterior) 30 30 -1.02 -999 imp:p,e=1 u=30 vol=0.00E+00 $Nasal layer (posterior) 31 31 -1.02 -999 imp:p,e=1 u=31 vol=0.00E+00 $Oral cavity layer 32 32 -1.02 -999 imp:p,e=1 u=32 vol=1.08E+02 $Pancreas 33 33 -1.05 -999 imp:p,e=1 u=33 vol=2.30E+01 $Penis 34 34 -1.03 -999 imp:p,e=1 u=34 vol=2.00E+00 $Pharynx 35 35 -1.03 -999 imp:p,e=1 u=35 vol=0.00E+00 $Pituitary Gland 36 36 -1.03 -999 imp:p,e=1 u=36 vol=4.00E+00 $Prostate 37 37 -1.03 -999 imp:p,e=1 u=37 vol=0.00E+00 $Rectosigmoid W 38 38 -0.49 -999 imp:p,e=1 u=38 vol=1.39E+02 $Rectosigmoid C

PAGE 134

134 39 39 -1.03 -999 imp:p,e=1 u=39 vol=3.40E+01 $Salivary Glands (parotid) 40 40 -1.03 -999 imp:p,e=1 u=40 vol=1.00E+01 $Scrotum 41 41 -1.03 -999 imp:p,e=1 u=41 vol=1.40E+02 $SI W 42 42 -0.41 -999 imp:p,e=1 u=42 vol=7.70E+02 $SI C 43 43 -1.09 -999 imp:p,e=1 u=43 vol=1.72E+02 $Skin 44 44 -1.04 -999 imp:p,e=1 u=44 vol=4.10E+01 $Spinal Cord 45 45 -1.06 -999 imp:p,e=1 u=45 vol=1.21E+02 $Spleen 46 46 -1.03 -999 imp:p,e=1 u=46 vol=3.30E+01 $Stomach W 47 47 -1.02 -999 imp:p,e=1 u=47 vol=2.08E+02 $Stomach C 48 48 -1.04 -999 imp:p,e=1 u=48 vol=1.80E+01 $Testes 49 49 -1.03 -999 imp:p,e=1 u=49 vol=3.20E+01 $Thymus 50 50 -1.05 -999 imp:p,e=1 u=50 vol=8.00E+00 $Thyroid 51 51 -1.05 -999 imp:p,e=1 u=51 vol=5.50E+01 $Tongue 52 52 -1.02 -999 imp:p,e=1 u=52 vol=2.00E+00 $Tonsil 53 53 -1.03 -999 imp:p,e=1 u=53 vol=6.00E+00 $Trachea 54 54 -1.04 -999 imp:p,e=1 u=54 vol=3.00E+00 $Urinary bladder W 55 55 -1.00 -999 imp:p,e=1 u=55 vol=1.61E+02 $Urinary bladder C 56 56 -0.001 -999 imp:p,e=1 u=56 vol=4.23E+02 $Air (in body) 57 57 -1.03 -999 imp:p,e=1 u=57 vol=7.00E+01 $Left Colon W 58 58 -0.34 -999 imp:p,e=1 u=58 vol=1.95E+02 $Left Colon C 59 59 -1.03 -999 imp:p,e=1 u=59 vol=2.00E+01 $Salivary Glands (submaxillary) 60 60 -1.03 -999 imp:p,e=1 u=60 vol=7.00E+00 $Salivary Glands (sublingual) 61 100 -1.10 -999 imp:p,e=1 u=61 vol=2.20E+01 $Coastal cartilage of ribs 62 100 -1.10 -999 imp:p,e=1 u=62 vol=0.00E+00 $Cervical Discs 63 100 -1.10 -999 imp:p,e=1 u=63 vol=2.80E+01 $Thoracic Discs 64 100 -1.10 -999 imp:p,e=1 u=64 vol=1.20E+01 $Lumbar Discs 65 101 -1.519 -999 imp:p,e=1 u=65 vol=7.48E+02 $Cranium 66 101 -1.519 -999 imp:p,e=1 u=66 vol=3.30E+01 $Mandible 67 104 -1.387 -999 imp:p,e=1 u=67 vol=1.06E+02 $Scapulae 68 103 -1.351 -999 imp:p,e=1 u=68 vol=3.00E+01 $Clavicles 69 108 -1.219 -999 imp:p,e=1 u=69 vol=4.40E+01 $Sternum 70 103 -1.351 -999 imp:p,e=1 u=70 vol=2.90E+01 $Ribs 71 102 -1.196 -999 imp:p,e=1 u=71 vol=5.10E+01 $Vertebrae-C 72 102 -1.196 -999 imp:p,e=1 u=72 vol=2.28E+02 $Vertebrae-T 73 102 -1.196 -999 imp:p,e=1 u=73 vol=1.95E+02 $Vertebrae-L 74 108 -1.219 -999 imp:p,e=1 u=74 vol=1.29E+02 $Sacrum 75 103 -1.351 -999 imp:p,e=1 u=75 vol=4.67E+02 $Os Coxae 76 103 -1.319 -999 imp:p,e=1 u=76 vol=2.20E+02 $Femur-proximal 77 103 -1.319 -999 imp:p,e=1 u=77 vol=1.63E+02 $Femur-upper shaft 78 104 -1.387 -999 imp:p,e=1 u=78 vol=1.33E+02 $Femur-lower shaft 79 103 -1.319 -999 imp:p,e=1 u=79 vol=2.59E+02 $Femur-distal 80 106 -1.352 -999 imp:p,e=1 u=80 vol=2.07E+02 $Tibiae-proximal 81 106 -1.352 -999 imp:p,e=1 u=81 vol=1.01E+02 $Tibiae-upper shaft 82 106 -1.352 -999 imp:p,e=1 u=82 vol=1.01E+02 $Tibiae-lower shaft 83 106 -1.352 -999 imp:p,e=1 u=83 vol=7.30E+01 $Tibiae-distal 84 107 -1.273 -999 imp:p,e=1 u=84 vol=1.40E+01 $Fibulae-proximal 85 107 -1.273 -999 imp:p,e=1 u=85 vol=4.00E+00 $Fibulae-upper shaft 86 107 -1.273 -999 imp:p,e=1 u=86 vol=1.40E+01 $Fibulae-lower shaft 87 107 -1.273 -999 imp:p,e=1 u=87 vol=1.70E+01 $Fibulae-distal 88 106 -1.35 -999 imp:p,e=1 u=88 vol=2.40E+01 $Patellae 89 105 -1.415 -999 imp:p,e=1 u=89 vol=4.30E+02 $Ankle+Feet 90 103 -1.319 -999 imp:p,e=1 u=90 vol=1.50E+02 $Humerus-proximal 91 103 -1.319 -999 imp:p,e=1 u=91 vol=7.40E+01 $Humerus-upper shaft 92 104 -1.378 -999 imp:p,e=1 u=92 vol=6.70E+01 $Humerus-lower shaft 93 104 -1.378 -999 imp:p,e=1 u=93 vol=9.30E+01 $Humerus-distal 94 106 -1.374 -999 imp:p,e=1 u=94 vol=1.60E+01 $Radii-proximal

PAGE 135

135 95 106 -1.374 -999 imp:p,e=1 u=95 vol=1.60E+01 $Radii-upper shaft 96 106 -1.374 -999 imp:p,e=1 u=96 vol=1.70E+01 $Radii-lower shaft 97 106 -1.374 -999 imp:p,e=1 u=97 vol=2.70E+01 $Radii-distal 98 106 -1.374 -999 imp:p,e=1 u=98 vol=4.90E+01 $Ulnae-proximal 99 106 -1.374 -999 imp:p,e=1 u=99 vol=2.70E+01 $Ulnae-upper shaft 100 106 -1.374 -999 imp:p,e=1 u=100 vol=1.80E+01 $Ulnae-lower shaft 101 106 -1.374 -999 imp:p,e=1 u=101 vol=7.00E+00 $Ulnae-distal 102 105 -1.415 -999 imp:p,e=1 u=102 vol=8.60E+01 $Hand 103 105 -3.00 -999 imp:p,e=1 u=103 vol=1.30E+01 $Teeth 301 1 -0.001205 -999 imp:p,e=1 u=301 vol= 78406.867 $ Air 302 28 -0.2995 -999 imp:p,e=1 u=302 vol= 10210.44 $ Lung 303 2 -1.00 -999 imp:p,e=1 u=303 vol= 27482.027 $ Tissue 304 101 -1.32 -999 imp:p,e=1 u=304 vol= 2304.37 $ Bone C ******************* Lattice definition ************************************ 1001 0 -100 fill=10000 imp:p,e=1 $ surrounding box 555 0 -200 lat=1 u=10000 imp:p,e=1 fill = 0:59 0:26 0:166 C **************** Image data start from here ******************************* Phantom Input from Ghoast-3-D C ********************************************************************************* c c Surface Cards c C ************************************************************************** 100 rpp 0 60 0 27 0 167 $(-29:30 -13:13 -83:83) 200 rpp 0 1.0 0 1.0 0 1.0 999 rpp -25 85 -25 52 -25 192 c ************************************************************************************** Material Cards C*************************************************************************************** mode p e sdef x=d1 y=d2 z=d3 erg=d4 AXS= 0 1 0 dir= 0.99763 vec= 0 1 0 si1 H 30 40 sp1 D 0 1 si2 H 0.0 1.0 sp2 D 0 1 si3 H 119 136 sp3 D 0 1 si4 H 0.01 8.0 sp4 D 0.0 1.0 c -------------------------------------------------------------------c tallies cards c --------------------------------------------------------------------c ---------------------------------------------------phys:p 4j 1 ctme 500 totnu print

PAGE 136

136 LIST OF REFERENCES 1R. Mohan, in XIIth International Confer ence on the Use of Computers in Radiation Therapy edited by (Medical Physics Publishing, Madiso n, WI, 1995, Salt Lack City, 1997), pp. 16-18.35 2T. R. Mackie, P. Reckwerdt, and N. Papani kolaou, "3-D Radiation Treatment Planning and Conformal Therapy," Medical Physics Publishing Corporation Madison WI, 1995), pp. 34 3B. A. Fraass, J. Smathers, and J. Deye, "Su mmary and recommendations of a National Cancer Institute workshop on issues limiting the clinical use of Monte Carlo dose calculation algorithms for megavoltage external beam radiation therapy," Med Phys 30, 3206-3216 (2003).29 4F. M. Khan, The Physics of Radiation Therapy, 3 (Lippincott Williams & Wilkins, 2003).36 5P. R. Almond, A. E. Wright, and M. L. Boone "High-energy electron dose perturbations in regions of tissue heterogeneity. II. Physical models of tissue heterogeneities," Radiology 88, 1146-1153 (1967).37 6A. Ahnesjo, M. Saxner, and A. Trepp, "A pencil beam model for photon dose calculation," Med Phys 19, 263-273 (1992).73 7A. L. EDWARDS, J. A. RATHKOPF, and R. K. SMIDT, "Extending the Alias Monte Carlo sampling method to general distributions,," in the International Conference on Mathematics and Computations and Reactor Physics edited by (American Nuclear Society, Inc., La Grange Park, IL, 1991), pp. 38 8A. Ahnesjo, "Collapsed cone convolution of radiant energy for phot on dose calculation in heterogeneous media," Med Phys 16, 577-592 (1989).52 9T. Knoos, A. Ahnesjo, P. Nilsson, and L. Weber, "Limitations of a pencil beam approach to photon dose calculations in l ung tissue," Phys Med Biol 40, 1411-1420 (1995).50 10B. Dionne, "Automated variance reduction tech nique for 3-D Monte Carlo coupled electronphoton-positron simulation using determ inistic importance functions," Nuclear Engineering Sciences (University of Florida, Gainesville, Fl, 2007), pp. 75 11E. E. Lewis, and W. F. Miller, Computational Methods of Neutron Transport ( American Nuclear Society, La Grange Park, Ill., USA, 1993).18 12L. J. Shukovsky, "Dose, time, volume relations hips in squamous cell carcinoma of the supraglottic larynx," Am J Roen tgenol Radium Ther Nucl Med 108, 27-29 (1970).31 13J. J. Fischer, and J. E. Moulde r, "The steepness of the dose-res ponse curve in radiation therapy. Theoretical considerations and e xperimental results," Radiology 117, 179-184 (1975).30

PAGE 137

137 14B. Wang, G. Xu, J. T. Goorley, and A. Bozkurt, "ISSUES RELA TED TO THE USE OF MCNP CODE FOR AN EXTREMELY LARGE VOXEL MODEL VIP-MAN," in The Monte Carlo Method: Versatility Unbounded In A Dynamic Computing World, edited by (American Nuclear Society, Chattanooga, Tennessee, 2005), pp. 32 15G. I. Bell, and S. Glasstone, Nuclear Reactor Theory, (Robert E. Krieger Publishing CO. Inc, Malabar, FL, USA, 1985).39 16B. G. Carlson, and K. D. Lathrop, Discrete Ordinates Angular Quadrature of the Neutron Transport Equation (Los Alamos Scientific Laboratory Report, LA-3186, 1965).40 17D. Shedlock, and A. Haghighat, Neutron Analysis of Spent Fuel Storage Installation Using Parallel Computing and Advance Discrete Ordinates and Monte Ca rlo Techniques," Radiat Prot Dosimetry 116 662-666 (2005).45 18K. D. Lathrop, "Remedies for Ray Effects," Nuclear Science and Engineering 45, 255-268 (1971).41 19G. Sjoden, and A. Haghighat, "PENTRAN A 3D Cartesian Parallel SN Code with Angular, Energy, and Spatial Decomposition," in Intl. Conf. on Math. Meth & Supercomp. for Nuclear App., edited by Saratoga Springs, New York, 1997), pp. 553.14 20A. Al-Basheer, M. Ghita, G. Sjoden, W. Bolch, and C. Lee, "Whole Body Dosimetry Simulations using the PENTRAN-MP Sn Code System," in American Nuclear Socity 2007 Annual Meeting edited by Bosten, MA, 2007), pp. 66 21G. M. Daskalov, R. S. Baker, R. C. Little, D. W. O. Rogers, and J. F. Williamson, "TwoDimensional Discrete Ordinates Photon Trans port Calculations for Brachytherapy Dosimetry Applications," Nucl. Sci. Eng. 134, 121-134 (2000).19 22G. M. Daskalov, R. S. Baker, D. W. Rogers and J. F. Williamson, "Multigroup discrete ordinates modeling of 125I 6702 seed dose distributions using a broa d energy-group cross section representation," Med Phys 29, 113-124 (2002).20 23M. L. Williams, D. Ilas, E. Sajo, D. B. Jones, and K. E. Watkins, "Deterministic photon transport calculations in gene ral geometry for external beam radiation therapy," Med Phys 30, 3183-3195 (2003).5 24A. Haghighat, International Training Course/Workshop on Methodologies for Particle Transport Simulation of Nuclear Systems 76 25(ICRU) 1980, Radiation quantities and units (International Commission on Radiation Units and Measurements, Bethesda, MD, 1980).42 26F. H. Attix, "The partition of kerma to account for bremsstrahlung," Health Phys 36, 347-354 (1979).43

PAGE 138

138 27F. H. Attix, "Energy imparted, energy transferred and net energy transferred," Phys Med Biol 28, 1385-1390 (1983).44 28N. I. o. S. a. T. NIST, ESTAR(Stopping Power and Range Tables for Electrons ) http://physics.nist.gov/PhysRe fData/Star/Text/E STAR.html.71 29T. R. Mackie, A. F. Bielajew, D. W. Rogers, and J. J. Battista, "Generation of photon energy deposition kernels using the EGS Monte Carlo code," Phys Med Biol 33, 1-20 (1988).54 30A. Ahnesjo, P. Andreo, and A. Brahme, "Calculation and application of point spread functions for treatment planning with high energy photon beams," Acta Oncol 26, 49-56 (1987).53 31L. A. N. L. X-5 Monte Carlo Team, MCNP5A General Monte Carlo N-Particle transport code, Version 5 10 32J. E. White, J. D. Intgersoll, C. Slater, and R. Roussin, BUGLE-96 : A Revised Multi-group Cross Section Library for LWR Applications Based on ENDF/B-VI 46 33L. J. Lorence, J. E. Morel, and G. D. Valdez, Physics Guide to CEPXS: A multigroup coupled electron-photon cross section generating code 11 34G. SJODEN, and A. HAGHIGHAT, PENTRAN: P arallel E nvironment N eutral-Particle TRAN sport Sn in 3-D Cartesian Geometry 33 35C. Yi, and A. Haghighat, PENMSH Express 62 36G. Longoni, "Advanced quadrature sets, acceleration and preconditioning techniques for the discrete ordinates method in parallel computing environments ," (University of Florida, Gainesville, FL, 2004), pp. 47 37B. Petrovic, and A. Haghighat, "New directi onal theta-weighted (DTW) differencing scheme and reduction of estimated pressure vessel fluence uncertainty," in the 9th International Symposium on Reactor Dosimetry edited by Prague, Czech Republic, 1996), pp. 746-753.48 38G. Sjoden, and A. Haghighat, "The exponentia l directional weighted (EDW) differencing scheme in 3-D cartesian geometry," in The Joint International Conference on Mathematics and Supercomputing for Nuclear Applications edited by Saratoga Springs, New York, 1997 ), pp. 21267-21276.49 39G. Sjoden, "An Efficient Exponential Directio nal Iterative Differencing Scheme for ThreeDimensional Sn Computations in XYZ Geometry," Nuc. Science and Eng 155, 179-189 (2007).12 40G. Sjoden, and A. Haghighat, "Taylor Project ion Mesh Coupling Between 3-D Discontinuous Grids for Sn," Transactions of the American Nuclear Society 74, 178-179 (1996).77

PAGE 139

139 41T. Tabata, R. Ito, and S. Okabe, "Generalized Semiempirical Equations For The Extrapolated Range of Electrons," Nucl. Instrum. Methods ;103: No. 1, 85-91(1972). (1972).65 42A. Al-Basheer, M. Ghita, and G. Sjoden, "Inves tigation of deterministic simulation parameters for a 3-D radiotherapy Co-60 device model," in ANS Winter Meeting and Nuclear Technology Expo, edited by Albuquerque, NM, 2006), pp. 511-513.13 43C. Lee, C. Lee, J. L. Williams, and W. E. Bolch, "Whole-body voxel phantoms of paediatric patients--UF Series B," Phys Med Biol 51, 4649-4661 (2006).7 44ICRU 1992 Photon, electron, proton and ne utron interaction data for body tissue (International Commission on Radiation Units and Measurements, Bethesda, MD, 1992).15 45M. GHITA, A. Al-BASHEER, and G. SJODEN "Validation of DXS: A Diagnostic X-Ray Source Generator For X-Ray Transport Simulations," in The American Nuclear Society's 14th Biennial Topical Meeting of the Radi ation Protection and Shielding Division edited by Carlsbad, New Mexico, 2006), pp. 9 46D. M. Tucker, G. T. Barnes, and D. P. Chakraborty, "Semiempirical model for generating tungsten target x-ray spectra," Med Phys 18, 211-218 (1991).6 47M. GHITA, DXS: A Diagnostic X-ray Souce Generator 74 48C. M. Ma, E. Mok, A. Kapur, T. Pawlicki, D. Fi ndley, S. Brain, K. Forster, and A. L. Boyer, "Clinical implementation of a Monte Carl o treatment planning system," Med Phys 26, 21332143 (1999).3 49W. Gropp, E. Lusk, and A. Skjellum, Using MPI: Portable Parallel Programming with the Message-Passing Interface, (The MIT Press, Cambridge, Massachussetts, 1994).72 50A. K. Jones, F. D. Pazik, D. E. Hintenlang, and W. E. Bolch, "MOSFET dosimeter depth-dose measurements in heterogeneous tissue-equivalent phantoms at diagnostic x-ray energies," Med Phys 32, 3209-3213 (2005).69 51C. Lee, D. Lodwick, D. Hasenauer, J. L. Williams, C. Lee, and W. E. Bolch, "Hybrid computational phantoms of the male and female newborn patient: NURBS-based whole-body models," Phys Med Biol 52, 3309-3333 (2007).70

PAGE 140

140 BIOGRAPHICAL SKETCH Ahm ad Al-Basheer was born in 1978 in Ramtha City, Jordan, in the heart of the Middle East. Generations of his family have made Ra mtha their hometown for centuries. This city, which borders Syria and Jordan, is the gatewa y to the desert region of the Middle East. Ahmad is the oldest of eight children, a nd his parents, Khaled and Fatima, are both involved in educational careers ; both are school principals. Ahmad was involved not only in traditional educational venues, but also in chess, soccer, reading and poetry. Ahmad competed in several reading and poetry contests throughout his schooling; he won the National Award for high school students for Best Poet. Ahmad graduated from Jordan University fo r Science and Technology with a Bachelor of Science degree in applied physics in 2000. Upon gr aduation, he spent one year as a physics teacher at the Yarmouk University Model Sc hool. In January, 2002, Ahmad was granted admission to the graduate program in the Depart ment of Nuclear and Ra diological Engineering at the University of Florida. He performed his research work under the supervision of Dr. Samim Anghaie, where he investigated a new approach for reducing scattered photons and electron contamination in Cobalt-60 therapy beam. In fall 2004 he joined the Un iversity of Florida Transport Theory Group (UFTTG) to pursue his Ph.D. with Dr. Glenn Sjoden. Duri ng his Ph.D., He won several honors, including best papers presented to numb er of conferences, a number of invited speeches at national conferences, and several recogniti ons form number of organizati on for his achievements. Ahmad married Ghadeer Al-Basheer, on June 16th 2007 and has one daughter, Ju de Al-Basheer who was born in June 14th 2008.


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 E20101222_AAAABS INGEST_TIME 2010-12-22T17:58:22Z PACKAGE UFE0022617_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 25271604 DFID F20101222_AABEKA ORIGIN DEPOSITOR PATH albasheer_a_Page_132.tif GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
5947636394fc8136ef865f313ac55d7d
SHA-1
dd70f61eac8c9fb6ac91c55ae978ec0df74e5ef7
F20101222_AABEJL albasheer_a_Page_112.tif
87a55471662ad1a81e297f8206072758
b20ce3583363c156fcc74ca458049aae16211d27
F20101222_AABEIX albasheer_a_Page_097.tif
18fbc1cb4af566901130a107c08e0952
71772ebbd396ee6d51dda3c48367ffb27d629e1e
F20101222_AABEKB albasheer_a_Page_133.tif
c58d4d53c8c6d23ed2578f85e3b2f437
c1da5c09bdba2930ace8cc59353dcfe546e1fdf3
F20101222_AABEJM albasheer_a_Page_113.tif
ff8330fa1af536f1077ffff8ebe755cd
e3d5985881f50614359b0f3296a1143a5ab93a17
F20101222_AABEIY albasheer_a_Page_098.tif
ca36d057b9669551ecb7d48a74508194
c661d2644e0b05f96a559c9c4ff80890b03edd48
F20101222_AABEKC albasheer_a_Page_134.tif
addc9761335ac968cc3c797ade4e7134
4ab44db10b5c66634e1a9a666260b7cc8f4abf4d
F20101222_AABEJN albasheer_a_Page_114.tif
9eb7f59c0de8e06b1f22a067c32fc2a3
66123d94dd2efa5f049b6fce2f876592f98169fd
F20101222_AABEIZ albasheer_a_Page_099.tif
00d776f8b208fcb3031a59cacaa971d2
8494afca9365f32fb315ee090d271189e6b7a851
F20101222_AABEKD albasheer_a_Page_135.tif
a250893f6a382e27d7f84044aa2c9c85
878e75f65e88432221a5aa8d1ac9d36c4af146a5
F20101222_AABEJO albasheer_a_Page_115.tif
e39afb7fd521034d7e21dfc992a8a717
c0fef9ea385e1db370b7000d3cad11374bae2821
F20101222_AABEKE albasheer_a_Page_136.tif
919e3b923c76f66060c36a4a500a3d55
6dcf3d91b803df1581b509c0eb3e17a6bcaa1308
F20101222_AABEJP albasheer_a_Page_118.tif
8e454d3b70235ef301bb7d1ff08456ca
cea2bb7c369872e428dcec75ab5338b6571d3f88
F20101222_AABEKF albasheer_a_Page_137.tif
0a3eef9bcf77087cf3241d797f151440
ae79a4773d8dab107f5a90ed4a7001925d006c32
F20101222_AABEJQ albasheer_a_Page_119.tif
52c78cec688edf19b5d97baa6ad798a5
b771c4755a0b40f5b8cdf63db7e0386dc9bc7d57
F20101222_AABEKG albasheer_a_Page_138.tif
f508b56a746f3b38e49182e510da26b7
da65d995fe0e7d958be718dd55edfc97ed821ee1
F20101222_AABEJR albasheer_a_Page_120.tif
a74aba11e0991b55077ac3c9a4f302ec
1be9a8ac93491956af7e5efe06a9d77b8e815814
F20101222_AABEKH albasheer_a_Page_139.tif
cf74c0eea0c2339dfca91f494d5220eb
49aea79e4b41330bda19e4f36a3b05615a7fbd2b
F20101222_AABEJS albasheer_a_Page_123.tif
6bba46ae9f9c2391ae90738b6e791d08
21b24764aa493bf16bcc73cddcc6c646df275c52
F20101222_AABEKI albasheer_a_Page_140.tif
8958fbc2c0cb9b64304f94ce9e5db788
d02742bbdd2d02a11d6ecf96c3f98cf63541bc4d
F20101222_AABEJT albasheer_a_Page_124.tif
e3cd24be03526c7b840f16f58f587b90
2c474ded775258238fe690fc0480e3eec160b6b3
8271 F20101222_AABEKJ albasheer_a_Page_001.pro
dd093eae9bc053228c0125897363e8a0
1cc2a014ba31aa81107e50f0350e869e9e8155b2
F20101222_AABEJU albasheer_a_Page_125.tif
3e137cc83c35ed454da5463b3d3c51c8
4fdd9646486a7db29f3e54b890b1605315f0f284
1457 F20101222_AABEKK albasheer_a_Page_003.pro
76268ea4f24fc306d3d55c224d08ac24
197de2f9b6fc801cd3503dd11bb0d23adc91c122
F20101222_AABEJV albasheer_a_Page_126.tif
84ad2c0621815e50db6c4f1e6c9ae1fc
71ab64c65c6c502d2917732c96a0af59478bb95a
31992 F20101222_AABEKL albasheer_a_Page_004.pro
912b2f37ef659d04d1f2925bc76f0650
618b2ba9b16ec50de7f346506bb652bcc5d9e38d
F20101222_AABEJW albasheer_a_Page_127.tif
1080c44302a9cde76107c00a2d1e73a2
22cccbc77a088c1e255c139d7d90e91bd202179c
F20101222_AABEJX albasheer_a_Page_128.tif
964e9b07b60d07b4cf5fe5d6b910e0f1
94fe4952de6678cad1b288c13635f4a123094401
47865 F20101222_AABELA albasheer_a_Page_019.pro
3814cbc12d108bc8aaf547fd8db3f0f4
af4ea1dea7c05d85d994df535defc6f1bd47bb3a
72050 F20101222_AABEKM albasheer_a_Page_005.pro
99d0bc76a517f26e261ea9d36dd4e61a
b9d8c6148f166513ee8861bf8a67433ac9f77d83
F20101222_AABEJY albasheer_a_Page_129.tif
649e27848b74551b8246a541afa08296
b68e697d86c0f29f089392c026e8d2ed2812f66a
43031 F20101222_AABELB albasheer_a_Page_020.pro
3a9d6c76af23bf440fd58cc6eb13c824
97bffdfba9a1012ccff797933d5830cc06dac48e
89169 F20101222_AABEKN albasheer_a_Page_006.pro
7940e2e06add6b091e0b25a73687216d
dbe9740e501c230337165b599bf0865298b4ef13
F20101222_AABEJZ albasheer_a_Page_131.tif
6bb755024bd43d11f5429014287f7a7a
7f0e2cc3168b951496216f51a66acb50ba6603f4
43692 F20101222_AABELC albasheer_a_Page_021.pro
c276aff70e8fcd68eaec49ef37b1dd2e
abef1d02ada4a718398f61378d1e7c05eb4de293
33588 F20101222_AABEKO albasheer_a_Page_007.pro
9a8543e00715b565bb27b274d1f31380
edd24d62376aa495153ad660a7a2bb12578bf33f
44192 F20101222_AABELD albasheer_a_Page_022.pro
ee9fb13b8e37f093efa6b344ca5ba0ed
a2ae30697ca8e3e3e0c7ea56dafdfdc2869d587c
53956 F20101222_AABEKP albasheer_a_Page_008.pro
727b424e97cc7f055fe3fe96ee605deb
ca1faaf2e07f3abe60ff38e049cfcd5f144f8a63
41984 F20101222_AABELE albasheer_a_Page_023.pro
95e46c1517a63690475028ba2ad3eadc
6f19d3fdeb47976857e92d28351daf4e992b526d
70173 F20101222_AABEKQ albasheer_a_Page_009.pro
de2a1ec2b4772714150a3249afcfec01
6ecaa3946af15f90fb301b4d15e9cfb08e6501c1
54456 F20101222_AABELF albasheer_a_Page_024.pro
7c24b2273420288ebe99a7e1a54fb6e2
15a5188638d5ef59a1c4e5bda53d1d305d3e9493
77509 F20101222_AABEKR albasheer_a_Page_010.pro
f9b2caeed11b1e2a0f381ecf5bc01613
d2ebfe8197685051f8dc87493214fdbe04dcac0c
35953 F20101222_AABELG albasheer_a_Page_025.pro
67ffb277369a1495b629fc8775a607a4
24012000fedca033ed4eb3759be012d3ef41d6d5
75138 F20101222_AABEKS albasheer_a_Page_011.pro
0e058dd9ded308a3b8090a22f46bb2c6
f27d28da9af2bbc5802790102adb4854629d2d9c
55373 F20101222_AABELH albasheer_a_Page_027.pro
636a71a8b789c94e43ca6b7182fa8373
7f30e9a1497c79b8fdc5f185bdd18aa1af7fd551
72397 F20101222_AABEKT albasheer_a_Page_012.pro
4b0b4b3c61ed3dc9ae4e81a628bfb94a
38cd00128c502e90d627595c11ce6bffe25e6ca5
37983 F20101222_AABELI albasheer_a_Page_028.pro
56706fa3804f783bb8c3999683039bad
4e7c37d244ae442af1cef334c09746ca61cdad0c
2983 F20101222_AABEKU albasheer_a_Page_013.pro
9fd496041b5208fd1bfe76ed3db126b0
2f6a5bf147dcd87803af4ddb6682911b3a00e68a
31231 F20101222_AABELJ albasheer_a_Page_029.pro
d8aa04ab26a6b2ca3554940ce5273849
47271b2f451cb8b3013f273747a3330443009801
48214 F20101222_AABEKV albasheer_a_Page_014.pro
c905657e8cf0736e006be092fad35534
b35fb295121eb88054935cec10297e8e530588ec
43086 F20101222_AABELK albasheer_a_Page_030.pro
783e85571be2cda2e135f0f2725b6677
8e2a4092de23f2658d180306d428ad885d751a34
24876 F20101222_AABEKW albasheer_a_Page_015.pro
f9abd2de8dd4337c42c364b4f8ce6a48
e67c02f649eb2dc8f5771a0f008d69620c9778d2
45746 F20101222_AABELL albasheer_a_Page_031.pro
0bfb8888e2124c29f75588627126af85
7fbcf9a3945b26041e41a041d89f17103062e602
51554 F20101222_AABEKX albasheer_a_Page_016.pro
551fea28978e8469625b4fe75e6b7c83
5dbedc94b202ae85c9c97488edeab15288e0e584
53615 F20101222_AABEMA albasheer_a_Page_050.pro
93b02c70e72e85770836eeac37bc478b
710453d6ec5a4c57b19efcd86c2a3d9c839b6cea
31049 F20101222_AABELM albasheer_a_Page_032.pro
c80e5cc1016cff1aa5cb23bc90edaddc
420c26e6c8df1fee9d20ba08633c37710f27a8ec
55599 F20101222_AABEKY albasheer_a_Page_017.pro
fa15b13a5c2dca5f8f3d512a35158d69
15b7dd63b633c411844bc988aa96e137e06c9275
22275 F20101222_AABEMB albasheer_a_Page_051.pro
b2cca672d5ccf99b429c2f7f19f03e93
f40cec7835cb73f0ee6ee71842b16bc2f3a54398
54729 F20101222_AABEKZ albasheer_a_Page_018.pro
f4d7c75723c336f23427bab5e4757e11
1e7fd5f1e366517849ea1364141b4256ea0d0fbb
32778 F20101222_AABEMC albasheer_a_Page_052.pro
15f7e7b84c0093752acfd00d09705daf
5ca1d85f16086421f2a0e841002ab0003bbe4fcc
30858 F20101222_AABELN albasheer_a_Page_033.pro
436fa8684bd6e160bc25c8a82956fbc6
ae230eeed331039553e78f7c7ead0db4bdbf6616
18911 F20101222_AABEMD albasheer_a_Page_053.pro
28926a39807bb205b5725e548bcbc98b
c88e43d31e1be4c2e6ce6a27cf29894cc2dd5e76
50284 F20101222_AABELO albasheer_a_Page_034.pro
5bdbbc3fc84f6c8eedc67f9d87e4157a
5dd4936d3d9f1161b6aaa5ec7964e92045ccd626
51743 F20101222_AABEME albasheer_a_Page_054.pro
1e8985ee5ea5a77d4d196f3632ee0aec
43cfe2a846953dbd2459e591ff13030c365dae02
45450 F20101222_AABELP albasheer_a_Page_035.pro
2665d60a20f7d43e20c5b3e4145da6f7
46e83987a31637c895200c6463326616f0a3e9fd
58476 F20101222_AABEMF albasheer_a_Page_055.pro
be784be4b160c21d11449257756cbbd1
096550a143a5fafea14f082c0ec174b0c0f250a6
57831 F20101222_AABELQ albasheer_a_Page_036.pro
515493700ded5c576542c692a38c1ab1
5a21a85df3dd02c28d3bcc86e841a4e170dfc774
49424 F20101222_AABEMG albasheer_a_Page_056.pro
f5b5a565363895ebda32aad22391df1b
3fddc31777fb1d3e51c3d031a3643e67313f56a3
48634 F20101222_AABELR albasheer_a_Page_037.pro
6e877800aadfc78bc6838df5495b5733
024c5499d633f5a4aa407200a48b05f46e0ea824
48005 F20101222_AABEMH albasheer_a_Page_057.pro
0416aa332a8a59491bac257632839b0a
85ec1fa540a02b246b6db799ef38c4151ac4468e
30228 F20101222_AABELS albasheer_a_Page_038.pro
f3827635b353e0ff7728362c5a8bb782
d84fc2c4ef208b2d1c830259270849ddab59bf8a
25215 F20101222_AABEMI albasheer_a_Page_058.pro
6c86b78e816be9cbe52e2d40ddb2add8
5b0ad56d8cf4ce754d8b0a280e46384dee3c2915
52523 F20101222_AABELT albasheer_a_Page_039.pro
84fa9d07c76757a3729d0e2b6f779e68
e5cc499ab026c74954f9d8eb53f5c0a32ea1ebcc
51097 F20101222_AABEMJ albasheer_a_Page_059.pro
8b58e87e970327493819c752317ba7cc
bd010d3ce4b7d5e62653ac887c6fe19e05163707
29167 F20101222_AABELU albasheer_a_Page_040.pro
18a157dc003b6c38323e0959648cdc2f
04493a905c364ca82a089c6fb8c6ccc733314dcb
24751 F20101222_AABEMK albasheer_a_Page_060.pro
81d11f3222132fcd86eefcacc114e7e4
59bf40eb31ad16ae78e929f544c186c1df95b15e
41626 F20101222_AABELV albasheer_a_Page_041.pro
7430f04e936a8444073eca643b46c855
4fb39260c06d90c6fc052d9cabdc7a5acac8e536
49381 F20101222_AABEML albasheer_a_Page_061.pro
95e4f554eb1a57d8b73655c315163e81
bf8fb07245d04f706a86175d87368e5735042adb
12018 F20101222_AABELW albasheer_a_Page_042.pro
7d08e484245d196092eda896aa807cc2
1b8034619496337f7333c3965a1e8167024e9051
28376 F20101222_AABENA albasheer_a_Page_078.pro
acc7f190b509c08cf270366074072454
b29813983020bfeb06cff2495b05382cc3c31c7c
51307 F20101222_AABEMM albasheer_a_Page_063.pro
c99e6be3d64435c591cdbe05406c326c
bbebdb4ebfbeb2723dad0dfae35cac40705ab9c3
40493 F20101222_AABELX albasheer_a_Page_044.pro
5d0ed26dbc5d829d368af8079383e75c
c1bdd3bb4e50b20691f3e987cb1a9835b2cb9d72
55119 F20101222_AABENB albasheer_a_Page_079.pro
a253f04e815b764e89ef5b5d9ced89e9
6e89539333697c32a92deece17637eb93f07016f
23907 F20101222_AABEMN albasheer_a_Page_064.pro
31ee09be6950758750faca163d40c2a2
928aab6f69b2991b913a57172c156a7f3e629b50
55982 F20101222_AABELY albasheer_a_Page_048.pro
256e9d460fb82d06aa46e88fbee654ca
673486e58ba647d2498d516318795c14f8242b5e
37659 F20101222_AABELZ albasheer_a_Page_049.pro
28e531fe675cb3df36052555643ad9af
8b5ec044aa17b4cdcd09b1494c9775c1881196d9
55471 F20101222_AABENC albasheer_a_Page_080.pro
0b5f2ea6b4a81236f3452c2cb45b803d
a6c965db74ccedf2a0ddb592ed3a97694e7c9ef3
7668 F20101222_AABEMO albasheer_a_Page_066.pro
4faaff87b50d699dbfaee07de07e4c68
71a4c2ab6f027e6003e2a36cbc5497796665f43e
51862 F20101222_AABEND albasheer_a_Page_081.pro
1444759d972d3e10dde05a3f920c7e77
456f6c9d80a5833269dd18bb45ddbb6e3761f5bb
22178 F20101222_AABEMP albasheer_a_Page_067.pro
22640bfa650cfcaa2ec1b5f64d8674b2
945a2d2f604d0e86c809b18b71291b72faf60690
31401 F20101222_AABENE albasheer_a_Page_082.pro
5481ccf14ae3e64fe2db7dc807730ac4
75152f56a42f5caa0ebfada36bf9448314a93ed3
26310 F20101222_AABEMQ albasheer_a_Page_068.pro
2767b1b6fcfa7d03c8db943b7c2b1a9d
d51e98c440bd764968158c7bb3f91c1b1c39c564
55710 F20101222_AABENF albasheer_a_Page_083.pro
f565bc13c0241722220b5cf4c9ec9243
6f07185dd5c38209059279f7d6b8523326953852
34094 F20101222_AABEMR albasheer_a_Page_069.pro
79544ca32ca5a39f9c39ec9e641d398d
76a9c278419a42db4bd0fb72382b12bc725b3d4d
13150 F20101222_AABENG albasheer_a_Page_084.pro
d755369a5302111db59342460c6d329f
0526405148a2bcde100a4fae2cd5f3c32b41e52e
38065 F20101222_AABEMS albasheer_a_Page_070.pro
c221b7e2469933341cf96d69e4a8a391
c186814d85920cec7c4926ddc75d00cb76e79a46
19290 F20101222_AABENH albasheer_a_Page_085.pro
ce812079ad39d9b5a0cfc8bc789abaf5
a18ee1172189a1036477ffbfc4162c8d779b22a3
6855 F20101222_AABEMT albasheer_a_Page_071.pro
0965bebe82caca9aeed84b7422f235bc
5a4223b29dcea76eff1c1c48d5a829f5491287c3
44400 F20101222_AABENI albasheer_a_Page_086.pro
9487f8b4de5ea45c67ebff3c5d93bb7e
fe401cd4b263879d83fa2b25ec659e8a32a0192f
6061 F20101222_AABEMU albasheer_a_Page_072.pro
3de879892c583fb2dc284159f28680b9
f3fff6ad44cde8282749a4bab625735971b76a69
23818 F20101222_AABENJ albasheer_a_Page_087.pro
223fb1623757533814969dce80773aef
cb11e7c227af2e8fae7ee103a81babfe1020d852
6035 F20101222_AABEMV albasheer_a_Page_073.pro
2be2666f0cd254bfad371a227cec1cdc
4a7f58cc7a160b6515c64edf6aad952c5a53533b
36006 F20101222_AABENK albasheer_a_Page_088.pro
507ada5d492836b91089ba3b2cff6711
a7c5f9fd739a02d48868a7397544c7926663e0bf
9096 F20101222_AABEMW albasheer_a_Page_074.pro
bb19d7843a487d03d344a7c9c37da43c
cc2fc9459542e30bda956eb34597657e38346301
35504 F20101222_AABENL albasheer_a_Page_089.pro
ef4f228f11ff46dce9a645870c1bd968
942954cd8391e735da9002b1f56f584dcbe5de34
44332 F20101222_AABEMX albasheer_a_Page_075.pro
54729fcd4421487d0bc3f0bade7ad6ef
a3b6aa50191c0d0e3139f4a0b253f07e4324a81f
36729 F20101222_AABEOA albasheer_a_Page_105.pro
fa16ca71d705aab06360c5b3532d7bc3
8fa4d7ab05c23f45fb517040656ddaae908571d9
44172 F20101222_AABENM albasheer_a_Page_090.pro
e9de0ec88a047927d812ec12dd08ac3f
b18b5c36577134c8978e070e312cbf86c623f113
44348 F20101222_AABEMY albasheer_a_Page_076.pro
079b76418bb65fab7b42e6fa7307fdda
130e39fb1b73f94d3ae60ff0d5d3ec25a8b0cf6b
23967 F20101222_AABEOB albasheer_a_Page_106.pro
eadf07dea7cd64ac1b0819b2f27ce8b5
64bff0cd267962ce3915caf4690f404da201a20d
6297 F20101222_AABENN albasheer_a_Page_091.pro
1271c4e02ad4fca65fa2e0e6921ab947
a2d95254361eed05132150263de936377213470c
52093 F20101222_AABEMZ albasheer_a_Page_077.pro
1eaf9199353160a0db65ec161e8e9ba9
87422001e206a84ba025c2b20573a3f6c6eaac17
34514 F20101222_AABEOC albasheer_a_Page_107.pro
773c9416dd0ddbaf9d748b921b29a5a7
38ef8e1c3a0548e654690390c986f600f2249c62
30324 F20101222_AABENO albasheer_a_Page_093.pro
7212f295399882bb40d4a007685efc78
752d35273318c1c786fe42bc70c174dba31db483
40390 F20101222_AABEOD albasheer_a_Page_108.pro
9d0e253d161f18e9922671719415b561
8d3e4a0ff6409dbfa37253b19d781dca1a27c498
38359 F20101222_AABEOE albasheer_a_Page_109.pro
ab0961fad3827d8a2a07fbcd1de561e9
8343d76040167a12a426b1b6e258286e5f0c325f
23064 F20101222_AABENP albasheer_a_Page_094.pro
58658bdd0e26ae0064122fc008160add
8b589dca5801b383cb5756b09621a80139461e0d
24030 F20101222_AABEOF albasheer_a_Page_112.pro
996309ad2ce05e1ffdd7de90fbe8b8e4
a9e941e3e02426c56767f818f9ff197c0a406ac2
34820 F20101222_AABENQ albasheer_a_Page_095.pro
04431ff9e574c22b0af3a314efb55c66
837f5e5e74d3989d0c06cafeb783a020c4b67496
29694 F20101222_AABEOG albasheer_a_Page_113.pro
c7dbba92ed6759d0d456c9421f0529ff
a18e244694967ac27148b93516999f61f80cc768
51834 F20101222_AABENR albasheer_a_Page_096.pro
de3443c9b0928d716ef17a5046f4b547
2fafbaf082f5fd933b320df50ea63f8540da1293
51712 F20101222_AABEOH albasheer_a_Page_114.pro
589c273e161a07aba931715f15988d88
110b7510fe997a7db72c0acd033e65f9d12ab061
25942 F20101222_AABENS albasheer_a_Page_097.pro
f06724fcb9f810c07efeae1d4f5b7586
b54f3b5b2dd846609924e688796775962e3bc352
24938 F20101222_AABEOI albasheer_a_Page_115.pro
979e9c04ca0c4d44518e38c9ef9b3e6a
ff786850a189ee556f9ed9aa780c34fb5a60a6d1
27883 F20101222_AABENT albasheer_a_Page_098.pro
9f706c98665a7e5117544d5a9d914d53
f85ebb8add2240085a22883cb744c1185607fd9e
45649 F20101222_AABEOJ albasheer_a_Page_116.pro
aa5e615cc1fb2471f09c258a05fd23e9
e5de5c56c7d0396a58e75206f211de2154a2666d
31465 F20101222_AABENU albasheer_a_Page_099.pro
549dc010066dcaca0f5a978a7041d2f4
5ee9eb31659932c75497170f701161e447ee14b1
52986 F20101222_AABEOK albasheer_a_Page_117.pro
e1b9b6fea9e524c537f2e01a0e785267
fd6f307e4a2e099638844a0cd47f87165a594668
53232 F20101222_AABENV albasheer_a_Page_100.pro
6e5b24c177b02ed50f9c7561ffcec6f0
b731c5e3f1e36448c307ee48dd27cd61abbd0b88
55086 F20101222_AABEOL albasheer_a_Page_118.pro
31ce79501e56143a2f8bef1cb7839a85
48d300b6155a0b7cda692c7ba059e2d4bdbc16f8
50528 F20101222_AABENW albasheer_a_Page_101.pro
71373685666a9f925f975beab9c88672
f9865ca57d60694f93a4738d922424ecc0ceca52
64066 F20101222_AABEPA albasheer_a_Page_137.pro
24b587873f36fa11c5eb41c35e45d24a
f2801b5943992cf2920c07dbcec6e628cba8ea4e
52455 F20101222_AABEOM albasheer_a_Page_119.pro
9f7d93bd46aed9ffa27179cbcb782b92
1bd9e16bb675e3d322f4336f88776b81447af5b6
40953 F20101222_AABENX albasheer_a_Page_102.pro
e627f2e74e60b75b186e19bcc3414232
c960cd5122b7073ae38d844474bc7ecd63588995
62271 F20101222_AABEPB albasheer_a_Page_138.pro
831ab5f2da0291d14841a799d9318ee5
5f93b6d40153a0aaca3896f75256e063f9cb307d
52592 F20101222_AABEON albasheer_a_Page_122.pro
1b0b1e09ce1b64abb461f3c16da403c1
416357493eea369f10e9c4dd90b78165cf1ef8da
50131 F20101222_AABENY albasheer_a_Page_103.pro
7833bdb6d1cca74334b2a4bd407aea2f
8975ff6212e9d4bcc9d47740ec32655a523f9618
54231 F20101222_AABEPC albasheer_a_Page_139.pro
1c6f7a82a8395f4ea1cdd0aa4601b984
1669ff5d38347d65a29529952f9ca62b6f93d470
55267 F20101222_AABEOO albasheer_a_Page_123.pro
398038b45f9cea3852235d021d634d43
76dcd0c6ca29d8b3afde4eec743c230f0e729c6b
35854 F20101222_AABENZ albasheer_a_Page_104.pro
84f6319f03a950bcc56ed696666dbc8b
6385268dd81a6b83c5b16f381d6b827473184977
451 F20101222_AABEPD albasheer_a_Page_001.txt
39393fc24951e10d3254b06c78e57226
ae8a5125a7072f2e3805c5e6038d11eea38e2430
33552 F20101222_AABEOP albasheer_a_Page_124.pro
f4a08e7b9707377916529ec8acf6b4bf
c7e0ac20742f24cef56b3f2dccf0e53e98b4da6e
96 F20101222_AABEPE albasheer_a_Page_002.txt
7f160930562edb792cd82ef980055a1e
7fc3bdd8ce6bed346124f72d0306184f0662d938
107 F20101222_AABEPF albasheer_a_Page_003.txt
cd44f919653d67b08b33b096664bfde9
eb039234462a2dd010b47211e51fecc7f183f64e
27495 F20101222_AABEOQ albasheer_a_Page_125.pro
447cee360a658e53ebc0f53368771441
8ab30f3c7ce5772e49983eb6cf13b04a6a0328be
1309 F20101222_AABEPG albasheer_a_Page_004.txt
b0a8cdc056163fa6e90e7023e83b984d
951fa518c94e1d31f4c638649ddbef77bc0f3d97
28164 F20101222_AABEOR albasheer_a_Page_126.pro
80369ba5f5915861d4fe855750f0570c
5d8b198cb8e6e122c3b6fab2624c1db65456fc7e
3179 F20101222_AABEPH albasheer_a_Page_005.txt
c71ce9c76885c693ab30a792a5b90b5d
64b7810b8043417b345a3f82ec25e08e2eed2892
15220 F20101222_AABEOS albasheer_a_Page_128.pro
9bb62cc3cb869544a1cf38ec547e0e4f
01408cf4b14117d065fd01d501c1d5d235502736
3687 F20101222_AABEPI albasheer_a_Page_006.txt
0e72815636219b28a42a36e8983e3a75
371c79f870ca652dce418c17472709147901c3fb
23689 F20101222_AABEOT albasheer_a_Page_129.pro
82e97a9b51ded502672a04b46fd203e5
1e278c4e960cf0a0401356ce23a068b21a1a209a
2270 F20101222_AABEPJ albasheer_a_Page_008.txt
e6ef4cb4d1b31a014342dbc7f6b98bb9
c8151218d8f0e65babf8a609a70a3417885e2e56
19708 F20101222_AABEOU albasheer_a_Page_130.pro
c65c0448d1b134cd6794acbaae560bf2
f824c75db95d1c8a504eedc7acc057dd15e441a6
3248 F20101222_AABEPK albasheer_a_Page_010.txt
6889360a6e970eb592daa4cfe20fb3cd
4492f54a126a7b8ac05980fa242b1445d49d1e64
44710 F20101222_AABEOV albasheer_a_Page_131.pro
0b00c41a403b851de4f215328906bf2a
7ed3110578dac508e6d75f26208c9ea7703fb0be
3073 F20101222_AABEPL albasheer_a_Page_011.txt
5c61b924756d52000a93bdda3e71952d
3283640152493c571d018a709bbd8ae4fe63f701
10474 F20101222_AABEOW albasheer_a_Page_132.pro
8139e38732264867c1675c7a40fb597d
8fd6c101c5de756ffcd37f0ceac230556fa7b2aa
2984 F20101222_AABEPM albasheer_a_Page_012.txt
c9a85111bc9e1a5d5eb672a0f013878f
e2bd9f26775545a999d9e7bc8f2630c5068d8c59
67076 F20101222_AABEOX albasheer_a_Page_133.pro
5f88726e44ba08210cd314268423e0c6
89ab6122242b73be51a61ced2edea1e106dcc931
2177 F20101222_AABEQA albasheer_a_Page_027.txt
f18ba3fcdd6645166fb28b79dfb1ddda
bc7664747b687fc2ea25c34fe06ab8d688b8a40b
126 F20101222_AABEPN albasheer_a_Page_013.txt
b24ebb33d27b5bf383c64a72f19e48ec
d0bd2b8e74cfae6b6b8644617e9012847bda242d
84412 F20101222_AABEOY albasheer_a_Page_134.pro
6d0e3523965fa2a610d55436649d6cf9
2c0cf653eadd3e12515d352f80c084a1d2f77f06
1660 F20101222_AABEQB albasheer_a_Page_028.txt
f0736ab250f5285c2abec40c1e120920
7912fbbfd1b4af08670c100ddf67031948a10bc0
2102 F20101222_AABEPO albasheer_a_Page_014.txt
ada215371f4b7ca4efcfd433e02ab301
081f1c51216932b73dc62d7a6404f38e32049302
44938 F20101222_AABEOZ albasheer_a_Page_135.pro
f3ff5fff7b3c8c2a50f8f5c8918328a9
2572f49188094c9df2813bc562e0a32468fd106a
1321 F20101222_AABEQC albasheer_a_Page_029.txt
7bd0e6f8070244f4754a3593cb6aad92
988d4ec264dbd281ee2aaaef62cb5de40b2d5a2c
995 F20101222_AABEPP albasheer_a_Page_015.txt
66daf86fc334ab48b0009bcc05601cc4
168731b3b27745df17fb2df7e6e98e26fc2f66ab
1757 F20101222_AABEQD albasheer_a_Page_030.txt
7f5f210f29b9ad529a531a2c3c29b24b
b4cbadb17c51c01e4220f3f1a77e1e8a69fef537
2213 F20101222_AABEPQ albasheer_a_Page_017.txt
929a9d997e5b4e2838bfbe9203328532
ddcd48bf2ee89a8ffeae63d38cfc5b384424df18
2088 F20101222_AABEQE albasheer_a_Page_031.txt
a4efae84b5df3c1c2dfd37a4d6840cb5
713043bf30be808caa9a431cb10c9553e723a3fc
1454 F20101222_AABEQF albasheer_a_Page_032.txt
1cdfe0cd7b84403f3af65769e2dd2d43
dab2f82a05be6f84bb66d2e11999dfdb83ab37c3
2161 F20101222_AABEPR albasheer_a_Page_018.txt
28cdff1dffa9f39d81421e6ee723e2fb
32f88a7c72fb4449769552f2e41aeec4f92a889a
1587 F20101222_AABEQG albasheer_a_Page_033.txt
c045c60bcafff7dc09fdef00ddbf2b9f
f9615a68c1693b9bb9dbfe76ae08eacf34480826
1932 F20101222_AABEPS albasheer_a_Page_019.txt
7331da8b78d87cf4e9387d926395b29b
bc2677691ee58de2c76519cf84c75b31141cbb61
1833 F20101222_AABEQH albasheer_a_Page_035.txt
d5edf0a237cac2c778e9e4713771cc27
a8d0eb8c481873979e127523496cb8cd5e115e1c
1857 F20101222_AABEPT albasheer_a_Page_020.txt
f51f9fabc2d55ae3307d5dcb32f4c922
40ffea6e980da5ab2926b81aeb7508c8ada48506
2333 F20101222_AABEQI albasheer_a_Page_036.txt
ce940ed375ae7bf830ff65c03b570cea
f30adc72e704034fd163080dc6778ec19cc0997f
1839 F20101222_AABEPU albasheer_a_Page_021.txt
5d8deea103401a3588e0893d31276ad9
573d265bc2b581ee78f8fc4309a3aabb3962e2fb
1961 F20101222_AABEQJ albasheer_a_Page_037.txt
a77e15e066f90202afeb2d9e44beb78b
e5249c03b264c983dde117710de3124a9b5957ab
1789 F20101222_AABEPV albasheer_a_Page_022.txt
2a9b56ce38c7c3a545d847193a28a977
711ee28d0709f8c22ff96bca44baca97625cb2bb
1351 F20101222_AABEQK albasheer_a_Page_038.txt
f4a05936515b94e42e1c09c98094edc1
4f95a2319e438350512797c45100c96c6aa34a23
1741 F20101222_AABEPW albasheer_a_Page_023.txt
1582ac77a8ca7986edea1e18b8536756
bc3061abdafca23a529ca48cdc23598ca73ac166
F20101222_AABEQL albasheer_a_Page_039.txt
67149dc5603ba7935626bd1620b8ae32
72fd2530709aa1c75153e82ab8f2e3624798ae12
2146 F20101222_AABEPX albasheer_a_Page_024.txt
6b5a4b9de5cb6768e0a027653f3bd261
972e7eceb63553ad945b98cbe32ca62597b4a793
2315 F20101222_AABERA albasheer_a_Page_055.txt
971994a733ec98fded1094fdebf126da
63594bd2414872e62f8eacbc724e0a24118fadfc
1237 F20101222_AABEQM albasheer_a_Page_040.txt
6f6fb032b358494c1d4e653bb4970e80
9ad228948d80dc382f4712be5105db7eb2f0dcbc
1518 F20101222_AABEPY albasheer_a_Page_025.txt
021445cee25b707c47b9373088cdbe28
631b60b3e6289583927562f03907add07e48102c
1951 F20101222_AABERB albasheer_a_Page_056.txt
e2892fbcc34561834aa850886c22fafb
c03f1f5fc1987e440ca3040d2b1eb7be5f462229
1644 F20101222_AABEQN albasheer_a_Page_041.txt
894c283bc7fc9aaed01ba82605b61e63
6928c24e39d551fe29fb241a99a66a8f3629e3d4
2045 F20101222_AABEPZ albasheer_a_Page_026.txt
ed0ad7ce6285cacad05cefbbdd023dcf
3820b4b51a40cee36c9b8506237ea4795c89279b
1941 F20101222_AABERC albasheer_a_Page_057.txt
4ad7778e92f3cfdc131a573681501df8
182fb4ad1c5b8145709d6d892be55ef4aeb8ddc5
582 F20101222_AABEQO albasheer_a_Page_042.txt
b628f3fb075adbe4912d5da8677b4527
cd2cadf21a2b5f4ffed6690e54d8c1e1b6e0137e
2037 F20101222_AABERD albasheer_a_Page_059.txt
f58ad93b6230f4ae686705475316a3f4
1a3609037bb550d51fc77fed90c3b6544b27a9b4
785 F20101222_AABEQP albasheer_a_Page_043.txt
5e230e95f9378709ac6e0bd86050f366
fdbf71c11ad0568609c4ca2f57d0e374e752565f
F20101222_AABERE albasheer_a_Page_060.txt
ab07282090628d58c6312b90141b5ee3
549ca2c98e00ad901f3c8079277ab2f0e50d006d
1901 F20101222_AABEQQ albasheer_a_Page_044.txt
73f85e3a031a3e7582bdcbda716fc3ce
6180d47dcd996316b96b26661b1d6a21f74a70be
1695 F20101222_AABERF albasheer_a_Page_062.txt
a784f6a3bd7b61b605e1800426cac258
a5b75ab8bb9ae477ed908a0051eb7bd52bb5ee95
1830 F20101222_AABEQR albasheer_a_Page_045.txt
d1ba0b05492e7debda85300f7d717ba3
c8d473afdd8cc6dd62afcbb54e17eb0b8832139a
2027 F20101222_AABERG albasheer_a_Page_063.txt
2ef5ca4996cb62e1d205b748c1e102cc
a73d1a5e392c444dbc86b5fbb7a7571d6b430f6f
1812 F20101222_AABERH albasheer_a_Page_065.txt
ccf45ecd39d7afa614d5e15e62168b32
7a437ac707f7e9efff9224406f1a4b21bd40a1f2
1512 F20101222_AABEQS albasheer_a_Page_046.txt
286973c63c4067c6ecc97834de5e2bed
1335452da47c256b75a274df1cdfc029da327b67
302 F20101222_AABERI albasheer_a_Page_066.txt
4c1209214608ad41cc50331efa33a134
174d4be7d320d9782773cd8cdecd7875fe69b185
1489 F20101222_AABEQT albasheer_a_Page_047.txt
56a912598aacb037e835597f18e8d332
9c69b78c3be1317e24da635e1d454519cf6d6dbb
985 F20101222_AABERJ albasheer_a_Page_067.txt
c57cefd4e460da2c37ead775b1cdbac9
f24d08e0314355a57fcd3813a6d4fd39286ad022
1620 F20101222_AABEQU albasheer_a_Page_049.txt
e701d9615a1054fa490985c45e0c7fb0
fc3292b609efde88722b50d490a09b12836b7075
1510 F20101222_AABERK albasheer_a_Page_068.txt
ff4ae57cc0672ec33b52971c4acb0178
cacebc1806a797c9b79c206a10afddf21135f80d
2105 F20101222_AABEQV albasheer_a_Page_050.txt
2d450c4e9c2b037181f4dc577ebe177e
fa3a7f8d41d0823bbe232bf5fe925d54d424fd34
1705 F20101222_AABERL albasheer_a_Page_069.txt
9e86b3f346d125851ebfe9c1fd6985db
7de5bb37704b551d3fabb987245b20a9e4335297
1264 F20101222_AABEQW albasheer_a_Page_051.txt
94c419d07fb7a2c5385689f2436dcc71
66343e98ddaa408a8b8722644074312849f78068
1520 F20101222_AABERM albasheer_a_Page_070.txt
5788d7f7de05dcb1deff404941b7920b
281872281a1520e6dba755c3f7a45a7589c684d5
1361 F20101222_AABEQX albasheer_a_Page_052.txt
a8f31de34b40921f3eb599a7a83110b6
1276b1d55b6f66bf378bcb8b98f7ad2042be2410
690 F20101222_AABESA albasheer_a_Page_084.txt
adbc046e388563f595765c8a4e262181
53e7241fb99ee8eeefc8388b52acb5d05c437d42
357 F20101222_AABERN albasheer_a_Page_071.txt
e2e7c88976b490f05e83adbd14ddc546
4c438b409413f8c249ee11c41b6690328c646d1e
961 F20101222_AABEQY albasheer_a_Page_053.txt
d3fdec1689d945a4dfeaa8d71508e89a
378758b11ba8721430e819209ffae4d81b2bbcdb
1104 F20101222_AABESB albasheer_a_Page_085.txt
6f0c01dde5a59ea015efe27aee855ea6
b8bfe46bef67e76da270ac9005c52df7765b534b
310 F20101222_AABERO albasheer_a_Page_072.txt
134f5ef64609fd70ddb172409f203b29
0e34d352895c7a6464a0d05b562930e969dc7d81
2107 F20101222_AABEQZ albasheer_a_Page_054.txt
a81d9fd398803a8d922e769cde0d71c8
45b1af70a44f2132ec042bcc10146edc7f76d710
1897 F20101222_AABESC albasheer_a_Page_086.txt
afb549ff2e1c849fd19fb436c5033855
5e2ae8754efb00ad0e9536f7f9a4fc9c36342ccb
309 F20101222_AABERP albasheer_a_Page_073.txt
79b93810c9f306735cda5bf9c4abf674
fc03e622aed67fdde74737b250240b4022d49b65
933 F20101222_AABESD albasheer_a_Page_087.txt
7f9415d6d17ffb7b7130185e99de3f97
c7c856ee4a6b6328a30683e0f5b20f77398d2807
427 F20101222_AABERQ albasheer_a_Page_074.txt
aaedc17f972fa7bd4656bf996c1dc8fb
dede770867b082e02fa723a4439bb9e0f0776e6e
1490 F20101222_AABESE albasheer_a_Page_088.txt
02cfd6be1500e07e6da74db17860a7ca
3640cd5205b43fbb28210b8ead9afba0fc5fc916
2013 F20101222_AABERR albasheer_a_Page_075.txt
6945a979592680ef46d2614d2e540699
7e429c78d07f4bf0ae06193fb0f25ab9e7a765e6
1544 F20101222_AABESF albasheer_a_Page_089.txt
04838161f1a3754c49b1ee2f5dc7ee5a
299c3a4d826cc1a4e79a9918e07159709b47d143
1867 F20101222_AABERS albasheer_a_Page_076.txt
95c23195ec15482c21e7ddefd505a794
8f6a880cbeaa79d35c4776d190ee54f6e62e91e7
1824 F20101222_AABESG albasheer_a_Page_090.txt
76304711ff6678a8f18cd5bf44038970
c4188273e061bd9010e4243d0356fbd0cc13cbd7
547 F20101222_AABESH albasheer_a_Page_091.txt
f0a13e5ae713148981331a8b5b114d23
26e611c247276eafbdf6c9aba0a26642fa8f4b3f
2061 F20101222_AABERT albasheer_a_Page_077.txt
8c8d00721565a8d46e41bb79d51cf8c7
4b276d1e3a5cbd7f0bcd4e9fcefe9a38244cb02f
1092 F20101222_AABESI albasheer_a_Page_092.txt
472cac8944037c63a01900614a869323
79d4d07eb5465a4682e45e93458b2696dc421f5e
1130 F20101222_AABERU albasheer_a_Page_078.txt
1fb32848c21eadbfc514f1165e6a6374
3f41e420013bc9d849011464b8d4ed071fb62531
1352 F20101222_AABESJ albasheer_a_Page_093.txt
51e537fa2707f3025ccda056b2b8cd4f
3ca2026936821332320b23643a0a1b846f603c3e
2246 F20101222_AABERV albasheer_a_Page_079.txt
d3dc66afa0bdecb6d884fb9294b7f8fe
8c646cf0d15fa78ac1e29cbad51e8b321f9eed04
1496 F20101222_AABESK albasheer_a_Page_094.txt
9128c3af32b80db54744a65d5c57af82
de2be0e4c9ce508445038ed23a23dc9a45bb6ea7
2197 F20101222_AABERW albasheer_a_Page_080.txt
a0c7f0378305671c73780ebd82897b2f
ecd8aa5b947eaf89576aa96b77aefb7f431ef108
1529 F20101222_AABESL albasheer_a_Page_095.txt
a5d4db6acbc39a46d6678327f5baba58
0e91b095b09418ce25f5b5168d2cca6d5e852b8c
2075 F20101222_AABERX albasheer_a_Page_081.txt
674be0b17b1ab9bea91a33d407e01a02
e7f2c2a0c423ae9eea6b6ca81fd5ddfd1165cc45
1803 F20101222_AABETA albasheer_a_Page_111.txt
31f7000f34ac133d6b2eeb4d69708d6e
90fa9aeadbac498a1528e929d1eb63856795d2d4
2047 F20101222_AABESM albasheer_a_Page_096.txt
3325d5f1a811bcfd91f2cb358e0e6dba
5cb8174b68f13cb68dd5b5014785652083b0cc30
1322 F20101222_AABERY albasheer_a_Page_082.txt
3b61b5c5828bb7ace9d6e1d7cc8615c6
dd7d7eab43cb3ee85b6bd0d60e4073caebeb37de
955 F20101222_AABETB albasheer_a_Page_112.txt
9b222a30f48d91107972d5ece707db0d
dc7769467dd718da7905e180fa259b9efd8dc73d
1246 F20101222_AABESN albasheer_a_Page_097.txt
044e45447bd250a4e3f15f9ac7897e80
2d1aecb75d8ed44f290a9a867b90d149a3e0d62f
2184 F20101222_AABERZ albasheer_a_Page_083.txt
e2f05023d426be26c5082665b64daa98
e09eb4af2695c43e39658d0c189a5953fe354b24
1239 F20101222_AABETC albasheer_a_Page_113.txt
b25a7ebd5e96de2d9c4339ac8aef443b
ed125defd960a64e862cb3152355d94794df5d59
1408 F20101222_AABESO albasheer_a_Page_098.txt
abee55ca8f8f4b1ce6fdca587a7ad66b
60984bb189b64a2b5fbfdd94fc3214f28c5e8d5d
2046 F20101222_AABETD albasheer_a_Page_114.txt
8ed39ec0549f25a9e76730001839dc8f
e5e831246341c0d12ddb1e3969c3daa505511715
1337 F20101222_AABESP albasheer_a_Page_099.txt
17ef58f2b75cfcdef0b811a2c4848144
a134de7736e82dcf542d45b22efb0c4ce9107f9a
1161 F20101222_AABETE albasheer_a_Page_115.txt
e59f28feb0bdc2905b5b268f43f4eb01
708c2e8b846c88bacc44d695c154635556f33e27
2104 F20101222_AABESQ albasheer_a_Page_100.txt
fb8f7372b8be0f3f7f2397bc619f7ecd
9414eaa0f4c6add47a76a657ed60b5e8307a1ff9
1915 F20101222_AABETF albasheer_a_Page_116.txt
86867b0c63c5578a3e6297d918085817
ecf4f011c795c29064f57f11f13e9291e8ce67e6
2074 F20101222_AABESR albasheer_a_Page_101.txt
321281c2660f5bb811bb194c9cef52be
92ec5e0b95d5526718053440480b5135ad208452
2257 F20101222_AABETG albasheer_a_Page_118.txt
be5ad90e1e83d56246dba03f8810a2a4
3df0987d8a5c3c9b0034d1a29b12e1a1f13fcd6a
F20101222_AABESS albasheer_a_Page_102.txt
bcdb09969cfec9e1889834a1a6ff7959
1fd70cb087f05f376f0bbea91e3b0b998f0483e0
824 F20101222_AABETH albasheer_a_Page_120.txt
f8833ac9c45324c693c554bb95a206e4
e5d6a1639b4e3fb9ec9a21b612f8fc3badd67d0d
F20101222_AABEST albasheer_a_Page_103.txt
fa02bac0828853eeaa1193e8093d9df9
6d0ccfbdcccb7bdeb7466de1402f8b3c05374d6b
2006 F20101222_AABETI albasheer_a_Page_121.txt
235dc0d012e2df806bf2fd9a309496a9
d7db550ede074df09d2adc51204379b9d9b271eb
2070 F20101222_AABETJ albasheer_a_Page_122.txt
c752c3eb6fb67fd649923815b70c0464
b68f1a946fd75581509bf28809bda0b827ff4a21
1685 F20101222_AABESU albasheer_a_Page_105.txt
136fdb8da264103181c83d895aecf7eb
c01899fcf544fbcf50818196ddf98555f9f1ce0a
2204 F20101222_AABETK albasheer_a_Page_123.txt
6c20f6782fff09ad4f4102c13aa6c409
d2ebae9e5d4cb5e53e108d1322df8b0e7514db42
1404 F20101222_AABESV albasheer_a_Page_106.txt
3e6e72d3030dacd9504fab6dc1f94237
f064740c913ecdd37f96ade8083ed46891c34439
1034 F20101222_AABETL albasheer_a_Page_127.txt
0dea38e9b4d7a42868998a775db2af73
8549be59f4f5b61d8a5f103906f625b1c5f35420
1424 F20101222_AABESW albasheer_a_Page_107.txt
527ef6abbe8db7567d3f8afa1d5f0527
e97e48927e92d2f9cae6b1cd58212d5f2ee6426a
41020 F20101222_AABEUA albasheer_a_Page_118.QC.jpg
d34215ffa6b9f03a8cafe6178dba87da
b06f535ecee7bbe91fd8b7e7564b8a5c749a1e7c
844 F20101222_AABETM albasheer_a_Page_128.txt
dac8172ba8294845b9c5b16563e791a7
cdf2291228824577a8848458c05baca2bb1e03a7
2020 F20101222_AABESX albasheer_a_Page_108.txt
89cc7206aa2f7d175f80eb3da72f819c
75ba6b794ddee6b967d20e713f1a43651a31fa51
7885 F20101222_AABEUB albasheer_a_Page_093thm.jpg
3924360af863275cb8c6b1ca898b8ff8
9fd4442e79938a70d64cae8b1a4aea993a48242f
1206 F20101222_AABETN albasheer_a_Page_129.txt
e5a9a536677a5500c44ee3d1a4ea4350
51201ceb48c626aa6475317a9f4ea7322a6453a8
1759 F20101222_AABESY albasheer_a_Page_109.txt
1126162752b6142f7e90f887160cce50
c4a5ddf485e7ac51ce124d8adae8262f0216ec2d
15338 F20101222_AABEUC albasheer_a_Page_120.QC.jpg
978c9e5d1cca49efb8ba302d7d03e940
0f93349817e1b8b961d41f9ade20286bf51c5c5f
1105 F20101222_AABETO albasheer_a_Page_130.txt
ba9ebe0784b99ce939180428f96dee88
f29d9eda4687d67cdca7fd86c3a9261c19512eed
848 F20101222_AABESZ albasheer_a_Page_110.txt
e5b37b313baa578e49ae865a10a0b2ea
8c1fe01cd8fb79b5b2f4ca9f46da52d17dcbc5d6
7532 F20101222_AABEUD albasheer_a_Page_106thm.jpg
e0e36230aa8cefe2b229aeb501ada61d
b84e99fd64aa6705f2c4d8737e91fe8e00f9340a
1847 F20101222_AABETP albasheer_a_Page_131.txt
df1115a9234c849cc1de28c264585dbe
4908801b03004498ed1f47ea724542f508db866b
9138 F20101222_AABEUE albasheer_a_Page_113thm.jpg
1d86a50e6ec98c3172dbc082e887cb18
185d58fb7ebc3b92289131899bda31883404349c
636 F20101222_AABETQ albasheer_a_Page_132.txt
9879d604b9bbe7d8732034e50b84668b
d3618b23b74abd217ac1222d186bc011f5ebf6bf
30727 F20101222_AABFAA albasheer_a_Page_076.QC.jpg
2fa83c20751d66f88d980ad7142d08e8
a3435b5b9e166e0280e8f19fd1bebd8d8ed3534a
32811 F20101222_AABEUF albasheer_a_Page_031.QC.jpg
e2f5a316d5cbad70249d039c60379a69
51d7e088e82c9d4eeb08bf07a22803409691d540
2852 F20101222_AABETR albasheer_a_Page_133.txt
6f77e32a5b74149985c0b153383f682d
4d8ece7681f3d9918745465df1578a690da4d1e0
7760 F20101222_AABEUG albasheer_a_Page_008thm.jpg
420777a28330260b965154749a7cdd33
97647e0b9a7f85d4b01913bee818ef1b042e9e26
3338 F20101222_AABETS albasheer_a_Page_134.txt
e53365370da7fa1c9daf0dbeba970f62
3624aba5500b70bd4b889027583862364e970e0e
8504 F20101222_AABFAB albasheer_a_Page_076thm.jpg
9caa4bea4d5805d670bf245458f931a5
fb49242b1658b75bb368a26d4a909516d0012a13
37984 F20101222_AABEUH albasheer_a_Page_134.QC.jpg
457dc1ae7c7b03b97a5e9e279af0a173
1b1c225017c9a11d47fcb25e0b111711598fdd01
1829 F20101222_AABETT albasheer_a_Page_135.txt
b3d4b990a8960af346f9753cb316b911
60b181f59f2a7b6b1aa3bcb8cd3e646c6d9b1be0
35421 F20101222_AABFAC albasheer_a_Page_077.QC.jpg
b14122fe280dee8e3ccd3ca09a47ef11
f48c0dfda71d4c1bc0cbe592759f3b0146f1f842
1193 F20101222_AABEUI albasheer_a_Page_002.QC.jpg
75e26b5ee25cc7819cc4d1cbf5661b66
a87d12570605395c7f0c61d0aa8b276d1c09123c
2482 F20101222_AABETU albasheer_a_Page_137.txt
131c837967f7bcc82d47e77ffce5316e
e66eddc5bcf4b63bd63aefed0d9c9c2e6e24329b
9088 F20101222_AABFAD albasheer_a_Page_077thm.jpg
39b786c649c9894c807d70b0d8b39a49
72ac814c00d8ce67a02d14a0863f16cb6702d006
36358 F20101222_AABEUJ albasheer_a_Page_100.QC.jpg
875023196a9c278a28223c1146bed580
3e297ec576587b60865b5e9921db6ab3783ea43b
4908 F20101222_AABFAE albasheer_a_Page_078thm.jpg
ac8361aa1857a62ec2794e18b2488818
7798a07e2b3664b93323489d6d47451a70236a34
8361 F20101222_AABEUK albasheer_a_Page_056thm.jpg
bf4031dd430339e3b3317c377a7431d5
dd61c594dfb0ef01eb9d7954b46d3a38f59c8913
2415 F20101222_AABETV albasheer_a_Page_138.txt
a2abd498a61e76455ed3ca9d8936822d
6abf3481cd1b4983a9b8ae6dd3e5f745bb24b344
38616 F20101222_AABFAF albasheer_a_Page_079.QC.jpg
57f6663ed6e874e738c83d5b1fe53508
a9547600074940f3dcecd63fec888442ef1a3da4
7602 F20101222_AABEUL albasheer_a_Page_090thm.jpg
67a0043a17f792aa00864913133fa857
025acde30d780685d948dfb2810485a7aaf09fc6
2126 F20101222_AABETW albasheer_a_Page_139.txt
ba4e0a406e0e86514355651cb5518c26
d73926f7717bfe48e94025e6bc5dd2146d0fa73a
9199 F20101222_AABFAG albasheer_a_Page_079thm.jpg
2df784e78f83af53d1d106693c9ea446
4e4d7b2bec1f4478b9f6392e3f5bc94874975e57
9389 F20101222_AABEUM albasheer_a_Page_096thm.jpg
8f05e9f3902cbf27b00b58bb2d0f9130
3423e6c63dd65b86ae6f512d135af3bd000cde4a
1866 F20101222_AABETX albasheer_a_Page_140.txt
fbae06cee8b29ab17e73dbdcf2a05f1f
b76cdb4cfe0f97e112449d0d7624096ebc0ded64
8558 F20101222_AABEVA albasheer_a_Page_075thm.jpg
0f1cbda34a6e7f3f56d986785a0d640a
eec02d8834c642dbb1c64ca32af20731db0d108c
39654 F20101222_AABFAH albasheer_a_Page_080.QC.jpg
eea462aab72079163255d6e31295a09f
e3886bc22d58e48219fbb91486cd56fd7338c156
F20101222_AABEUN albasheer_a_Page_106.QC.jpg
7bcd389a7c2f191624346c7f9bf113d9
8bf58da347b182fded7d878b63fdd2ceb626fdc4
2103 F20101222_AABETY albasheer_a_Page_001thm.jpg
2a95ee013449433288f062cb4478a1e1
b688c0a6d27e0612d95b78427da6e4630474d3f4
5876 F20101222_AABEVB albasheer_a_Page_032thm.jpg
6a8d1adea9d91a26624c3a7b618aa4d3
77c9e645ae13f613309d50d2b823d3c4c763552e
9695 F20101222_AABFAI albasheer_a_Page_080thm.jpg
adc07fba36dc06c242aec5c405483271
dfd4c125afa27aa14484a9453c623c68e6b40e22
5643 F20101222_AABEUO albasheer_a_Page_003.QC.jpg
7d6cdede6780c90aef0898cbc7b64923
959838cae028fca942c19f6380001aefcd1b281d
8572 F20101222_AABETZ albasheer_a_Page_121thm.jpg
ccbd542de2677266eeb38156b6ead4b0
59791852896dfa2c90a5e4decd8733bd1251d757
3397 F20101222_AABDSA albasheer_a_Page_132thm.jpg
203fbead30818e2e0f0c58ef2ff20402
9c00ec8def89b980f060c389816ecce5ef2d35b8
7806 F20101222_AABEVC albasheer_a_Page_047thm.jpg
a8a1371e15be9e0d11ecc89eb9de359a
ea88ba7c5c79aa8a717e19fa5155d992a6410722
34663 F20101222_AABFAJ albasheer_a_Page_081.QC.jpg
0708c9b8024f3a11fe94f7f4ae2dbac9
8648acb8f6468b8fb3c79147694a1960f5bcfea8
6138 F20101222_AABEUP albasheer_a_Page_029thm.jpg
3f09572d4cb51b2296e6a3121ad2d430
052b9778f8ce3eac5007f574f0479d92d5f77006
1051972 F20101222_AABDSB albasheer_a_Page_121.jp2
60683f80bf84d166439fdb0fab39155f
f63ff45f469166611a8282d02f65eaaea952f141
9513 F20101222_AABEVD albasheer_a_Page_138thm.jpg
33e009d45866972ab4856f80b25b4cac
1c2dd8c58ee2a7f16530de0e0d113c32d8da159b
9059 F20101222_AABFAK albasheer_a_Page_081thm.jpg
712f04d9c3f60094b3538a523fd01a5c
9b94104c2b24243959fa73c401368a889d667d0e
9086 F20101222_AABEUQ albasheer_a_Page_059thm.jpg
60874199b8ef9a3de8d7e1483644eff3
bcb84bb257d238a0ea3ea6687821c02ade86ad2e
8163 F20101222_AABDSC albasheer_a_Page_086thm.jpg
72a4a495dfa3251436943f67deab9b10
625f618189dc479baec0665286f8e309834f0da3
10158 F20101222_AABEVE albasheer_a_Page_067thm.jpg
65919b43a4aea6965a83869988c6c058
d60353d8c7d398d90dd3d9acddc04e178feef1c9
7034 F20101222_AABFBA albasheer_a_Page_094thm.jpg
d70bea54ace41f0a8ad7121505197405
52e633ed983ee1e556861a5f517c8f22db935c2d
6748 F20101222_AABFAL albasheer_a_Page_082thm.jpg
342d63a4151fc3e446c4ec3deff5549b
777c725335ee85f511d69475bab12674cfbe2371
8398 F20101222_AABEUR albasheer_a_Page_107thm.jpg
c516aafbd2c09ffb566e44b8b71f9ca2
c7d52cf566b92f89b398e79fbbd4e0520167c63f
9393 F20101222_AABDSD albasheer_a_Page_013.jpg
1fd21c1ce667eb1e829d72e804b7ef23
b136b62451ec59a7e3e02eafc555a708776e897b
4882 F20101222_AABEVF albasheer_a_Page_135thm.jpg
7b48e3d971b39e80f28326db1462e387
bd9dcb0f85086c2fae95355f9207e02a54ece9aa
28740 F20101222_AABFBB albasheer_a_Page_095.QC.jpg
83d39d07f039dc3df07f84f79b88966f
329316f0d6f7ab35329aede1be1f8c6b293897ab
38838 F20101222_AABFAM albasheer_a_Page_083.QC.jpg
7eee3bfe4f9685c34dae270f7741146c
4464d76197a5986d3fa7d87129105bdf73a3def0
26968 F20101222_AABEUS albasheer_a_Page_082.QC.jpg
7cecf26609dc37f562f2f732c99fd368
af75a809eac58b8eb3c42d10c103897520a320f0
1403 F20101222_AABDSE albasheer_a_Page_126.txt
f9458139ebb48ba3bd8cf49a33b66910
4f31bcb6edbbbab4977bb2a165812a5f6fd63b19
30698 F20101222_AABEVG albasheer_a_Page_086.QC.jpg
04b01a8a5ec6c9ede8c739061021eff6
98dee45793c6ca4493d58af072fdb565e7e80477
9263 F20101222_AABFAN albasheer_a_Page_083thm.jpg
8d790cf62588c1f441a93022e2d33b29
5126133ffbc3c93eed1a4917fb54e3908e56375f
29923 F20101222_AABEUT albasheer_a_Page_066.QC.jpg
708c3b58ccaf1b7298ca562c4059fc43
77908110848fa9c680b3bd201b83a77bb21e2d86
F20101222_AABDSF albasheer_a_Page_119.txt
8f4bba57e5c51f80720f99fdb01d15f3
ccca120d5622ad3125d09c75c88ab3d12ed706ae
7879 F20101222_AABEVH albasheer_a_Page_020thm.jpg
e8e51f8521184a04be43ae67470099f7
3fc8965e379c7cb39797db0730eed1ff08d49544
8368 F20101222_AABFBC albasheer_a_Page_095thm.jpg
4e27f9a2b412b68c4ebd7b02cc5d9da6
1c85c9fff0dd49210b95ec1fea32dcc97b63c546
23097 F20101222_AABFAO albasheer_a_Page_085.QC.jpg
a40c8db845b61f413959984525876b7e
50bd3f913988f8bb9af19972724f810a9ee820f8
8546 F20101222_AABEUU albasheer_a_Page_124thm.jpg
56e8a9efa6b3c4114adee22b21709500
95bf00c9a733922641ade39bd42fc2571983d575
60611 F20101222_AABDSG albasheer_a_Page_078.jpg
366c83f43b57ccde5bed592e50f9b3de
9013c23557c48118f147942dd90e0f05ad18c525
32874 F20101222_AABEVI albasheer_a_Page_014.QC.jpg
d8e1ddd4297cf4e2a44c47a7be06a4ad
3bd53520145b31e578b29e85dff9979e67b75203
36360 F20101222_AABFBD albasheer_a_Page_096.QC.jpg
5a98f6ae9adb7a81a2c0c658d9e1702f
e644f4e14ec508961936e049dd66d90658b6752c
6608 F20101222_AABFAP albasheer_a_Page_085thm.jpg
ac940298a14b9dee97dd57582f4950ca
7f22ca16929b3bbbad0d75d3ad3d9a5f0d28c925
32046 F20101222_AABEUV albasheer_a_Page_065.QC.jpg
c4f871c4f301d1eac033f5dde224b664
8753468c69c4c3f42ffe0ca2b0628750f1b2d15b
37004 F20101222_AABDSH albasheer_a_Page_047.pro
709c02c50150ca6bb867baa904fcb1bb
c801a165cf00c064677ce4195b89c17a1915b1c2
16381 F20101222_AABEVJ albasheer_a_Page_091.QC.jpg
9503afac139b20ccc62c56b24800ea06
48f645ece10d01cef3510c2d26bc9778dadc9ffe
34762 F20101222_AABDRT albasheer_a_Page_034.QC.jpg
70fd28b46144afa52c2b9c94601bc454
522dc83701569c4d2597f40ce043c8fa4df47c72
8374 F20101222_AABFBE albasheer_a_Page_097thm.jpg
2259dba3c53eb9e3188446a9e0cf0fb6
d459d640f3a891a3f8a08a031519b9fe832bd49e
6933 F20101222_AABFAQ albasheer_a_Page_087thm.jpg
d6a239ed5a40046767c8876bbb65c989
8aceaed27b76759ef4cfbc15337b193a938a3060
F20101222_AABDSI albasheer_a_Page_014.tif
e0a8e19358caab410e9c0f86b91b7380
11b9cca48e8bbd25b5efca61afa46ba533b7dd89
7624 F20101222_AABEVK albasheer_a_Page_139thm.jpg
2bc72e953d1156e2c0b0b161b18c725e
b807eb234362611b6bd2e772fb72b73effb4e44b
25681 F20101222_AABFBF albasheer_a_Page_098.QC.jpg
a203a1dfb8dedfa1d499b27cb4279b0c
2fe23f7fe1f271d235f9708782e5abfb7a299e78
32491 F20101222_AABFAR albasheer_a_Page_088.QC.jpg
03d9e1ef2e7575d8c85bc1d18e1b269e
0927d73a9fedad479dac6bf2c7204323b6da9616
8608 F20101222_AABEUW albasheer_a_Page_006thm.jpg
1063d485eec306c837cd2e877f0685c9
da76d586a7e2ff25614a6658dcf4336d3911a528
30408 F20101222_AABDSJ albasheer_a_Page_033.QC.jpg
7355120bc9e44813b74f82eecf53c290
7ac3097e75f18e043b7491a4e4f66dd8842a7b13
8245 F20101222_AABEVL albasheer_a_Page_105thm.jpg
ff919f836f263d15c956a1c44a997cea
dea0b5891fb67548166a60f211c01b9fb1757750
71904 F20101222_AABDRU albasheer_a_Page_112.jpg
2187f5a832e87180d7924f306a59dcbd
376b909119f738f289e290283c0e8e8145e53767
29080 F20101222_AABFBG albasheer_a_Page_099.QC.jpg
8cb2986c5e1423405fe8fb25c55bc3cc
fadfb461ed58900e8efe2e7b55d84a5f5948ac0e
8084 F20101222_AABFAS albasheer_a_Page_088thm.jpg
98582630b721c49185471ab41d744714
9a77e49c125030aaf77548ea8f1fd40ef197d696
22490 F20101222_AABEWA albasheer_a_Page_004.QC.jpg
4cdfacaf23e7f02b101f29e7a0677a9a
0608d71867ab69267591e19701b48e3b335dd99b
36695 F20101222_AABDSK albasheer_a_Page_111.pro
d1d81cd79110efa88f072303137e34d9
09995fe7e8ee3adef21a2f01e2c7acf2df79fe35
29697 F20101222_AABEVM albasheer_a_Page_097.QC.jpg
be4369faede0193316b92f18ed854a9c
adef1435778d2dfd4098d194ed45fdaf3d3784cf
F20101222_AABDRV albasheer_a_Page_116.tif
7512d4c58b2d8397915efab4f8cab3dd
7ae48a194337137768117766e30857515bf1e0d2
5530 F20101222_AABEUX albasheer_a_Page_091thm.jpg
be9a63eac5d8ff6289cad15d3fd4d3b0
edcefff0a0c4b6b3aee968d04923e5c911a00069
9530 F20101222_AABFBH albasheer_a_Page_100thm.jpg
3d6c7c8f89f19b09c048ae199b168625
869f392cbe99c09d3debeb5857a16b9191af591c
30624 F20101222_AABFAT albasheer_a_Page_089.QC.jpg
b301e77a479d5a965c5dfc80272be3dc
13949f7a3b94f12f888337fe052c9fca0c40582e
5500 F20101222_AABEWB albasheer_a_Page_004thm.jpg
23b18532bc636df96059b47a30d5ce1f
a7f54adb0e09cf7538168f066af73cf9073428a7
51070 F20101222_AABDSL albasheer_a_Page_026.pro
c1b0328c769929f49f3a93238e87c7f3
6cf94915c34f14225f4b45bca562a3405ddd2871
34179 F20101222_AABEVN albasheer_a_Page_045.QC.jpg
28b4ef7a16995783871dec4222bdaa18
6d81d9dac9673301a0fa727a96696147721f7105
65077 F20101222_AABDRW albasheer_a_Page_136.pro
f05f74aa96da496455c72a495368d8ec
96e0512701177b8d62132e9ed97ced35fc4cdf3a
32767 F20101222_AABEUY albasheer_a_Page_035.QC.jpg
8a21503b2d06398dcbd3547e782e40d9
bee5e1a084aff49d245a529b5525d2aaa5a4019b
33433 F20101222_AABFBI albasheer_a_Page_101.QC.jpg
c278e799a728260a6a691ccd9331d23a
7183074acd8139448f45c789f200dd2e315d3742
8325 F20101222_AABFAU albasheer_a_Page_089thm.jpg
585dba7011a41e6c2d233dcc851e9c51
4806b19ddaf9c337e78571256ab56b27e20a83eb
27439 F20101222_AABEWC albasheer_a_Page_005.QC.jpg
2c46172e9141f965360a767d0abfdb1d
d0b6c5e3ea3ade15ea1b244894db1dfbd0144e29
45880 F20101222_AABDTA albasheer_a_Page_045.pro
b97b1f339458c1b17f87d5d0fa289513
8011efba7c0941d945d4de4420e5e9be565ed6fc
1988 F20101222_AABDSM albasheer_a_Page_034.txt
ab90d6c405cadcd43d9bc17331c0d3e3
c8b709cf24f8b595733ccde9216e8f69c371f0a2
7251 F20101222_AABEVO albasheer_a_Page_064thm.jpg
8e23b2d5aa25cde95a6dccaeb1619deb
efac04a8c3597556e8ed3c6738b8b2f2ade276d5
34794 F20101222_AABDRX albasheer_a_Page_056.QC.jpg
0786f13f276290f70d8716201dceb713
5af561c8e097d451bfcf1f0cc1eb489fd0ec50b8
41367 F20101222_AABEUZ albasheer_a_Page_012.QC.jpg
f03c858a0de2826dfdd6aed5b950d89c
f801931c6f790278759b1aaa522b009c0dc74a6d
F20101222_AABFBJ albasheer_a_Page_101thm.jpg
50d9c15e73cc747860d11400f9cf9f0c
c4eade426e23e6414080a63efc1ca136030792eb
28792 F20101222_AABFAV albasheer_a_Page_090.QC.jpg
db55c1312773e3d1a5aeced3df87d785
5bda03ab3a2491f16e5373122e29534b67b282cd
6859 F20101222_AABEWD albasheer_a_Page_005thm.jpg
250a7bab1ddd55e7df572e807fe6c1ed
48861cdbd89363314f82579c358bc36a06558c91
53337 F20101222_AABDTB albasheer_a_Page_051.jpg
e9cdeae6b9cfe17b4607e5b2c8159978
5feb3c55a8755b32cd3fbcb6ff42a67f9b818c32
93605 F20101222_AABDSN albasheer_a_Page_089.jpg
69f699883f581df95c683ed9ee7f3aa4
627de2f6b658f99f7f5874ba11ad8cfb0c15cf50
7139 F20101222_AABEVP albasheer_a_Page_098thm.jpg
64dee3476040997d41df44a47f636954
ca02f017335fe4204ca821668a5ab37efd492bd5
F20101222_AABDRY albasheer_a_Page_056.tif
34bb04a9282b5a3962b4af81ca984954
c185b478b24d5dddf9c4a5b72aaeab4995dac39a
29149 F20101222_AABFBK albasheer_a_Page_102.QC.jpg
ad4dd326e487c1852271c5640c4fd876
25240210c76b305413b1b6cc7ee1e1f048384111
15645 F20101222_AABFAW albasheer_a_Page_092.QC.jpg
6ec395ae3e8e832312c02382614ef2c7
dd6010185df34af13f28645ee7b5296ad6f15dfa
37236 F20101222_AABEWE albasheer_a_Page_006.QC.jpg
41efbc6afd941bdde675c6b9cac40a43
96cf20186ba106270079e62a7116b9eecbe45ddc
F20101222_AABDTC albasheer_a_Page_090.tif
81df765eb5260399652cdac542128a8e
5d3678543eebc7633de6fe60306e41e2bfb46dbd
12177 F20101222_AABDSO albasheer_a_Page_043.pro
e45f0225e521ace8d34f3a5ba6d04c19
4ab9a7b7746d2eded56f198432b5dfe3ad1ee696
38067 F20101222_AABEVQ albasheer_a_Page_009.QC.jpg
92c6ffe433630e26f41c1318631d33ae
79f5b665d7d3908203e5f020a134a53c8dc933d6
1051893 F20101222_AABDRZ albasheer_a_Page_097.jp2
d5b6900beb4f59923e9e8a89af5ddf0f
3fd7abb982f346e4667c7518e3e4e1e347f2c73a
7264 F20101222_AABFCA albasheer_a_Page_112thm.jpg
84960983e314b332624dd73d1daa6ef3
b33e575719bc074987a1c05684b982a628638e62
7292 F20101222_AABFBL albasheer_a_Page_102thm.jpg
78343cbeb85abea2662ddc7e10841a0e
fd786f9419e4f493984860c54e04e6705a4ae8a3
5032 F20101222_AABFAX albasheer_a_Page_092thm.jpg
834a614484569e066ed55b0f2084de63
6f129e88156fb50b0015897ea26bb11c65e90a6c
17237 F20101222_AABEWF albasheer_a_Page_007.QC.jpg
ff257cc22ee66efce8cbd25639be891d
b34b0debf0932df8a93e12770a1cf4d27fa2aed2
117108 F20101222_AABDTD albasheer_a_Page_048.jpg
112ed47703f07213c65f2318f1cafcd9
455c8bf440b041bd7ffd93ca9b4fd4c129ad2404
1051950 F20101222_AABDSP albasheer_a_Page_027.jp2
5cf3995f771192a364b1a9681010d1b8
b8fd6b54bedc64bd3a5b3ed6a2d2ae011c03fd17
16818 F20101222_AABEVR albasheer_a_Page_128.QC.jpg
50c5928c286bcb6601f3386f3d1771b4
b1a0f289dd204766c4a00bfddd67b8d127c05063
36073 F20101222_AABFCB albasheer_a_Page_114.QC.jpg
9363b9ccb0f55a27c1933c7e804b1d31
0b3376874b0f5ca8b9fd20b4a4e606d210afdb96
35467 F20101222_AABFBM albasheer_a_Page_103.QC.jpg
5263ba65e8a3e57e79d114d3c7a3ae63
1e80bbf126927546b86283459ab20062e51beb77
30670 F20101222_AABFAY albasheer_a_Page_093.QC.jpg
c8d44344f158db15139ca746e382351c
03bfb87b2d6478916cd644677433fc9d16a5e70e
4252 F20101222_AABEWG albasheer_a_Page_007thm.jpg
c428149bff1832dd1c7c9f57b676acc2
08e951b40a73c6c8c89c41f0448658027711f7a9
30992 F20101222_AABDTE albasheer_a_Page_008.QC.jpg
6fca740748d962d7b414315f5d8a453e
371f7ef2734c13b68b66a9aa6fb02c3377c63556
32872 F20101222_AABDSQ albasheer_a_Page_074.QC.jpg
d0aa5e38c6a98935913b27649b6c7aa2
61699be10c100611e825d9bcb7e8e4cb82c650f7
5458 F20101222_AABEVS albasheer_a_Page_084thm.jpg
ad452e4333b6fd18e716f344d94be8e3
83d9c0e8b87a9ab19bb51e857584187235d8c12b
9535 F20101222_AABFCC albasheer_a_Page_114thm.jpg
27fd1695110780a30d6fd20c89a256bc
36e3ad5593feccb326f436739674ea15f4bdb41e
9205 F20101222_AABFBN albasheer_a_Page_103thm.jpg
d7a2bc424b8a2a07faa6e31eb97a8bce
077997e9e16828b8d8b91547406a12f8110c772b
25380 F20101222_AABFAZ albasheer_a_Page_094.QC.jpg
216aba7ce418399727fbfd51d6c2d986
d76be265240e79e8c435668130ae18c3a669c634
9501 F20101222_AABEWH albasheer_a_Page_009thm.jpg
1024b8a264a64df6c90a5f28d5fe6cf3
f85de0cd200eb71ce049aa47d46f852fa9c5c209
1051970 F20101222_AABDTF albasheer_a_Page_079.jp2
837779cb65a2d67e60d4f95d1290d1a1
18a3d0549350bf382bd2dfca88c931aabcb0df45
3660346 F20101222_AABDSR albasheer_a.pdf
18a878bb750431ff81f0ddf75d675830
c2877b3dc6c26d1821dd6d14e0c94ed2e583bbda
8520 F20101222_AABEVT albasheer_a_Page_068thm.jpg
14a1bf5da63613803ffd6e7043efe374
969942312c3848c70c71236c96b68adb98eeb9bd
28850 F20101222_AABFBO albasheer_a_Page_104.QC.jpg
073331899671841d913b2a1a6ae8a4e5
884574d48701d5ee365a92405fed3c3fb2d383c0
41533 F20101222_AABEWI albasheer_a_Page_010.QC.jpg
e371b07537e2b784b0e55e53c97f7834
c57f7e7a196fab186cbc93eb8ffdf143732e92e1
31673 F20101222_AABDTG albasheer_a_Page_131.QC.jpg
d2ac90d952ad1c163e861612d4dd4367
1b0165c3511266c48eacbce134d88e495b9c6f24
7150 F20101222_AABDSS albasheer_a_Page_025thm.jpg
6bd73532ac493d08bc06493ea68ec043
0ca0603afa82e2588c3bba34d2340426aa3939bf
23885 F20101222_AABEVU albasheer_a_Page_087.QC.jpg
c435f419c0f7fd6101e255817115902b
c03c390ebb9726a18a8633f34395111e36d9fefb
30119 F20101222_AABFCD albasheer_a_Page_115.QC.jpg
68fffa328a548e5eff48abe4287390b6
6e721a67240e9cdc8a8acbc286fa831041210913
7826 F20101222_AABFBP albasheer_a_Page_104thm.jpg
d5ec803068462653e81d07a11abd6bdf
9f8b966d7259f77fdd108cc6afceff45c3bd93c9
10094 F20101222_AABEWJ albasheer_a_Page_010thm.jpg
0d791cec19803b6a54b6f141b22b2849
5be20b9488e1b2316484eb9b539a18692e8aeebd
32064 F20101222_AABDTH albasheer_a_Page_047.QC.jpg
edcf28cb50bbd422ded1ff0bb60dcb85
a5b107bfb91cd61afb764685b4958a0dda02f2f4
108714 F20101222_AABDST albasheer_a_Page_066.jpg
28749ed471f640e81bd41f193c9a66cf
300a9c9c37f9bd367acbbdc92d2b7defb722c5d1
37222 F20101222_AABEVV albasheer_a_Page_039.QC.jpg
a94777ddf27ea1384b3020010f8ce48c
159586546d227f99f26bfc7079835f67d0b28b40
8963 F20101222_AABFCE albasheer_a_Page_115thm.jpg
ea370e2b22283cf5c3b515d2220de0e1
84b3b676f070914fffd437f4715cd8c439963597
31119 F20101222_AABFBQ albasheer_a_Page_105.QC.jpg
12618f002163658cbecb8ba614cfb15e
00f8d1a593023967377423bfeb456180aa97a0b8
42815 F20101222_AABEWK albasheer_a_Page_011.QC.jpg
6694a72b44184e393ab5c810fafeb5e4
885df42d0408d90aa5e8c823ce570ee1b589ff4e
42525 F20101222_AABDTI albasheer_a_Page_062.pro
7a974788c2d9598ecc6590fcb65b049c
66855395ddbe7b65b49ed44f0430588eb1c91651
2188 F20101222_AABDSU albasheer_a_Page_048.txt
1c4fd31ef7240c782df4680d0ebf844b
207589affeb0457b66adfcdb271af832caf4839e
212716 F20101222_AABEVW UFE0022617_00001.xml FULL
09c015ad2e2ae2f52c26cd00362528df
2a0c7c6446af4887403d40ad1dea2279ffbb8b9a
33540 F20101222_AABFCF albasheer_a_Page_116.QC.jpg
0ce9b1745613ba3580dc6a7d1575ed10
1dc7e4bfad4c9473f7a820826db8900f8c311514
29847 F20101222_AABFBR albasheer_a_Page_107.QC.jpg
6bc4e7fc688fbc2a1b726416dbf35da5
2e68c49dcf366fda4d895a65f350ae02d9010cf1
10089 F20101222_AABEWL albasheer_a_Page_011thm.jpg
4c22e07094a0c4c017f20d1a3c760036
ccbeb4b3e7f99a65513eeecf358ef07c5f160e2a
1391 F20101222_AABDTJ albasheer_a_Page_007.txt
096e9ea5f00a28fc4ff1f8a6344c1a13
fb26062fcb2aa9474be8d333653a1e99d716086f
9017 F20101222_AABFCG albasheer_a_Page_116thm.jpg
20b8b5c9ab5c1dc6f91dc6b0e4002c35
6e57c506458fa83cf4a620aec0cc4bee09675010
32785 F20101222_AABFBS albasheer_a_Page_108.QC.jpg
9bbd215c1b843293144010d7b0039dde
910bc4d8c9c76aa40aa55b266656b9b427503fd3
F20101222_AABEXA albasheer_a_Page_020.QC.jpg
9de22f1a5b63e3dd8cc1191e06e222d6
1f1d51f75aeff71ba5fd86acee92c02883195c87
10032 F20101222_AABEWM albasheer_a_Page_012thm.jpg
b4b60a0126af5e656ad8137211e06aac
ef47f6ad5069a43968bfb52250643b23e3d0b41e
8490 F20101222_AABDTK albasheer_a_Page_031thm.jpg
e0e62dfb816def35a00853ef0b17aaec
0a8ece52145e62d0200dfeb99e398dbb30c94b19
1280 F20101222_AABDSV albasheer_a_Page_064.txt
e45977188d354e00bf09d8b00576572f
f94a9ce0f38d73262cc746efc53a25e153f0f7b2
8269 F20101222_AABEVX albasheer_a_Page_001.QC.jpg
12bee8ee81f4f0ed5a5a91b331b9bfbc
4eecccf1e875aa5405d141640ce9c93e0eb0535f
38125 F20101222_AABFCH albasheer_a_Page_117.QC.jpg
e9800a709a90984e48632e2230a360b1
8816db78d744f4a43f8274194afe981f81a24d2a
31537 F20101222_AABFBT albasheer_a_Page_109.QC.jpg
2a30d950dc0a5fb85e471bbe4aa92054
a07713f7bef6a4b6aa394a38f25c62835fc02394
32729 F20101222_AABEXB albasheer_a_Page_021.QC.jpg
c676282e2942ca5eb4109767628b3b97
6f31287f6e4a4d1cba86dc9319a6c751a13718eb
3210 F20101222_AABEWN albasheer_a_Page_013.QC.jpg
f23ce12eb266be7e4eaf9bc8ac5282dc
d2c1a35d622b39750e650613df1f979014119fb0
F20101222_AABDTL albasheer_a_Page_117.tif
70d5fa4a9ff02b930346b9fb6f2e3e92
a76c21561fcf74e5fa44288f0c74f1e31acd7472
F20101222_AABDSW albasheer_a_Page_003.tif
12794433a383afd01132311d2a86ac1c
96b71347d7ff4eb7fba48d4f3026b95bb2634be2
546 F20101222_AABEVY albasheer_a_Page_002thm.jpg
1a21a43f2c621c80893b0be94e067154
b9e57620faf41de487af27566335a6be66769c4c
9947 F20101222_AABFCI albasheer_a_Page_118thm.jpg
2b5d4926b04961fe20ce4ae05ab070ef
dc672f32a2c193f0ca24dab98d8ccdd766e276da
8400 F20101222_AABFBU albasheer_a_Page_109thm.jpg
7cd0d324cb54d980fc7b7dbe73fe27bd
8edd5cfdd7bc2a6d4902ee6e5f9a0318d36f894f
33052 F20101222_AABEXC albasheer_a_Page_022.QC.jpg
df3e2bf903fe7911582aee990b6bc092
c0b600be47a4ef054044b8a0bdd77cf84e1b46af
1032 F20101222_AABEWO albasheer_a_Page_013thm.jpg
b2a821feb0443ea4dd75cfe6b3f0f1ac
172efa094cd56543d6d95864f69e4294be033a6c
1831 F20101222_AABEVZ albasheer_a_Page_003thm.jpg
55b2149e1bc8c350ee1c9cb375376334
9431ebc76a39003e32b84e0068c1a08df76367c4
F20101222_AABDTM albasheer_a_Page_130.tif
ce639f5bdf16da554fe1b9165c7bd131
9cc9f80d2e85ca076b515e9ece8214d6e3fdadaa
7975 F20101222_AABDSX albasheer_a_Page_099thm.jpg
80f31bd1e1863868a2f44f2213968fc7
7405b0fbc09129a54cf0944658b372522dff349a
885 F20101222_AABDUA albasheer_a_Page_002.pro
97dca9488fd01df9ddb244e1fcb56cc5
58e7685ac3ac4694fb83f0e0b1617c7783f856cb
36908 F20101222_AABFCJ albasheer_a_Page_119.QC.jpg
578c4010d6762ab557a78e2eeed9f1f2
fa842987ccf3dce713e63510f41647b61df5844f
23960 F20101222_AABFBV albasheer_a_Page_110.QC.jpg
51f519d20d14aca348f72344080b80b4
f34c3200e9b492a262559f8edcc686a070b883e5
7988 F20101222_AABEXD albasheer_a_Page_022thm.jpg
ebe4242281cbea599193381d96316b84
1674ae68823dbbfcb569b3c588dfa13355dc6670
8081 F20101222_AABEWP albasheer_a_Page_014thm.jpg
659b378c544366376e00da7e5ecafbca
07433620affb08925cd84cb983c65e6c9b159ebe
F20101222_AABDTN albasheer_a_Page_107.tif
c3b73cfebd80906fc05bfea9c606000d
f76f86be4b56a23642bd8f369b7eebd5598f797b
33978 F20101222_AABDSY albasheer_a_Page_113.QC.jpg
95a2c1c06951753bfde8d5eb96051473
4d98f8b28922535e5d2ce82c3ca43ab1497c3153
4543 F20101222_AABDUB albasheer_a_Page_130thm.jpg
1941165644dc316674192ffe58d4639b
cc5c761f75fece3ba11495d3ce0e5ccfa9dc5ca3
9361 F20101222_AABFCK albasheer_a_Page_119thm.jpg
717c06e7f5fb95ffe0a32c1b2d6a2e52
75c815f7e41a49953ad10db2c5259d2ac935554b
6803 F20101222_AABFBW albasheer_a_Page_110thm.jpg
7b30c66b9a5ef2d86b00489a306b87f7
f65f91bbbe9b2fdcaf754586c4ba1d31a886f151
29970 F20101222_AABEXE albasheer_a_Page_023.QC.jpg
16831257dcc570d8552a93abc741d182
1dd50f171158f9dbeb0db116bca85a0872f60f87
18402 F20101222_AABEWQ albasheer_a_Page_015.QC.jpg
3af65e35c641171d59f956558b624928
37d7ddfb6476c4ce866a42d5e288532b434e70ca
45595 F20101222_AABDTO albasheer_a_Page_065.pro
937cafe0806d52bd9deb448f35232b4f
f1ca006bf60b276a7ad61e3214920592c0221480
109010 F20101222_AABDSZ albasheer_a_Page_059.jpg
335ed4609845f6047f2097ba5fbdee67
9d50429753138ac43dd79850eef0b7c17e9235d0
105682 F20101222_AABDUC albasheer_a_Page_034.jpg
c2b096c329a38eec9e5c709d6934cea0
01b4ee4e0f0c324b27974e902836468586949e25
5015 F20101222_AABFDA albasheer_a_Page_129thm.jpg
8825f6c693de2efbaf5b51dba7e13615
9ae9766421e9b05abfad3727800811b7a7c96ed1
3788 F20101222_AABFCL albasheer_a_Page_120thm.jpg
bdd5db5a278416e498ace1c8323c9c8a
b8cab439ae99639515a2c068f5603b6e307566b9
31174 F20101222_AABFBX albasheer_a_Page_111.QC.jpg
22b70a8e8e90180394d5c623f36ee742
7ac6d1a9df618d9e06fc6dfe7f487746dac6f1cf
7483 F20101222_AABEXF albasheer_a_Page_023thm.jpg
3ac813d258f5aa7c605204b99dd0b768
eb6935950da1cf0b19e7ad78feba5f2dce81eb52
4693 F20101222_AABEWR albasheer_a_Page_015thm.jpg
cf0eceb653fafff735679db51745770d
480a23e78f0c7a2f635864cd5bdaba80e5cda46a
F20101222_AABDTP albasheer_a_Page_022.tif
a30d1d63438cc68e147e5b68a63a3d6e
fbf119414da182c97af26370e2d73a4ca58c1f52
9664 F20101222_AABDUD albasheer_a_Page_036thm.jpg
dd6cc0fb9b203d488dcfa1321e8b455b
59e1a7067bd84121a3b0ae6e452253f908280ec3
14254 F20101222_AABFDB albasheer_a_Page_130.QC.jpg
316c9dc0096a0961851a97bbf11c61f3
3baa6c6f59fea31c6d3127db426ff67b42f83744
33076 F20101222_AABFCM albasheer_a_Page_121.QC.jpg
189364eb7a7e24ce28ad2d846d244756
e95c48889eedf4d5dd9689d284004c07232be20a
8277 F20101222_AABFBY albasheer_a_Page_111thm.jpg
0f6e88581b8e712e1d92221b8da2fa0e
1e75beed2dfe74295ebff68dd717c42ab4a8a44d
36867 F20101222_AABEXG albasheer_a_Page_024.QC.jpg
8bd154158c780871da0ff50db1be2ab4
a3f77b18b2e7575925405903967ea5e575c125f0
36273 F20101222_AABEWS albasheer_a_Page_016.QC.jpg
1dcef346bb2e61a9be9eaa6aae3a0afe
46f80243721ea6cc8b55339222c531d9a7b90952
14761 F20101222_AABDTQ albasheer_a_Page_127.pro
b54444ff4723099db8bb3b460388d72c
b822d9abf06201df9d7d28f0f07aaed88f981a25
8910 F20101222_AABDUE albasheer_a_Page_026thm.jpg
99e1e33c00eacd69cf3151173fd0032b
099e5462e215fad3745c1e48d26248e177a1a2c7
8062 F20101222_AABFDC albasheer_a_Page_131thm.jpg
737ea74c11abbf657962ba6f04b99ba8
59bf3721b40b452e87e4585c7289fbd312d3a328
35863 F20101222_AABFCN albasheer_a_Page_122.QC.jpg
0269c332594e546408166a80c11b28c6
1f4ce0901079b2ef2d6016aac4e084c5b5cabdb9
25359 F20101222_AABFBZ albasheer_a_Page_112.QC.jpg
a4baea671042f392311e9e9472759eb9
310e94ac3ab582fe643c7c141d7361813313506d
9521 F20101222_AABEXH albasheer_a_Page_024thm.jpg
ca4807deb634bbebc9790f37158e33bf
59f682eaa8ace6c2c8f156781807e96a54bc1c1f
8867 F20101222_AABEWT albasheer_a_Page_016thm.jpg
de676326a17eb27c240e9d4226ba19fb
a97f48f7454666473e52e90d69f55bf481b30184
46072 F20101222_AABDTR albasheer_a_Page_140.pro
bb1de7d21ceb8c1c3a0b0d15bd9ec766
0ea955dcd05bf4b967be64fac5c15ffd646745e0
103275 F20101222_AABEAA albasheer_a_Page_121.jpg
7758f1e8478a9546f97def03072101fe
1aeaff4f04181f11ba3349190c5d63cf504dce98
F20101222_AABDUF albasheer_a_Page_027.tif
6d985ac223c68f6127b075b8c5170c88
1f2f66c8c68543dcd81842dab391209547ad1f77
10473 F20101222_AABFDD albasheer_a_Page_132.QC.jpg
30fafe490f8cf33102ca0d4aa927e1ed
b37ebc25b53785047f8511aa94100d810b155ffc
9210 F20101222_AABFCO albasheer_a_Page_122thm.jpg
71551b40b7b5d19cb937bd06a05b8b72
5fb2862191e71b42d4c261e2f4e89272ae78d9a7
27293 F20101222_AABEXI albasheer_a_Page_025.QC.jpg
2ea29982e5d04e139b1b78b1ed05977f
c70a5c4d25d6a9c4754222dc4a857aaf9452e972
38524 F20101222_AABEWU albasheer_a_Page_017.QC.jpg
53c86b8d8054c21bb41d9147e9499da8
a3e696aacc87ff9b240deef9e1ed65f991ed10d6
88551 F20101222_AABDTS albasheer_a_Page_090.jpg
936aebe4849c10c5b2414df0dc4825ac
899c980336167dc0c9f49da8b056715400da3d77
113619 F20101222_AABEAB albasheer_a_Page_123.jpg
fbb528265aa8c5b51c6a11a4c70dcf19
9dac1273044a94404d4434adc3eebd632745f29c
954916 F20101222_AABDUG albasheer_a_Page_023.jp2
17a21886538c675aadce973df401a433
7d358b1f25813ab6702b74ce0548792ba1173f5b
35264 F20101222_AABFCP albasheer_a_Page_123.QC.jpg
3e5feb0067b2d65f41dc4c19143f69b6
76228eb76c9ee1807ec1a032e4f8d83b6ede5987
34057 F20101222_AABEXJ albasheer_a_Page_026.QC.jpg
e909f0f92b226250c5fb964c49951fc4
78bcf01f1b560ebe3601fa3aa4a27ac170e9e65d
9430 F20101222_AABEWV albasheer_a_Page_017thm.jpg
ac48f30afd8d26e3e09ead83fb02b90f
b4df78becad4dbffed9bec2d786526c3c13a0b83
23618 F20101222_AABDTT albasheer_a_Page_092.pro
763d496bb2537e703a553ac8b824b47e
f7cc027777af8e6aae81a87b253d92062cbbb293
20918 F20101222_AABDUH albasheer_a_Page_110.pro
62d01de185023d6f601f0c2c463c2b04
2ddbbd176a6bb74b93e5b40943777af830013fd4
31253 F20101222_AABFDE albasheer_a_Page_133.QC.jpg
027e637081eaae0f60617e46833a7ef4
c21e2cbff9ff6f595679bb0da450a9c18114d8c2
8904 F20101222_AABFCQ albasheer_a_Page_123thm.jpg
3a9b14e28088f4bdf50ff47c801939f8
9abd2a762a1a9843645d22ad856fd1621e8109cd
37821 F20101222_AABEXK albasheer_a_Page_027.QC.jpg
86cfc4a9bb0fd1d5ef78551d1b61aa8f
03307a89468c942f77a7b2772524cf0fcc1f79b5
36765 F20101222_AABEWW albasheer_a_Page_018.QC.jpg
4c2298535ef3d8aaf5c3cdacb7d3d800
298ebb4c18adf02eecd88531bd54e1f17c8b9253
F20101222_AABDTU albasheer_a_Page_117.txt
1c9a51f6a911dd798fea66cb1d09f8ce
9651dec40763c0782f67745ead2bf3a4a39df106
91952 F20101222_AABEAC albasheer_a_Page_124.jpg
4e95da2d28d480543016b02e6274d9d9
b6f6246ed1b57ff40b84065a3801532afc9912c4
7069 F20101222_AABDUI albasheer_a_Page_070thm.jpg
ed966910644021a6c129de6e33021895
c32c348f68450d4f6d3dd6c86bd579304e566454
6638 F20101222_AABFDF albasheer_a_Page_133thm.jpg
e11abcdc6d612d115d342bded1dbff8b
ac950aac36a76183afa85f9941d5bdde6cc18ab6
29488 F20101222_AABFCR albasheer_a_Page_124.QC.jpg
78822039dde06bf1a1b5959b840d284e
99fe261773f3ea2d26c14d3552b907a30094c725
9145 F20101222_AABEXL albasheer_a_Page_027thm.jpg
ce9e886b862fcc232d5216123ad35b01
0f11b946c5ecfd1e983982b718b94779f778754c
9274 F20101222_AABEWX albasheer_a_Page_018thm.jpg
8208a2db03e9ab55fb15825722b47174
c980aa074731bfbaa84b87d9e276f3c116229ebc
83884 F20101222_AABDTV albasheer_a_Page_025.jpg
10a8f1bf6e2724d0316e252f6f0ca01a
cd6c822fa49243f55749210386238f26f784a937
62570 F20101222_AABEAD albasheer_a_Page_125.jpg
b22c4219c4859abf7215d08537d9392c
6f787e06465fb910ac587afbcdf699c9c1716781
9617 F20101222_AABDUJ albasheer_a_Page_048thm.jpg
2d4e3c5464cd37473997a0904c137fa8
66aa59c51157879cdc679e0c64f5e760ac72231b
7695 F20101222_AABFDG albasheer_a_Page_134thm.jpg
dfe1e63aea03004c776c8807203531e9
c7cf587f181143091c71811e78e849a859771d5f
20468 F20101222_AABFCS albasheer_a_Page_125.QC.jpg
97c86a5c91c70e98a36271c9246af969
5fcd85a66fbb2d29044f119bbbbf2a20f41abaaf
10028 F20101222_AABEYA albasheer_a_Page_040thm.jpg
ebd80ae8e70df7045a0b0af10a62caf5
62fdc91ec981f486c78c2dab36e23796875ea857
25875 F20101222_AABEXM albasheer_a_Page_028.QC.jpg
8f8cf99c3fa64c24c3311ad3355d7b0b
e883b9b0c451e036616e855bd3541db807d32ca7
58633 F20101222_AABEAE albasheer_a_Page_126.jpg
f1b82f42d4631b73fc37ee8ad53d7b16
49c09fd1f09bc5ad9455c6c4bdd0318ccc37df09
1545 F20101222_AABDUK albasheer_a_Page_104.txt
3b112b1e380cfc720f8754a82d1e9c13
455d5c6f33c48d9b12ca82b3f414b484a66865f3
21773 F20101222_AABFDH albasheer_a_Page_135.QC.jpg
93987f66e9e2cfcc57d863d85d362768
b6ec70f180fbf67cb577e29beeb390b9206be5f5
6152 F20101222_AABFCT albasheer_a_Page_125thm.jpg
d57f0534b9a0323f24fe5347ff1eca9b
946d1dedefb86eaaacff1e05e8e71cd0f4465e4e
36299 F20101222_AABEYB albasheer_a_Page_041.QC.jpg
2043f2faa18048a2649e72c114ef361e
f18e628277f6f04fdf817d087a924270086e9308
7379 F20101222_AABEXN albasheer_a_Page_028thm.jpg
3c47dccc451a131bf4f227fd5a3eeaee
4894b27f0e3042c8981f89a85e68484666567712
32040 F20101222_AABEWY albasheer_a_Page_019.QC.jpg
114566100bb7ba40b9c54c42808df985
8d1fc3118e263d8ac04e01e031e348e43855116a
1051985 F20101222_AABDTW albasheer_a_Page_014.jp2
bc58c1cc6ab5df51f3204361e53d312e
83f6565ae17a937368c115da4812557f784241f2
37653 F20101222_AABEAF albasheer_a_Page_127.jpg
5cf950092337a24ec60145188c2e4fa1
7feb13cfb05b81d0fac5f6ee51501a1b6034fe37
70522 F20101222_AABDUL albasheer_a_Page_110.jpg
48c02b56e2c999ffa8c26f98f6f0e529
f2508e28b0292d5697f96f16d4e38510be47c9fb
36383 F20101222_AABFDI albasheer_a_Page_136.QC.jpg
041d1be34bd292aba8940b6c247ce80b
3c1c8f1506a3a92ff0af7a3fbbe1073af512fbbb
21020 F20101222_AABFCU albasheer_a_Page_126.QC.jpg
0e5bd078dbb4f1c1cc7d89496abe1b6f
149273c994643a795077c5d560e419906af5f4bc
9333 F20101222_AABEYC albasheer_a_Page_041thm.jpg
4851133104c2eb6bae61f33ac4e9a495
cda0b20606535a6063338d7b4a196bf84d92a108
23420 F20101222_AABEXO albasheer_a_Page_029.QC.jpg
1f20d1b32e97de154e2cac0a62bd80d5
ee51aba60c7a04c4b12a25741329dee8345686b7
8431 F20101222_AABEWZ albasheer_a_Page_019thm.jpg
2dfd7f1ddba1fb3a225b01ab0f8cbade
ff029ee8e8f1a91e673e61eb7c07811a45f8c165
F20101222_AABDTX albasheer_a_Page_058.txt
3a45d43b7ad5c031b9fee2aaa9d4819b
89cfeed51b227ba7c9a1d3e4af2beb3c9be0881c
54156 F20101222_AABEAG albasheer_a_Page_128.jpg
e558694ca248faeb202c6844218088db
b805d29a56ef2d76d8a1841d2966f874b59e0ac1
18405 F20101222_AABDVA albasheer_a_Page_084.QC.jpg
547b51951b488073f112b3655b99dddf
9cb8f22f564323300c116a1ce9624f36a77679c1
7970 F20101222_AABDUM albasheer_a_Page_021thm.jpg
48f1b3bbd4782d04007d8dc1b7cd9f23
f32bbb3bb1bebf90ff6b04b582387e44b0d8a8bc
9166 F20101222_AABFDJ albasheer_a_Page_136thm.jpg
f29e50fc252a049ea03abe4520919f5e
669bb8c12562f2a0948bb81c13be4c75edba99a3
6187 F20101222_AABFCV albasheer_a_Page_126thm.jpg
2cfd04ad9e5962495a755c9833d8b043
e670de81edc9d01617d4813f7279ad57a73e3e46
22842 F20101222_AABEYD albasheer_a_Page_042.QC.jpg
b0da5e80314bd29136f5d200157d57d7
61e58a26eaef00093690ff53e7378d6c28807176
28743 F20101222_AABEXP albasheer_a_Page_030.QC.jpg
af4445febd425e91d7a7497695f4fe50
7cd6b881f27509cd07c82595a4220797506c0ab4
8970 F20101222_AABDTY albasheer_a_Page_072thm.jpg
c8b59e5777506e831ac441da57fa7092
43e862ceda4aac91e834cfcc45222837c1e252ca
51483 F20101222_AABEAH albasheer_a_Page_129.jpg
2c2afb1626d0ee0705037c4d5edfc10b
64d167725910f2aeb479e24aa24c481f6d84cf5c
1051965 F20101222_AABDVB albasheer_a_Page_007.jp2
1f6d1817a97fee8257a736ca4c689088
5e8daf10db2ece2edc09eaa8bf7d20affa179691
2518 F20101222_AABDUN albasheer_a_Page_136.txt
30c2071916711f330c8ccf6abd1e87c1
1f112e959ff580aa2efedc834622265d550aab42
37763 F20101222_AABFDK albasheer_a_Page_137.QC.jpg
d9a8c8eba9bd801b5894d4bd4297de74
07ebd7f1e55fd7a640e6960fd998bd3eea1fa723
13308 F20101222_AABFCW albasheer_a_Page_127.QC.jpg
fd57111771fce2f991426f449fe33942
53bb6063595d8f8388d41f2c732f0825167d0897
7288 F20101222_AABEYE albasheer_a_Page_042thm.jpg
06fed8595d883d991831adcfd07083de
bbe1fa938c3d08f8d7a9708a3cb17a0cccb5a682
F20101222_AABEXQ albasheer_a_Page_030thm.jpg
5d47f0d1bc358b84f8aeadb746cc7582
95cbe07fdfeece54a9520078f1ff3407322cca06
2975 F20101222_AABDTZ albasheer_a_Page_009.txt
2ed34da73294b00de0741b1b1d99dacf
cd57eab76326b9bf7a92daf9e2ab345ce00b4f87
45160 F20101222_AABEAI albasheer_a_Page_130.jpg
3b7f7978669d7557fe532617f8f86c1d
a1e1cd513ce6f3618d8774931fbb52b8118e866c
8345 F20101222_AABDVC albasheer_a_Page_108thm.jpg
0be8c7509e84e61979a858277bcf181b
7d8fac87dfd5bc8bedef18074fc7612335a318e8
1973 F20101222_AABDUO albasheer_a_Page_061.txt
0d02f7ae0bb47a0497c1033a3071d3b3
96951b7b837a865c6f4e450f61d44b91fdc5d76b
9379 F20101222_AABFDL albasheer_a_Page_137thm.jpg
459cb56da79ad8125c940d987e7b7edb
e3c1608da68f391f7d1c65eeef5b60abae25c734
4065 F20101222_AABFCX albasheer_a_Page_127thm.jpg
2ecfa88cc5a13951babca6660ec1867d
7011a61c32c1b4a05cb2362a2133c8df81f6b715
22141 F20101222_AABEYF albasheer_a_Page_043.QC.jpg
24e1cfb648cc96bd0f27a49aa0d83f10
6642034d0f4a09705bd32e899c497606c2d483ef
22274 F20101222_AABEXR albasheer_a_Page_032.QC.jpg
7ad11c2e08d411f66e42b76f54df292e
867d101381bfe67bef3ed1c87a702c0f9808a733
95528 F20101222_AABEAJ albasheer_a_Page_131.jpg
93f9bb5d87e4cb28a932f0b419cf3144
27311cb034388b295171c970eba8d3a34c20cf8c
F20101222_AABDVD albasheer_a_Page_121.tif
44ad928e5bb7e00b78a736bdb766261e
5da577a62f47761e621a19ba293662e703eafc4f
F20101222_AABDUP albasheer_a_Page_052.tif
bd7e34236376433c5efc401d85a516e8
62a59bc84950c614b4e90a1380ed88dfc9771114
37880 F20101222_AABFDM albasheer_a_Page_138.QC.jpg
7d34b1ed9156a7098b6c95d8f3fe5641
cf7c5ebdb81885226e70399568612e88df839ca3
5105 F20101222_AABFCY albasheer_a_Page_128thm.jpg
72507db495486ff1d27217c95c93e0d0
6972bd4cbe01b9de90171e40f822fe37d1020de6
7210 F20101222_AABEYG albasheer_a_Page_043thm.jpg
a0dcab524c6e7a0adf79089d1946edb6
ab220a95f866684d94e108e596128bf1f3320f5b
8429 F20101222_AABEXS albasheer_a_Page_033thm.jpg
00330983f2b0d78faf16dff67cd1cf12
84aaf045997887a40e571bca030637700b22618d
30238 F20101222_AABEAK albasheer_a_Page_132.jpg
8b11cc61671a4de3baa0c4d2d0f625d8
5515ce8882090bb0edb5d84b1c99871bba8e6fbc
103639 F20101222_AABDVE albasheer_a_Page_037.jpg
98e5a9e1d1a57b0abf64722bdff4e91c
7c04b484587c104a22bf6e63f9fb6f431b19aded
1051986 F20101222_AABDUQ albasheer_a_Page_037.jp2
ce9ee793a50eec76aa26ca386bcb3fb1
0b70427df67c35643d67bfd49163f3dc557c2324
31820 F20101222_AABFDN albasheer_a_Page_139.QC.jpg
81a1105aba25e8eebc0872a04450fdc0
890a95242abf36a75551c339da485c1d428eeef4
19104 F20101222_AABFCZ albasheer_a_Page_129.QC.jpg
dd9cadc1e0624f647c38aed628efeee6
ea65f025d53e4e6211d914caa02639f02debbde2
31162 F20101222_AABEYH albasheer_a_Page_044.QC.jpg
88d64b554a781adeb8b954a20a2fbb34
2844e50373b6815157d30d8dc0c4e2f912f08725
8627 F20101222_AABEXT albasheer_a_Page_034thm.jpg
e017106057fd4c1f015afc1cb9ab1586
1c945ff6747c9bc5caac08310813072b260942ce
119894 F20101222_AABEAL albasheer_a_Page_133.jpg
4f9e4f79bcf7a00655fd5d033f8825fa
d34d2759b557abcd3bb043993e2f6eec3c1e6a01
F20101222_AABDVF albasheer_a_Page_122.tif
57e87af1c2a143ba89ad2eae67fe44fb
9861594cfa7d3e44e5919bab29f0ac5a962594e8
20069 F20101222_AABDUR albasheer_a_Page_078.QC.jpg
c0a17db87e96294859e56b6ca866a19f
2f6b30503893c711e3610d54c0692d82b4c0d3cd
1051978 F20101222_AABEBA albasheer_a_Page_010.jp2
4f107d8549f8bdb3a87c6d50cf16a7ab
dfffe5de47f14a5ac35a03fd9a61d3ee816f9370
31680 F20101222_AABFDO albasheer_a_Page_140.QC.jpg
db08d030386d82a75219549f2bcdabae
ac556792c66cd476e15d8081362fb504c77fc096
8623 F20101222_AABEYI albasheer_a_Page_045thm.jpg
1c0a62f9147b507eaca07803d451320d
09ec6498bf6b7eae65b0208e9fbfecfd14d28f16
39791 F20101222_AABEXU albasheer_a_Page_036.QC.jpg
c923916cab4cf602f6dd8bfad75d898a
0bcea9076f1850ea3164532063180e930a3b7675
154459 F20101222_AABEAM albasheer_a_Page_134.jpg
4b82be639beff645b0b6bcf063ed04ac
38818ab3424798cadb557ea07607f4e9072de695
8333 F20101222_AABDVG albasheer_a_Page_044thm.jpg
57f4e1837befa0b42bc8485bded53385
0a027f5055c724ac32b1a5a004f76cf7a8207cdd
10004 F20101222_AABDUS albasheer_a_Page_117thm.jpg
4998470d9422cae16773f5e08a0c6b17
c18c3384df424dd841004183c974df838637e350
1051962 F20101222_AABEBB albasheer_a_Page_011.jp2
6975f0483ad005650af0c086d1827399
36b7992ed248ce7cb61e1d85f74de60ec68dac17
8097 F20101222_AABFDP albasheer_a_Page_140thm.jpg
7875a548663415f4336eb6485447a3b5
36197b3242873b72ab28d69d83d0aaea91adcba2
25262 F20101222_AABEYJ albasheer_a_Page_046.QC.jpg
240af9b5448e6522de4729cc3df4bc57
7a1b74bdfa12f326bab7027518e7e49e3286b5ba
8629 F20101222_AABEXV albasheer_a_Page_037thm.jpg
52f2cf2fc89fb69ce04bc378a0e49c35
0e4a7b0ed599afe97b1b21a673251ff753811c74
79543 F20101222_AABEAN albasheer_a_Page_135.jpg
c29baefceaf001282bca323788ec114b
e2bbb99757b8c0108d2e59bfd51fb4118dc7323a
F20101222_AABDVH albasheer_a_Page_042.tif
231b15148e5d4a03e867498442428585
0e31af1700d0a87684e8e624e45d3ff379553ef8
95317 F20101222_AABDUT albasheer_a_Page_075.jpg
aa69c0bfe62867d5febaa9001d10ddbb
801350a3c36a59a40736784092e7bf0a6be9027a
1051983 F20101222_AABEBC albasheer_a_Page_012.jp2
34ee9c67142cb6f3c6ef6342a8c95444
0f0eac8c7a97cea4baedd0c007b1a528e24737cb
6533 F20101222_AABEYK albasheer_a_Page_046thm.jpg
26dd24ab4b6510f7e986c9b7ed7077fd
4864d437f30f431e4b5d4021ab707cdd53d0a130
29422 F20101222_AABEXW albasheer_a_Page_038.QC.jpg
a023221341893bedcc005c651ccdaaca
668084ba875f82d0ae06f38f01908b331456af41
128024 F20101222_AABEAO albasheer_a_Page_136.jpg
6e18deae9648faf298724ef54fb972aa
7cd24404d05e1613f281886f664423baa05ef1ec
31575 F20101222_AABDVI albasheer_a_Page_046.pro
beab3f6e733bc3df636cd042a9692b21
043a6f09ed3fef7bd1c8506f805d0b30dfc6da03
30762 F20101222_AABDUU albasheer_a_Page_062.QC.jpg
d67115f6882eec3a4ff3f534c795c931
eb928644b22a4662281e9f91ad58c7a20322176d
38460 F20101222_AABEYL albasheer_a_Page_048.QC.jpg
fb511486140657a5cf5b56992981fbfe
106c8452baeea00f7dc45f0e1c5f7408ec4738b0
8262 F20101222_AABEXX albasheer_a_Page_038thm.jpg
52d986fd0f7aba84ca2ce2225e76c5aa
41f892d83e9ea070720c50ec3a28cd8f22e17895
132216 F20101222_AABEAP albasheer_a_Page_137.jpg
f01af6398fdf6a50a54fd7c452f8bfeb
d0aa0dd9a8dc5e189a329a0a50dc5499525a1ccb
70019 F20101222_AABDVJ albasheer_a_Page_004.jpg
18358cbd6846df282c748a8cf513a199
dd5a737539b2199d2e7844a839f83210dba755e0
97880 F20101222_AABDUV albasheer_a_Page_031.jpg
f3e9adf413b7b04ce65711a16c29d70a
1e1812023bd4c5026a3880c3e128a3d9aa59fa18
116370 F20101222_AABEBD albasheer_a_Page_013.jp2
3cd941b3d28103e1612da04518b52c51
ecb1b52f4c3dc99819adb492ea42206d01f08682
35175 F20101222_AABEZA albasheer_a_Page_057.QC.jpg
91d21340dd18c5d65db4986818c468e2
5789b47a2d0d22a56257785b8b55a3b4beb61594
28979 F20101222_AABEYM albasheer_a_Page_049.QC.jpg
5b28338d727536bd0d38cd5f42bbfa64
93709a4bacdf191a430ee58114772f3cd252fcd8
9188 F20101222_AABEXY albasheer_a_Page_039thm.jpg
0c529696d61c85cdaa0da6d2b6186180
84090ec5e34dfceaf09e2ccc0557b4b3f49d78a8
131957 F20101222_AABEAQ albasheer_a_Page_138.jpg
8078d7c40f92969d57cae9e73315d04e
48118f4cac896390efd7f3e00701593d4f8b750f
25953 F20101222_AABDVK albasheer_a_Page_072.QC.jpg
4d2828ca5382e3eef819fc0711fc5f2b
354a474ebf3413347ce4d41f32e4d9c202517a24
109592 F20101222_AABDUW albasheer_a_Page_016.jpg
a0992a2980cec439ea3bec158dc5d905
e4670fa87552a78e996b0ae5b7911abb7453030b
580525 F20101222_AABEBE albasheer_a_Page_015.jp2
9405312290965cf5b16e02f98be97185
27a36c8baee951da6b091334f9e5559109c3652c
8645 F20101222_AABEZB albasheer_a_Page_057thm.jpg
32c4dddd225ec34ee6eb8ea08fa97ecd
41cdcc814858d25ae0220703afa97b3ea14f18cb
F20101222_AABEYN albasheer_a_Page_049thm.jpg
a452461c9c58c6834aa3839fcc43ee20
485a2d1295926bb86622c68892741fc69b5ab0b1
111612 F20101222_AABEAR albasheer_a_Page_139.jpg
9e88fc5cd048002b3531975859a440d8
b123197fe7d356e00181ca80a513e6eda0ece977
1051980 F20101222_AABDVL albasheer_a_Page_059.jp2
ba48d3f4c79e4a63f1410360a6dad5fb
801ccfc30a093b501e04ef61fe095c28be3ba730
F20101222_AABEBF albasheer_a_Page_016.jp2
c8f5fa3ea356b5b33ef494e49741495d
c9fec0ce7825c8ecb676ffb1f328271e852dba99
25371 F20101222_AABEZC albasheer_a_Page_058.QC.jpg
6a7eaf7c69fd2b4c79c4495c73915486
123794f5f4eedbedb89835f37d1c2b5eec3397f6
36197 F20101222_AABEYO albasheer_a_Page_050.QC.jpg
7be7daac6f0797febf1b75b1b005a6ea
c4d68f813a7c005cf60e36d2520b6c9ddd989c68
35663 F20101222_AABEXZ albasheer_a_Page_040.QC.jpg
7153e4192a056d88b320ee9ac40ac299
d6eccb1062cf0cb0465e22618a031ab1df483b2d
34097 F20101222_AABDWA albasheer_a_Page_037.QC.jpg
3f6d09b8d938660360afe7c75c4dfd70
a7f90ddb06aa8bda441de6f4fe9fee97a0a29a1f
97751 F20101222_AABEAS albasheer_a_Page_140.jpg
1ac1bb6a71796a455ac6fbd18903f7f5
c47c3af0fc976d869fa316d433e5c696a5acd843
113346 F20101222_AABDVM albasheer_a_Page_027.jpg
d843a35a4489db849362226a1f11be62
464da26299f8612ce999725d37c9bc8e6d26e0a9
127122 F20101222_AABDUX albasheer_a_Page_118.jpg
33bff6c127abcdc9f01115e35d33d2f3
f723e2653f08190c1524a4c70d97ec0d3bf3b914
1051963 F20101222_AABEBG albasheer_a_Page_017.jp2
c70b42c7194bc5c95531d91372a89d21
41d4c58997be0d5646a76c9118fee1871133c183
6718 F20101222_AABEZD albasheer_a_Page_058thm.jpg
4877d8777992bb976a74d2cd34ea62b6
d35daf9e93752a74bebaad7f64caddc231b1e2f0
9225 F20101222_AABEYP albasheer_a_Page_050thm.jpg
83bb10b0fa5412eb1e9a5ceaa892e05f
300e85c5496580b03120f7660ee972b5303347bc
48462 F20101222_AABDWB albasheer_a_Page_121.pro
eab4dfa6ec638325d54ee5a2af8c2ee8
87b0243310e145512d61b105c9dd3c4c1b8e1093
255404 F20101222_AABEAT albasheer_a_Page_001.jp2
e54fa3f38fe6346d65e8721c7387ff42
b0d5f9df9076ef4956b47c49ecc6c00ecbcfb984
64809 F20101222_AABDVN albasheer_a_Page_007.jpg
61f4f45e1952f1d26407b9faba07e247
fab183300416d0f7feef6e2abc2721cb941d7a0c
1284 F20101222_AABDUY albasheer_a_Page_125.txt
58d650483a5e52d9b40ae4cd473a6ca9
ded9a6421c8f8d9b60fb354f954330a789c89161
1051937 F20101222_AABEBH albasheer_a_Page_018.jp2
ac53f7c329d2eda3bd320376f79d982f
c77b694a59b2a539726e77ceae53106a4bce8df7
36119 F20101222_AABEZE albasheer_a_Page_059.QC.jpg
deeaec5586c37ff58281492813dcf581
5f20475579ed9c33158706b54b060888afa221cc
17278 F20101222_AABEYQ albasheer_a_Page_051.QC.jpg
f471e75186b0829fe84855f88f74391c
004d722a1e5a2de65d49317fb525cfd84eeb91fd
164163 F20101222_AABDWC UFE0022617_00001.mets
c4d05ad7e9851c8032e7fe60156fcfea
916567e036405e25e22507fb50510c1673a6ee08
27643 F20101222_AABEAU albasheer_a_Page_002.jp2
405018343c89e8f0ed6ad2756b35a239
a79fb9263bfe9181aa2650590804c4de1d11e8a9
F20101222_AABDVO albasheer_a_Page_076.tif
b880427c1b03d050b637f379c21955e6
811603e2709c87dda408de44864fe7f66661409f
1051984 F20101222_AABDUZ albasheer_a_Page_101.jp2
c0a0a27f9c41b09e963531a5abbda8d1
91715d2e3bde7311521c891b1d9ceba80651b5af
F20101222_AABEBI albasheer_a_Page_019.jp2
d34338dfe77c5bc9eaa6fde7ad4fed7a
8b8e7786a7bca059fde8cf51977fdeeed389f0c8
30495 F20101222_AABEZF albasheer_a_Page_060.QC.jpg
c087e8f598c9a6e9821bbf7b2a9c5915
291473cffae29c17a7820aba78979eba51af8fcf
6078 F20101222_AABEYR albasheer_a_Page_051thm.jpg
892dc206da0eb83bc25af19c53ed671b
26f63112b52dfa4079d24b9aecce883c57fa38a4
116160 F20101222_AABEAV albasheer_a_Page_003.jp2
fa2df6bb093078e4d44bc7e11c8b9230
693251299dfb0179c571487fbe61d85606e7917a
F20101222_AABDVP albasheer_a_Page_108.jp2
b6f0cc2621d391623b4ab5d7310c93a9
35754c8a1f50d09aebcdf0c123ad585348750f97
940083 F20101222_AABEBJ albasheer_a_Page_020.jp2
29d359b6afdbb7d149db7c3a2b68d46a
5cb1f0eff1830d5ece4d195c1cec34e6442b8a1a
9115 F20101222_AABEZG albasheer_a_Page_060thm.jpg
24963cb8449496de551f46b644506da1
7ab8db9b99a27260895562b691d8d04b69c12596
27593 F20101222_AABEYS albasheer_a_Page_052.QC.jpg
74eb8050070630164228e90627887068
c131f38c3e2ab61055c654c07075e9f753a7bc93
735917 F20101222_AABEAW albasheer_a_Page_004.jp2
29f4889636789835e6ef1e2758e3ebc4
0ab84f3529fc807698fecdd5ff53e3cf418f18c3
20710 F20101222_AABDVQ albasheer_a_Page_120.pro
749e99b6da3d3b8003786b4e6438327c
9a7c58a96fffa88fef78f9af665aa32a15b042de
1004071 F20101222_AABEBK albasheer_a_Page_021.jp2
0aff4c7aacb842c699d339973045ffb1
84e6c71cc53269d4c18226174366696a34e7eeb9
36600 F20101222_AABEZH albasheer_a_Page_061.QC.jpg
be5ee8857d81ae4e7cf1e22e200a0354
cc0edffcd62b91d9a47e3606932c3bd1b4f716f0
F20101222_AABEYT albasheer_a_Page_052thm.jpg
4ee8ebce3d82f5e5c1f23f29b6a5dd26
9a06771c2e51ba1aa0e9b492b22168875fd9ca35
26209 F20101222_AABDWF albasheer_a_Page_001.jpg
fbf5d152e5fbfb451dba4e0494b44b20
a912b86b12f81e657fa39961ba8ab23a7b67eea3
1051931 F20101222_AABEAX albasheer_a_Page_005.jp2
587b2f2dfd55e918f5254f82f035f112
e614c6f3f8ec5914e91ff7d62d070ec7e40fb45e
7940 F20101222_AABDVR albasheer_a_Page_035thm.jpg
06af60603a821b6a8dc64b1a485ed3fe
b0eafd605e288ff37dc504809f934fbe8e859876
1051977 F20101222_AABECA albasheer_a_Page_040.jp2
8027d1311e5d18bad563028ba65bc0fe
ec86e6af71543b2026ca76df4e18b675d23dcdbe
1014343 F20101222_AABEBL albasheer_a_Page_022.jp2
263e6e79f279b0ffb0ef3f02753511f0
ee3195b39718418c14423867cf1f5271ea1e859c
9035 F20101222_AABEZI albasheer_a_Page_061thm.jpg
9e8e3bffb768c9209b6b0d7f7d7b6ed4
422b156f8b26a5ec8a2db55bbce9bfebfacc8f3b
26192 F20101222_AABEYU albasheer_a_Page_053.QC.jpg
759c500b0523262e1f432f04b52d99f1
da1494f75e44581de264dc50f9b7fb3de308bbd2
4071 F20101222_AABDWG albasheer_a_Page_002.jpg
c81344f2f93a6bb970186c2475bb1948
7ced0bb2600dea130760ef5d3eda7a4e58bb6d03
F20101222_AABEAY albasheer_a_Page_008.jp2
796220c2fa6acfc350daa5b288bc1936
f098f5f86a958acee5a4b96839dc7f4b3917ade3
F20101222_AABDVS albasheer_a_Page_071.jp2
9d0946e9e76932f1e4e122e91674a9ea
0def0bf964b259f8740bb8e960f92e17b79fa1cb
F20101222_AABECB albasheer_a_Page_041.jp2
b9422615dee9a11381bb569c9280e0ce
43c734852110ea1977347c8723c631251453193d
F20101222_AABEBM albasheer_a_Page_024.jp2
4c7a13630e2a549ca72ab3346518d452
abb5cb08cf22ad192194a6785cecae7309c0e4a2
7665 F20101222_AABEZJ albasheer_a_Page_062thm.jpg
89a73af191599af753959708b1e6243a
bada1a5d2e45921051d97f74220b018727d690c6
7228 F20101222_AABEYV albasheer_a_Page_053thm.jpg
3abf9f259f48558f6ee098de0a7116f4
9a5b41e6c8b97f9051bd07339a0d9e23b34f169c
15813 F20101222_AABDWH albasheer_a_Page_003.jpg
ab9d51fcfee16dbbbdc8b6e62a75072f
c2892f9c51cb605d48c1c64d340e2d43481c9f82
F20101222_AABEAZ albasheer_a_Page_009.jp2
2b6b43ef00ce0a8963c41c9f61e8b290
9ee91d77b97a772226f764b4497fe30946cf5c2e
111490 F20101222_AABDVT albasheer_a_Page_122.jpg
5e9ed88dbc90232150b1567be8a5d205
b0830051f66765389c7dae8bcfb0c99aed2678a2
F20101222_AABECC albasheer_a_Page_042.jp2
26e9e4ba6ae20f42a056ca17845bf755
abb6ab9a32f961c1aec880d149d88fc375331e82
894504 F20101222_AABEBN albasheer_a_Page_025.jp2
3208eb24bd2a51d46cb9423e437d9d24
63a3586f09497ab4db8d37674bac46e79e6dfe12
36925 F20101222_AABEZK albasheer_a_Page_063.QC.jpg
6e0e41f6e17daf22a3dd1ae3b37ad7f9
3c6f8405cc3b68bc8806375e30e5973759c7eb64
36130 F20101222_AABEYW albasheer_a_Page_054.QC.jpg
ed81ffc54f4c4ac3103168c7456a935d
703d54e093d01e8de2abbe47b7a7931d41394882
117850 F20101222_AABDWI albasheer_a_Page_005.jpg
e250c8c91fde22630ea59c01c92ca2e1
bce26069c594534de538705ea897c38d2f9ce693
1599 F20101222_AABDVU albasheer_a_Page_124.txt
d4f4765fe0f5e9c7f82fcc46b94a72cf
e0046c848306afb2830b3e64e07c769d9a12b7e8
1016308 F20101222_AABECD albasheer_a_Page_043.jp2
5c87a712ecf2ab79c54ecff9b09f09e5
502deff1994489bd30f249b56701f5ffdf7c109e
1051939 F20101222_AABEBO albasheer_a_Page_026.jp2
8cade184f9ac4c68799307b85185fbe7
34193f3ace9827b1c80e026e3010c381492571c4
8894 F20101222_AABEZL albasheer_a_Page_063thm.jpg
4302395bd6619d0930531e3c56be0638
0fd23905d37b544a34c563170f49041e8e51c9b3
9354 F20101222_AABEYX albasheer_a_Page_054thm.jpg
648e096dd06c79ff323a0f6f049bfff0
219a3eb9b53e9ecd45d6240851a64244940e7a0e
162723 F20101222_AABDWJ albasheer_a_Page_006.jpg
e2d5974bb72b6c98e3ba4b44254b771f
c049036705023e40d39755d893f1ed2b58123de6
F20101222_AABDVV albasheer_a_Page_016.txt
47556f3f10def222b554064dfa798b8e
aa2dc68f8e4edc55b0f5118875b1a5f9ac5a0de9
1051880 F20101222_AABEBP albasheer_a_Page_028.jp2
09c374d9ea79e5f6604f87baa2886ee6
04512220ce1a1a69c183fae8dc586d231b892976
24643 F20101222_AABEZM albasheer_a_Page_064.QC.jpg
1efaabd546e2d5d26535c3985aa110bc
a0c9873270d3cda136a4db2af59a623b8b0f2739
39335 F20101222_AABEYY albasheer_a_Page_055.QC.jpg
6aa99017475e48da77c8dff4f81de766
fe88956ba86ae29eb3c024ac9c61e41c8d30c6ec
1051956 F20101222_AABDVW albasheer_a_Page_006.jp2
23d9b3d938b8bcfa385fa31294cf55fb
1de5bd5ebca3d727e3fb3978ab02c48fe250fac7
1030813 F20101222_AABECE albasheer_a_Page_044.jp2
146a0757dfabfd181a4a8c14183419ba
2ba998392c7143e33285b48c63ac32028855bc2f
714899 F20101222_AABEBQ albasheer_a_Page_029.jp2
d86ad145929efbc9574b2f07e20a904d
262f0312fa2aeaf4c5d0a5600e81c05d96797c4a
111415 F20101222_AABDWK albasheer_a_Page_008.jpg
e0284d6410ce1ecbe18ad9288f470a33
671e30071365a9653d6fb898a967871489a0c473
8129 F20101222_AABEZN albasheer_a_Page_065thm.jpg
9b52b32018f9b2e4a56aa888934343d2
1be8176540edf25e75723a0b9ab26172f24d12ce
9620 F20101222_AABEYZ albasheer_a_Page_055thm.jpg
6198aa57f35a68533682dbbc7de42397
02416ec60e8b6bf14c6f2293b9eba17e1bd6a900
96148 F20101222_AABDVX albasheer_a_Page_033.jpg
4cc701ea53e9993e0edd871e69dab5e1
09376cbd1842721360adf3a1db819119bd281af0
1051949 F20101222_AABECF albasheer_a_Page_045.jp2
a10055a6be228c1fa9172ea981b5aa89
3741b3a5366021562a2e1e74a25082e1ffdca7c8
958822 F20101222_AABEBR albasheer_a_Page_030.jp2
8b79264a9617488f74807938b6fb204a
535bc38e99e8d0330dd9aacf400bb3fa2b0f79cd
146411 F20101222_AABDWL albasheer_a_Page_009.jpg
58a5b9b43716b9711da6c8b71548bb32
6d4a3382962c378dac83918889170d11145a2a38
8625 F20101222_AABEZO albasheer_a_Page_066thm.jpg
39f9cf3eeaa3c7ad232c7926335eb970
0ebace2b5508fcf2430773786db9f1b3cb41e90d
756948 F20101222_AABECG albasheer_a_Page_046.jp2
442eb93f0fa349565a2ad90e5249cb51
d83b58b087c43eacaa4cf9d4aaa308101203fb98
85048 F20101222_AABDXA albasheer_a_Page_028.jpg
dd5e8f6f3d7b283966a13bf9c08c46e6
27c9c6a0c916ffc71e9d4665d95a0e3a045ec64e
1023734 F20101222_AABEBS albasheer_a_Page_031.jp2
530226009a47b2eb66bd3e3125822ec8
ed74164b79400a44e567e243176554125631402a
156386 F20101222_AABDWM albasheer_a_Page_010.jpg
3b270c06f43df76a7c29bddb034a9277
e9e0aa0c43ee7eba5662ce91cb934f56996a4911
33841 F20101222_AABEZP albasheer_a_Page_067.QC.jpg
062d30e118c4da3977c3c4f916dcf87a
15eb283d45390df371669bd17575dcf2ced9fe57
F20101222_AABDVY albasheer_a_Page_067.jp2
e66a9a2206854e0ef1312b2adb05d887
2cad8f2fcbcbfffab117c7642c64ae99ce494905
1051936 F20101222_AABECH albasheer_a_Page_047.jp2
6c5911c06da5371b8bb99790cf77fa45
bb45c35b63a1049a21cd9fd0a066bf06f28b5365
66833 F20101222_AABDXB albasheer_a_Page_029.jpg
88611dca2af23970c7e1fa07351eb671
f99e09a0c5fa322ef1e60fb4a39edcbf89bdbef2
696123 F20101222_AABEBT albasheer_a_Page_032.jp2
2f9145c418a789985a1ace309f3f080c
49586767f7641ce7c6caf6f828ccea357e7aa6d6
154825 F20101222_AABDWN albasheer_a_Page_011.jpg
6290b064dee1d59aa5327109d34b192f
a4ba8682e01a996dc6c7607836715dc8c5136b60
28158 F20101222_AABEZQ albasheer_a_Page_068.QC.jpg
523c6f5889aed750da29d771eaa930fa
f04784d57f8c889b221d5266b8c4defc210e68c0
F20101222_AABDVZ albasheer_a_Page_033.tif
f47761aef5a8faa993d07040d960e74a
892c7105213d8d868163c5f1bbc1c7277413cbe9
1051966 F20101222_AABECI albasheer_a_Page_048.jp2
f3a6c004b888c3e371ecc25d7030868f
69c05b7d4bb2483afce390af75fe83c3dc991cc9
89828 F20101222_AABDXC albasheer_a_Page_030.jpg
444be889b604adc0ea086a5aa0890f83
43377f1b3cf21491722e954a16133cad0a60719a
1051914 F20101222_AABEBU albasheer_a_Page_033.jp2
24665ca91bcadb868ae26bbecf8030e8
5a22cf3d3a95c3971f7a3337bac714b3c7850ee2
147497 F20101222_AABDWO albasheer_a_Page_012.jpg
699840e8b68fc3c915f534511a76d2ef
5e20184dd19a11d12776944b5820c8216e88c016
26732 F20101222_AABEZR albasheer_a_Page_069.QC.jpg
b0de941d68713cc617d38745021d8896
af8e54fbdb8c76ed7507fa9d903d5e623cd7133b
914226 F20101222_AABECJ albasheer_a_Page_049.jp2
678ae1ecdbd010b9feb2e3763e07bbe0
4fb73760e701f209301208e877c6f049896791b9
68913 F20101222_AABDXD albasheer_a_Page_032.jpg
131997921350d13cb7791d3f8eff0667
4b63345937d79ce581ab8d5c757713941212de18
F20101222_AABEBV albasheer_a_Page_034.jp2
dbdedf9a23a5a9dc2f5be7ff231de400
5a9d4882753f193d0011df76f540a1c921681642
103124 F20101222_AABDWP albasheer_a_Page_014.jpg
da8db9ab1989d730bc7882b2a72f537f
217fdaebbeb6edb3ceefcb2a0f53842ae486a233
7450 F20101222_AABEZS albasheer_a_Page_069thm.jpg
1fe9cc633e83ba4ffa050f9a2ef62627
b96baafdb2db8160406c62a90152279fc3c2c563
F20101222_AABECK albasheer_a_Page_050.jp2
f422d61fff370f0d5b99da7fb4a39668
4e6a6923f13fbe4687bf9c6a9bae400d6ab55123
99328 F20101222_AABDXE albasheer_a_Page_035.jpg
61a7cd28db0f91462adf6ced570bbe06
910a4151e3d1dd73875be4b409cdef0049998e27
1051948 F20101222_AABEBW albasheer_a_Page_035.jp2
e061e4acf9fbe5009b44a8110b257976
2bb5be671382ea98c25122aa3a7f9dcb1c5b3be0
55705 F20101222_AABDWQ albasheer_a_Page_015.jpg
1bb268548a34bb581225d0cd22d22934
bd7b1f8efbdbc1989b2873e476f2483305ea4cdb
27475 F20101222_AABEZT albasheer_a_Page_070.QC.jpg
5b29108f5f594eae48cacd1414e9efcd
06b5a2f1eda9aa713825896ea9f43bd153f605c1
F20101222_AABEDA albasheer_a_Page_068.jp2
e1d9c5c26c856a2c4b46f99b2b5075d5
0398a89b24c57b532da46f2e33447d8d7cec04e1
544690 F20101222_AABECL albasheer_a_Page_051.jp2
bf57e2fef91f389a76e0f468f9ebde1e
4a2c84cf32f3e40c72ce12dd152636a8286db014
121843 F20101222_AABDXF albasheer_a_Page_036.jpg
677a7ab06f4ce5df7a6483fb866586d4
65290d2c2e36a857cbe24b0a12dde17d8820a9de
1051969 F20101222_AABEBX albasheer_a_Page_036.jp2
0000966fb58df80fe4974bc07963ce39
3f4adebfa00756653a708fa2aedb83a8b6109a8b
115273 F20101222_AABDWR albasheer_a_Page_017.jpg
0dccf81d6c5d2d42b07d4b61b9a34548
a7b730f2dd51e4bfa5578132e6c6fd97f49a66d9
26237 F20101222_AABEZU albasheer_a_Page_071.QC.jpg
2d91548a146a245ea15ee7d305f80640
f7f62f440d0b6db0d0b92b274dbf417685e3d423
954149 F20101222_AABEDB albasheer_a_Page_069.jp2
c283ea8700152e640cc427b3673b6eb7
37910f7e88ac671c8a507b6bc4adba545d29a4af
916860 F20101222_AABECM albasheer_a_Page_052.jp2
380628fdb45627dd39020904ca10ac7f
d78e5ea7684280e82eac63eb8fd886b4afb9308a
89515 F20101222_AABDXG albasheer_a_Page_038.jpg
17f635ba8bec6f7d84a97d1ab3937f51
c6a5fd9e91fbed187fcbf490be48b7096240c934
F20101222_AABEBY albasheer_a_Page_038.jp2
d58dc50eb86f81daca34239f8b3a2ab8
38c45e946f4f9543aab7c28cc9397744524d444a
112845 F20101222_AABDWS albasheer_a_Page_018.jpg
ee3a23698309cf224c4aa3e876e22e92
9d4f9816da7b137f60ff3da625e6617db292a8c6
8831 F20101222_AABEZV albasheer_a_Page_071thm.jpg
c512046182f5320b514460b672aa6cdc
a12d52ec21ed3233508150d24f4f393bb2a35a0e
879716 F20101222_AABEDC albasheer_a_Page_070.jp2
5bdf37edf805af7defb62b2c61e165ae
f7df1751d51cd9440eb20dab50ecb7b2b25e5c93
810918 F20101222_AABECN albasheer_a_Page_053.jp2
3f974297c47617238afba5913f65799d
88586b2799e71b0b278b32e4a23f26294e0359a1
111054 F20101222_AABDXH albasheer_a_Page_039.jpg
6512a2bde81d374c145d397c77bf0d40
bbec59789d565336c58c36756804242bd0c674b0
1051899 F20101222_AABEBZ albasheer_a_Page_039.jp2
1c2e36c6c917374980c43c22a4303f47
e585acce3bcff8fe3ceebfeb9b853bf8d585f517
98311 F20101222_AABDWT albasheer_a_Page_019.jpg
effc4b775bf026b941496cfdb3cd94bb
f9c29a5fa603b3d30b4ebf32e9adff79dcd1c07c
28894 F20101222_AABEZW albasheer_a_Page_073.QC.jpg
6e9dbac53efeff23e16e7ceccc4a9a30
0e085cabad88fb1d81284ae8b11a5ed4e5de0537
1051952 F20101222_AABEDD albasheer_a_Page_072.jp2
db0d1487431b9b25bdef18e4a24d476c
57a9f5a505527cf3072422c17c0be093f9b0fcc6
F20101222_AABECO albasheer_a_Page_054.jp2
bd3ac83145462fc0a98508a1291265d5
4f70c1c38acc8a7a40b602257c2333b96d88c97a
107978 F20101222_AABDXI albasheer_a_Page_040.jpg
8d115bb4252ce03888232a875f0517bf
e1af444f5cb98a029f04d6265a7fa8f1bd64ea29
87802 F20101222_AABDWU albasheer_a_Page_020.jpg
712dd3668fba0a392d47c9b1574002ae
e9d77c7092643d7397d2d5fd01b152cb20f0d659
9978 F20101222_AABEZX albasheer_a_Page_073thm.jpg
d61488d03e5979915d5f2faa85d81855
79c855bb29a7e80102a0d1bb24e0d815d96b5f2a
1051973 F20101222_AABEDE albasheer_a_Page_073.jp2
794fdb089032ee2ecd84615c79ac0643
30a47b5d619ab8f127aeb3228dc6ec206ca6657e
1051976 F20101222_AABECP albasheer_a_Page_055.jp2
2c6af274db5db871ca13c6bb9a2d1500
db46b5a197d44c4958231288092686601c58748a
112118 F20101222_AABDXJ albasheer_a_Page_041.jpg
4a2024a10247d2d944a19da4a613455d
e85115c4fad724b36e7460e57294674a2e23a498
93637 F20101222_AABDWV albasheer_a_Page_021.jpg
0d288c3c99067d8bc200ddc30908fb8d
1b36b58ad768e725801ea756afef7315f14f25a8
10093 F20101222_AABEZY albasheer_a_Page_074thm.jpg
0c4b53906464a62fca5cffe4c8ebed26
a5b0cc1fac81d32690882258813447cd54897767
F20101222_AABECQ albasheer_a_Page_056.jp2
ffedcf7043b5fbe0d5edd40c9232cbfe
5965c8b4b4ffef97744bfc3540d30665eeb7da6f
74705 F20101222_AABDXK albasheer_a_Page_042.jpg
75d872b377ce23eca7c020c65c174f25
b5343c9d9395789a17ddbb795a01c79e9b9bf165
94969 F20101222_AABDWW albasheer_a_Page_022.jpg
be4347c46ada516b2f3b358b8d66bc39
d1aa98065381865c191df2e7ec232dbb3df24fd7
31070 F20101222_AABEZZ albasheer_a_Page_075.QC.jpg
80f81a6e0db9794e589e640dad755c7d
ad05719442cd892e1636b214c20d8b26b4271f6a
1051954 F20101222_AABEDF albasheer_a_Page_074.jp2
99e25059c61687c040c1d5678e54e8f9
c7b119fd5cfdea9af0a1a2e15534fd4589255db7
1051967 F20101222_AABECR albasheer_a_Page_057.jp2
fac18fb76178f6fcb91c96cdbbfccbe6
08970f539bc6e0fb692908286f79a098cc7c5b08
70047 F20101222_AABDXL albasheer_a_Page_043.jpg
90f6ee3e5a79658da02092a5470b4ce2
8ee1d3aab9433bd7008d1be8ced57025b73001c1
90352 F20101222_AABDWX albasheer_a_Page_023.jpg
8dcc896ec645ab8868b5f3c68c0da194
e96d9c85f987a60826ecac314f2ab6801c8529fd
1014948 F20101222_AABEDG albasheer_a_Page_075.jp2
7793f5281ca6277b50820e5a036ae529
7a48c3a03fc7695926adbd650a6b84d492af5d84
107207 F20101222_AABDYA albasheer_a_Page_061.jpg
a6ab541cfdfafe899e1a3e395d464a14
54cee05408c17d3fd032bde81ab8e46618a907df
863854 F20101222_AABECS albasheer_a_Page_058.jp2
f02e8f116b098c818f6c7d8f5c03feb1
a8b226c079fb4081628f37f9b12e6fb926b284d8
94416 F20101222_AABDXM albasheer_a_Page_044.jpg
857718d233976e902306b1c947909858
9423736bd8ea608f665c08576c1a9e427849840d
113769 F20101222_AABDWY albasheer_a_Page_024.jpg
2675eb88f84bd105550bc434865667f2
27b56ae0d1f594eb6c74bee760da8636ec7f4c2a
92542 F20101222_AABDYB albasheer_a_Page_062.jpg
670832d4a3efa710c1ee1c2b21877732
9b8328663e0c9adbf65d22904f0398ca053a4520
F20101222_AABECT albasheer_a_Page_060.jp2
539cc5b1831b14483acff7b02736a5f5
07c745a41218f513655d74f960d166a79c7d1050
101024 F20101222_AABDXN albasheer_a_Page_045.jpg
eda13a8d93d9b0a269c0c5423d175319
dc795893837b13b2495cc0ff2f46d6672bf94edc
1023666 F20101222_AABEDH albasheer_a_Page_076.jp2
a073a6f8ed8f7e58b615864c537bcba6
b011e560e66aa487a6f05e178f6615fa0d64b460
108196 F20101222_AABDYC albasheer_a_Page_063.jpg
47700ccd772664f5c30e284d5fb815ab
bf189c1445bec1ed12187ea81c5fa78b6854016d
1051947 F20101222_AABECU albasheer_a_Page_061.jp2
9e4228b3bebdd8c7f7e6a1faeb3471fc
fb10afef170c9869752fa5023adf45d94579e0c4
71417 F20101222_AABDXO albasheer_a_Page_046.jpg
e9afd0f0ff92768b911d6ba95ed94d68
330394baa4441ab8274dee7405d7c9b7bdede3b8
105644 F20101222_AABDWZ albasheer_a_Page_026.jpg
9799be7b8d0ac21284139791e7639f29
d85a0863f3ba3f4da9c734900dce4764237828e4
1051982 F20101222_AABEDI albasheer_a_Page_077.jp2
475c5f8ab3d52b9891526774788808a5
900893f9a7b22b0cb0052e0c35c902056d66f3df
F20101222_AABDYD albasheer_a_Page_064.jpg
d61ab3773c41ed1c593f086c6b4935ca
650222fbb303aa544630049b36999eb04fed2528
991325 F20101222_AABECV albasheer_a_Page_062.jp2
1aba0b4f5c88dee1f6e4ae2f8f64603b
8868d2e3d4aae6ce3812fdea1bfbe0009e3d783d
96822 F20101222_AABDXP albasheer_a_Page_047.jpg
8e7d179fb6ccd65903d82919c472de6d
67b096514307745847b9f9606708c9044c933d5d
650807 F20101222_AABEDJ albasheer_a_Page_078.jp2
23ec1e5b445a406af84ba498a47828b3
ea216acb54cfd16aada7da1cb4330064c14bc16d
97102 F20101222_AABDYE albasheer_a_Page_065.jpg
b99286ad55abe63265c77295ff00f675
dd8d1ef48e53e9196077f7b3b2fac7eaee158ae5
F20101222_AABECW albasheer_a_Page_063.jp2
a5edc46f0e615e356b0fac7ffefb4c7b
d218dc12cc2471688b3c7e7be33f4c5469a90d70
85292 F20101222_AABDXQ albasheer_a_Page_049.jpg
5f1f8e11d12baa25ad99e152221446a7
590e876890be540c319674b1d041da4b206fd480
1051934 F20101222_AABEDK albasheer_a_Page_080.jp2
b2af116aa9e898fdbb2d4a117681fb12
e1a5cd2b68b2dd064da071cb39db8972a5425575
100761 F20101222_AABDYF albasheer_a_Page_067.jpg
d3f2e52802614078351a70f276aa0337
f2dbf20b505754fda849dc04dca1043be2c18f74
761716 F20101222_AABECX albasheer_a_Page_064.jp2
bd59a6f81ac9797c5b9bbf7880a102e7
e2be31f84a3acf4663cc2c908a1ec3c1cf3ad424
111506 F20101222_AABDXR albasheer_a_Page_050.jpg
51ffb2ad4264b6973dc7716640d7503c
bd09bde429bc3f9f68ac46ed7a1185955d827588
F20101222_AABEEA albasheer_a_Page_096.jp2
fd43e2f8957eacd27fe4c87b8a478bf1
8e5c79b2e7f967d9d732822537d1ad6f3e4bade7
1051906 F20101222_AABEDL albasheer_a_Page_081.jp2
e1abf44e4205d44ad7f59bc640775954
d49de919c8615ee98a0399d39459ce8311a7fe8f
88648 F20101222_AABDYG albasheer_a_Page_068.jpg
4c954d587af663ff238573f38842f1de
3b7f2d36f0f859857cec36df86f11b066460d430
1042240 F20101222_AABECY albasheer_a_Page_065.jp2
8c33a7edd2956c2d5c69a6d0e0a80f57
5120e4308d9a97079632121464d47a035902385f
85776 F20101222_AABDXS albasheer_a_Page_052.jpg
f8e77bdc901be58ee5995e3ee95b8bcb
54249cffe6ec437add82c063b19f01219c6df5cd
F20101222_AABEEB albasheer_a_Page_098.jp2
9d584db6aec44d8244e32f27dd110179
ea74db6ce9fb11c1e90d090e08f7840a7898d4e9
835064 F20101222_AABEDM albasheer_a_Page_082.jp2
4c594e03bd30193b26b5de6e7bfbb69f
7ca61cbcc29653d7310b7a669adabcb054ac9f2a
86580 F20101222_AABDYH albasheer_a_Page_069.jpg
ec4b7998e82af43fdd7199d2ae6144d7
1fd23627f920f1e0c3c85775c02f051789e9ca31
1051842 F20101222_AABECZ albasheer_a_Page_066.jp2
c6ecba04e65e14a796054f33a28f2367
a559f786a1c38975a900427e502195615aed6d1b
80083 F20101222_AABDXT albasheer_a_Page_053.jpg
b8e92c9aa21e32dad2b65cd16e9f3292
ffb55923890ac3cbe3d7e8d0da5be0894cb84876
1008382 F20101222_AABEEC albasheer_a_Page_099.jp2
c302e04b9992a51f9787c70b2b6fdc12
695518efead8f483e87f0bd7e21e0133eb16a8ce
1051951 F20101222_AABEDN albasheer_a_Page_083.jp2
25ad436520fe3e70a7aa87965985b976
bcfb68df5ad748d195fc5225b4e4d8b2f977b41d
84981 F20101222_AABDYI albasheer_a_Page_070.jpg
f62a1648ffd842e7da5aca0a42cb96b7
0b98d94a0acdcbde5a87f6564e65fb35cff1c3a7
109714 F20101222_AABDXU albasheer_a_Page_054.jpg
402531fefca42bae07315e56d668550d
1040f641a36094a409121ebf6d5cdc5bf2a8512a
1051960 F20101222_AABEED albasheer_a_Page_100.jp2
89703354e4c345f4b925fca057e94a2d
05411f2ad91ee134a1bf823430e77f6726cb4ed7
716463 F20101222_AABEDO albasheer_a_Page_084.jp2
087d71ba37cc3c92668f4bbbb7478645
1922fd2e24a14788abcc643fc878a9718b6b4d01
82031 F20101222_AABDYJ albasheer_a_Page_071.jpg
1b30c0051c27c66851c2943cda88386d
3a30bc1a722a8e2bed527adb5f17d89b4d07a4ef
120752 F20101222_AABDXV albasheer_a_Page_055.jpg
ffe0f7985385974cbdc55b95ff39d636
012652681967787f420625b9ae78c19c02d96409
945969 F20101222_AABEEE albasheer_a_Page_102.jp2
e7e4a1c6cd3aeafee8968158cbfbba08
1027736492b0deeae65aa7cea9a4f4fccdc080a9
869483 F20101222_AABEDP albasheer_a_Page_085.jp2
7a68593703906297c395499d3d9ad5bd
001e0c04d0d7b16f9a4ff3beddb698f8824643be
73898 F20101222_AABDYK albasheer_a_Page_072.jpg
55a20b3b636557bf575f05491821f539
7bdd4ac1a075606afae7b37339b5b79378998026
103839 F20101222_AABDXW albasheer_a_Page_056.jpg
f1c2f9ee5d0aeb7447b3cc38eea83f3d
370c60fc7c7ad746d2007ed632a642abe27557db
1051946 F20101222_AABEEF albasheer_a_Page_103.jp2
4f0e3e82e66add8ca50e9fa82684934b
a27a76156a75e0c62d4d5cd607cb06f36119fc26
1004418 F20101222_AABEDQ albasheer_a_Page_086.jp2
cdef3bcd05ae505462def6fe7acb8cd1
5d1382f0fcaad1d7adf33fc907fb64f60cbdeabf
84479 F20101222_AABDYL albasheer_a_Page_073.jpg
e769a54361f54720479d924b1c352564
495df9b3c372a65ea13095c7d8490befecd383b6
108437 F20101222_AABDXX albasheer_a_Page_057.jpg
9a958250707276d4e461fced206cf5c2
aed4818fad5e7019525868dc27a0332e03238ecb
1036117 F20101222_AABEDR albasheer_a_Page_087.jp2
aafeb6a380c9029d4e5a147da8839cc8
13798c76fee1d23ba1a6d66cd60574674117b5c4
101707 F20101222_AABDYM albasheer_a_Page_074.jpg
61da90b6d5b1b2ee02cd4ee189cecb32
8cfbf17f21d4867e0cd371ac8fbd10a1c078c35f
74293 F20101222_AABDXY albasheer_a_Page_058.jpg
0a8644ae8ee61bf80ff6aa32a32e878b
5f9192773a53a61829584a722c0ad347defa7fb7
878495 F20101222_AABEEG albasheer_a_Page_104.jp2
3f253dcd7fe27d5a4ebf417e3638d594
b522f93f39c6685e4aab82a89892104aa4ada696
51095 F20101222_AABDZA albasheer_a_Page_092.jpg
ab1d36d0f44bcb10df1995a15a97df87
a3c618cb3ced14ba4d13cae296aaccf6e101c2bb
F20101222_AABEDS albasheer_a_Page_088.jp2
ab629a5df11df61e7397dcc0101e36a2
1dca3609a0e5353284cbde205652f3b4b377fcad
96600 F20101222_AABDYN albasheer_a_Page_076.jpg
79bd9351afa5bc329ae33e2ac91368cb
0218b67ae3113d0e38e88c1159d36080aeda124c
91568 F20101222_AABDXZ albasheer_a_Page_060.jpg
9625830a4eb002507101173f86971955
823aafc03251e9e2d943f91c9fed5e284a893744
F20101222_AABEEH albasheer_a_Page_105.jp2
97826cb5655b1947d1dfbea7a1ce2eeb
c4417a6b18069a3298bc19a17fe1e8c0b8058e73
96578 F20101222_AABDZB albasheer_a_Page_093.jpg
ea9d4edf02af2d3623855b4602e3fa10
f7cf19fdd1882f0da45c86bfe129c960c918b701
F20101222_AABEDT albasheer_a_Page_089.jp2
51b98a08e1b6eae6747a702ae1fa3e01
4076a02d825682ba8b025db376157105c3a39777
108562 F20101222_AABDYO albasheer_a_Page_077.jpg
bc87818506595260392aa8022dc376ff
0e0f79dc2a5ab5aa3b3524b390d26a62de859049
F20101222_AABEEI albasheer_a_Page_106.jp2
c31cfc3febb3a38ac090e146070e4af4
ea743a842e9bfe464f014d89d80399f913be6410
74397 F20101222_AABDZC albasheer_a_Page_094.jpg
dffb1ee86ce99f9598718b88bb00b156
f623ce309e19249e541f5f5452b522c72220b41d
945571 F20101222_AABEDU albasheer_a_Page_090.jp2
70a1db77ecf389def81183a4ed67cca2
86981454bbb550328d0e40137776108df6e8af33
117123 F20101222_AABDYP albasheer_a_Page_079.jpg
b82df5716af546b9ef19c1a7d480cbda
c54b840e3f755bed833ca40c8e1db0f701be961e
1010339 F20101222_AABEEJ albasheer_a_Page_107.jp2
76e96271f87fcb9f88cf22bb49b5bf15
2354ea2ecd895a55e2ac9539ca1c7f9d05c7cb53
85407 F20101222_AABDZD albasheer_a_Page_095.jpg
2cb18a1fc98d8dc5710461e0ff27cac6
3119574429d3e300a8ce1379a947fbe7884e7657
429824 F20101222_AABEDV albasheer_a_Page_091.jp2
2f1351ab70dd502efea4ecffad3ee2e8
295daa930654de059b83470855ac98ed9a17f2b7
116223 F20101222_AABDYQ albasheer_a_Page_080.jpg
b71f46addba6cef521b24d448f123e51
101ab42a0f9503f6193f377fde9ec3e021a0b221
F20101222_AABEEK albasheer_a_Page_109.jp2
427341ece3da7ec3d9e5aff5e86f1a4d
9d60c4976afd7d8394a6463c726b11dd0da32fb1
110059 F20101222_AABDZE albasheer_a_Page_096.jpg
04cf4adcefc097b4421041e484c6b687
1b03cef848dce097ec3fa1ce76e3b01c0e44892b
521495 F20101222_AABEDW albasheer_a_Page_092.jp2
687ca7dc65f42ccb631055e6df3034ce
554e8025ccbc89697a1b2bc2f2b31661d1a1d2f2
106675 F20101222_AABDYR albasheer_a_Page_081.jpg
8796cfec901336c5015c93a75102866f
38efa232c4fa2256374f2f1ae18cfb1f39c6fb0d
608509 F20101222_AABEFA albasheer_a_Page_126.jp2
7d79ab18efaaa9d3d2b750265878a116
33ad13e83f3565c122e9c7ec4c5979eefa839791
704278 F20101222_AABEEL albasheer_a_Page_110.jp2
6840817efbf60d573651fedc2364344a
14331c77913e7933338992a68c1caf72512ab57e
93337 F20101222_AABDZF albasheer_a_Page_097.jpg
7c3d2f7f55ae7b7c62d66ce8dd90c5fb
69d7563e7c7fd6a74023ad35df0db9af848bf4e0
1051935 F20101222_AABEDX albasheer_a_Page_093.jp2
bf89fdf1ffc8161380acb33a4bb5720b
db9387a5c8e182309a37b362d99c7b46ee42261a
77612 F20101222_AABDYS albasheer_a_Page_082.jpg
16bd71b22f26f08b8427a19ee5300f38
1d67bb4138085e848f034d94ba95ad5926670a05
366473 F20101222_AABEFB albasheer_a_Page_127.jp2
0cded4ae01df668f48497d5cb91a8bd9
fed8a128c7ccf640ebd8b22fc322a3894164f1d5
920882 F20101222_AABEEM albasheer_a_Page_111.jp2
a2582697ca5ae8848e8b0add78631bc8
5fde29355565d30f1e1e338d8ff22ae5c92df6b9
82165 F20101222_AABDZG albasheer_a_Page_098.jpg
c4cfe68476d0f3338cc4cdf747af7117
827aa14a05136586a759729f502ff20b534bd565
641762 F20101222_AABEDY albasheer_a_Page_094.jp2
d5f4faecd46c515ce1d1c2a72e626bbc
39ab209105971d3b7791f799e1e019c58c7bc396
115926 F20101222_AABDYT albasheer_a_Page_083.jpg
fdf2c4cfe043e1857ecdbcba71743fc6
e05b3b05a9f529280363ea6fc4a1f5123c4008ce
532123 F20101222_AABEFC albasheer_a_Page_128.jp2
565b17fc0290cf3121fe96c1bf00ecb6
af4af377d64807395424386af170d633cec3b1e0
805927 F20101222_AABEEN albasheer_a_Page_112.jp2
2f9d7437bbc127c4c5ac243a780afa9a
9cfddf2b24bea4ba0655ccb20aac4e8da73272a9
88672 F20101222_AABDZH albasheer_a_Page_099.jpg
88916cc0acb19f3a42530fbd3dc9db03
496a2f927a0517c3657155a8cf5dd0d62db85633
900522 F20101222_AABEDZ albasheer_a_Page_095.jp2
5357f1bfe787a95d843fe14b9fd1c92b
7b327517d4ad4057293f655a58e650cbfe883b98
59786 F20101222_AABDYU albasheer_a_Page_084.jpg
95cb49efcd9d2acf17df2c2c7e8143f0
0c88ec76d335e21e8d230a0975b2497bf4cb0b69
533601 F20101222_AABEFD albasheer_a_Page_129.jp2
d21c54a59f44e28e02a79e68f5341b03
313fd37bfba9d6be259fca90b9c78dbfe37fce22
F20101222_AABEEO albasheer_a_Page_113.jp2
f29670d71d46d8d7e43f03b37a2050d6
7535db5ef1b9baad2667649cb8b732436eacc223
113633 F20101222_AABDZI albasheer_a_Page_100.jpg
afbde8f1b7ce96cb507af21a9e4d0fc8
d9e8af9a9f6d00cc4d3f4d91cb4359759a3b0053
65206 F20101222_AABDYV albasheer_a_Page_085.jpg
35579b17b42dddebef91b8f88618e254
016af294815e000dce5444bc875250b8e2f96625
459704 F20101222_AABEFE albasheer_a_Page_130.jp2
881528816e31be55d61a0442d7e93e51
b19f33b9597821d381bad36fbada27476682394b
1051955 F20101222_AABEEP albasheer_a_Page_114.jp2
58082094263b705471b1294d42ff7ce7
3aa049a8cc4ff9ff9657d222ab17654d4d4d4fbd
106639 F20101222_AABDZJ albasheer_a_Page_101.jpg
a1fbbec1b5b236f33e028610e456b8a8
2efaaf0a4606f8eaac79e08ba9fe0033feef910c
92747 F20101222_AABDYW albasheer_a_Page_086.jpg
9a92fc62cab018888c4b453e4a42549b
25f3db8e3d3ca9fa4a279e30eebc75b5ac538306
1012888 F20101222_AABEFF albasheer_a_Page_131.jp2
8314a036f761448b6ccc1f72a70664a0
49b55f4c6a4fc1b1a77dec553e00f4a69aaf51ed
1051975 F20101222_AABEEQ albasheer_a_Page_115.jp2
9dae70afe6fbceaa6dcb975eb853c59f
fdb02f3caf5b82dd4f624e9e568418d9957a6343
87479 F20101222_AABDZK albasheer_a_Page_102.jpg
1cb6280feef7af3a65d94553f10a3ec6
f529f5e7f33dc47a8bbc8859ed2f9181d8e80686
76654 F20101222_AABDYX albasheer_a_Page_087.jpg
afcd23ec6afbf72f57caab53c1dce74b
8cfe690f51bdcf09dc69c00295f930897dedd09d
286623 F20101222_AABEFG albasheer_a_Page_132.jp2
a1bd8b092cf043cf3eab8b6999b7b8d5
d5629745f666ea751039694a62e3d2b3d547b7e0
1051971 F20101222_AABEER albasheer_a_Page_116.jp2
848f843812489daef47d1c6039c337e2
4b4d3337b504ce1005df22d2c649614bca6473c9
108110 F20101222_AABDZL albasheer_a_Page_103.jpg
6e7d39fcbf1ce16af8d448995d8b46b3
4106da4916a7b2d41e8f98c69a07167ada8e807a
98962 F20101222_AABDYY albasheer_a_Page_088.jpg
652fce798f89841a98c4082483ad6ce8
858b3e2138a9573ce9ce1c3f535573549e55f615
F20101222_AABEES albasheer_a_Page_117.jp2
830f7a761c639780c659c6ee4b043d16
54e6e9ea7ccd26181e10d7d18d79104a9cd14239
87964 F20101222_AABDZM albasheer_a_Page_104.jpg
82f44483719a08c94e0b9cb35fc438ff
9830da1ec21aa560ca84645a1dc6a193a7ed8417
46124 F20101222_AABDYZ albasheer_a_Page_091.jpg
020255cdda72d0cb035ff98ed77d8b35
dc415e1801625e85f3505fefbed4f0f8e2c713c5
1051981 F20101222_AABEFH albasheer_a_Page_133.jp2
e04ddcbe4245753c6b50c8d640db033e
4cc7485e38cd5d70c445cca065ba8c6cae9bfc00
F20101222_AABEET albasheer_a_Page_118.jp2
f269fd95da25e178978b1c53b8246853
8fb9bb9f2b2ecd017e88986bd1cb2a06408c5f83
97042 F20101222_AABDZN albasheer_a_Page_105.jpg
88a8912324bcba277c512f76394382b3
05408b6775ca9a1ae1160a725d4402aa09d37eb8
F20101222_AABEFI albasheer_a_Page_134.jp2
4d6e4f2e8ad7b7d571c51302ac88f8be
963844eb25efde98c23df1f8be10b4409df101e1
F20101222_AABEEU albasheer_a_Page_119.jp2
7f001c36b3d03b5b3380bb07138e9802
e8b4f381985ed111986cfb45946bf5baa27762fd
83173 F20101222_AABDZO albasheer_a_Page_106.jpg
ec1493fb6ce42a7ff28f58e85209ed61
d8fd4dfae385431cd8588c4b6b89c548199f008c
843674 F20101222_AABEFJ albasheer_a_Page_135.jp2
74ebdd600f59dd5d871d051203aaa965
5852dfbd9ad52b26b5cd7478b9e8d9b43fcdc480
478022 F20101222_AABEEV albasheer_a_Page_120.jp2
96a466cb7b256504ee3e2e2cc5266c69
ec21380a123d219e31cd1a08844a008c7c3f9b3e
88854 F20101222_AABDZP albasheer_a_Page_107.jpg
2cdfef2b89762c1fb738eb6c2d2d80fe
7a779fe166119b333fe5de54fb048780fc46c84b
F20101222_AABEFK albasheer_a_Page_136.jp2
966fa8eafe3b279ed5c3c278ea2b7f63
026fdfa8df00c3d0f709bc0c1db1a4d52afc27c1
1051902 F20101222_AABEEW albasheer_a_Page_122.jp2
e7212d7a4ca147af462ea3dc1f0fa084
437eb07a2599f631603595773f8788f6820651bc
101389 F20101222_AABDZQ albasheer_a_Page_108.jpg
f01d2648daf020bfa550e260c22edc5f
bdcfe686b986b070c60cd8ae188a87b269904fb8
F20101222_AABEFL albasheer_a_Page_137.jp2
cb3a8bfae7a2d043cc538e95df66b0dd
80ca0515f82eab26d9adb893d04cb91396cf5859
F20101222_AABEEX albasheer_a_Page_123.jp2
564c832f5768aa297f02bc2bc1da75e1
a36379d76e6d7669e30c8c40b95b0b66c887b250
97109 F20101222_AABDZR albasheer_a_Page_109.jpg
bd07792a8112b705957e6169f42ed671
f38ae8dac4d0420f0f71f75daf4029b80e914e23
F20101222_AABEGA albasheer_a_Page_013.tif
8eefc134c0e9e40508900e207ea6d907
fc86fbaa9da3cc4359c7ac59984d83deb1bb8233
F20101222_AABEFM albasheer_a_Page_138.jp2
e688240f38f4c0d8219b0dca0a123129
375d9f813eae4a297e0045a917f07f7d194b5eb9
1004152 F20101222_AABEEY albasheer_a_Page_124.jp2
40553c1d85562e68dd9e3d59971c4d07
5dc32aeef05118791aafbdd37e84d909eda3ee7c
92032 F20101222_AABDZS albasheer_a_Page_111.jpg
a1a1b631c5dbaae07514046be1fa1abc
b7f6baeeba401b78f26e38f17f0990891264635c
F20101222_AABEGB albasheer_a_Page_015.tif
32cbb31393bc008ae40ffbe182c2306f
a81279ffc5824eae1a9c7c79b7d8f2e3b0a966dd
1051905 F20101222_AABEFN albasheer_a_Page_139.jp2
ae72f0fc88f60ec145de21d935cf131b
7ec6629160e152767ec9c2487e43c9ebf2f54181
655052 F20101222_AABEEZ albasheer_a_Page_125.jp2
c78687c88cfbd7362fede8eeb3aa3271
1451af9f70bc5c67899ba2814e8bb26f6be120f4
105011 F20101222_AABDZT albasheer_a_Page_113.jpg
85d57175447898b33eb0d12a973106fa
628306c5cf3e637516947ca48b61f7109af2245e
F20101222_AABEGC albasheer_a_Page_016.tif
83c86f00bccf068d3ea9f22c95f1303f
7349fb7b20853a512250ae24468cdd84374e742f
1051913 F20101222_AABEFO albasheer_a_Page_140.jp2
fdc37bd8cb23f2c9def9a16c330d3c3b
d1825897fe12e5eba43e370ce2102b306bfae77f
110627 F20101222_AABDZU albasheer_a_Page_114.jpg
3c091bdbbb1c7a20faedb89bdf8dfed7
54f19910517ad55cb44baea0b65bfa3cff316bf2
F20101222_AABEGD albasheer_a_Page_017.tif
2fbd1626d58b57759083fb30cf974341
1e61c7fa9ff710c4a3d3a02439003f03f1aaee04
F20101222_AABEFP albasheer_a_Page_001.tif
2f1d4e71b968936164a2c334cf5901c9
5e41f947fc404a51080dd4a1cfa3063d5137e259
99830 F20101222_AABDZV albasheer_a_Page_115.jpg
589b5d76d3955c3ef9576aa5d24d1d4e
f57bfebb557c92f4b6cf01ff45a3c5a70080dc9a
F20101222_AABEGE albasheer_a_Page_018.tif
bd2f1db459cba41d59166237d4c516b0
0743b0498364221dafe0f110b4e5772780ce4e5f
F20101222_AABEFQ albasheer_a_Page_002.tif
04a674bbe7734f3dda74c1e9ab7c9b6b
bde7f8356e9bed031874c73848d32de824a4fad8
107888 F20101222_AABDZW albasheer_a_Page_116.jpg
9be56769e11e3e1fca7d6fa6ed2b1da4
6256c3fe375e6f689270fcb7c8c9857d599c79a0
F20101222_AABEGF albasheer_a_Page_019.tif
40f7c7e7df4be18d1c9f9286efcb44b6
1063362833a7e3885de742e519ea5fa9bcecf7ac
F20101222_AABEFR albasheer_a_Page_004.tif
001c486d799e0c51ace902216303f7fe
b036b0ee1780d902a108be867e584cba2184189c
118718 F20101222_AABDZX albasheer_a_Page_117.jpg
6e92291f828401a4845657715d3c1667
bb97d8d8b2f46e3a1e2787382dbfaabd02d697db
F20101222_AABEGG albasheer_a_Page_020.tif
adaa6c8c55c72e5d46754f7fff2c27fd
e5ffbe549e93dca19e35f87bb2bd42f1c74b284d
F20101222_AABEFS albasheer_a_Page_005.tif
0357225490c32f3695cf61e9fab5ff4a
953e6d8c8671715ad87ea523a20151d68e94acdd
109653 F20101222_AABDZY albasheer_a_Page_119.jpg
625d89362dcee03736e6cd3c58d21f98
3b27617d1ff6ba23daeaf217167b233bfa26fca8
F20101222_AABEGH albasheer_a_Page_021.tif
967949bd33d3ddfabd16d3d3ecb3f9f6
497654f31c0a731a459bc45578d3dd011c6a4d2a
F20101222_AABEFT albasheer_a_Page_006.tif
d010410542152feccbaf6626e6a28f72
ee1a7fc33beecc0967d9c820e58bcbb806df9ade
45846 F20101222_AABDZZ albasheer_a_Page_120.jpg
68ccd0a926c59984d8727c86b7726c42
4544e1867f09421ff68004bda27442c08faf186f
F20101222_AABEFU albasheer_a_Page_007.tif
e237acd29265da031320f66c09d5bb35
5e438b26a64dd8f230eeaf3f1e9f5992e326d506
F20101222_AABEGI albasheer_a_Page_023.tif
c8a4808f3b9eba87c424a41f18dfed85
4e5124b64852a2bbdfb25c87bbd2a2b54f5c0113
F20101222_AABEFV albasheer_a_Page_008.tif
c8aa59c34ffaf23138390ebec4db620e
6af60b9fa90819cf7c2416b6c3f1c8c37a193076
F20101222_AABEGJ albasheer_a_Page_024.tif
f95548b6c64952b0aeb0b87bc5331c56
3d21d8bbc2cf0273056a5c67cb84a529690141e8
F20101222_AABEFW albasheer_a_Page_009.tif
c49fc79515bcc651abe0890d83543df6
2a73ac8a5bd4bd58810dbcaba60de9e19529e98a
F20101222_AABEGK albasheer_a_Page_025.tif
872bab54c6704741597f0475a09cdf47
6d218318062c5780e1855be1d6a40863a3da796b
F20101222_AABEFX albasheer_a_Page_010.tif
37fca6c2734de51302d2924daaf095ee
a0c3b81edb7b65ec46c4860a16c8a3c50311d733
F20101222_AABEHA albasheer_a_Page_044.tif
2ed922a9621868136bebd092aff6a999
3bd1baf0f879775b2d0200a4ffac82c7b6d969ee
F20101222_AABEGL albasheer_a_Page_026.tif
afb7399cd68862d9ce91c1b0e2d95464
0d601f7c643dac15b5f459e16fa97ddb60c1b930
F20101222_AABEFY albasheer_a_Page_011.tif
6f24af4e487fb393dfc891f9059a93f0
b8766cda4616a111a83ef4372908dd40f997925d
F20101222_AABEHB albasheer_a_Page_045.tif
826b7d4be026e75c39dda88dbe064a3c
b03070aed540006b848f785c8c1ff658fd0c4423
F20101222_AABEGM albasheer_a_Page_028.tif
25725a28407670e5f6630f3b757eaf34
1432cca9d8b48e97041ff7ef685547c088d25913
F20101222_AABEFZ albasheer_a_Page_012.tif
8e10b7b8eb2c7410b214dc8f034ed324
d0627040b8d8159c32c64bc084a82a87f5749a1e
F20101222_AABEHC albasheer_a_Page_046.tif
7b879eedfbe76dfbb0051cd1cc96aa36
0e87959614295db501f2f43408a4d38627ff9732
F20101222_AABEGN albasheer_a_Page_029.tif
16fb728ec7037952cae04614d7cb694c
20069af746811b223a30e0d595ee53b660d4b44a
F20101222_AABEHD albasheer_a_Page_047.tif
6e476ba7ce2d3d4f80d9f47885ba0b59
222e46b74dd949ae2fdfaf136b27421d0f4f63f5
F20101222_AABEGO albasheer_a_Page_030.tif
493de7f38a7b205db8e15e835045f988
75bde7b93f6167138d2919bea1595425a7e96e06
F20101222_AABEHE albasheer_a_Page_048.tif
5af39f5e1a011b20b1cf2419e59d91cf
465f4f60a8f23389679f27fac982eea2d2e07417
F20101222_AABEGP albasheer_a_Page_031.tif
8d9b6b887813e70909b60bae2cdad843
1eb6321b85721355b8f2f72a084cf3ebb9a96a04
F20101222_AABEHF albasheer_a_Page_049.tif
8757c5507a7fb922d7ca8082bd6c9b8b
7bda64dbbe7102d80b727fa18d61aa1bd84d45d0
F20101222_AABEGQ albasheer_a_Page_032.tif
39ddaeaa4e791f894124ba805a492bff
76346fef1ed665d501d70ba0c7dcbcae5c9abdbc
F20101222_AABEHG albasheer_a_Page_050.tif
72ae42b9962a5e34fa4869d91eac6fdb
0fe97594c6f2652e06e1fa8acb880b93fd0e3413
F20101222_AABEGR albasheer_a_Page_034.tif
2b080075b71d5fa64cc6cf412bd7d9fa
a8d8f005b7ef97718057c0a6bebf51fc76560d3d
F20101222_AABEHH albasheer_a_Page_051.tif
ac5c789be891ff79b30fd8596c0460fc
1943af75292dcb9c4ff8446045a51fdb3db4d086
F20101222_AABEGS albasheer_a_Page_035.tif
acfd4248f31ea45d48a34da34594fa04
c37e3d6f33196f758fd07fdd0ac85707bd6d6cb5
F20101222_AABEHI albasheer_a_Page_053.tif
5f58ed437f64c150e1063bc9cfce12ab
b1c8a9cdb7f7b2b95f18f45857338931bdbade52
F20101222_AABEGT albasheer_a_Page_036.tif
f8a9c75087935e64409bb36b57c46211
9f82f96fa50aaf874e2168265f42a9a85e4fbe96
F20101222_AABEGU albasheer_a_Page_037.tif
abd65a7f649f6f291d794e86ed450f38
cbcc8fd18a7c8b6959521f1c8f8e50efea92bf20
F20101222_AABEHJ albasheer_a_Page_054.tif
3221f90dd92a14c4a64421c4211e544b
fda0da319b60aca11a410cba6c40a96081ef5690
F20101222_AABEGV albasheer_a_Page_038.tif
e1c46fe55e35fbf1899e7f8d593f1233
e57198c9996827d3f1055fc4e3d131cf06a3ed72
F20101222_AABEHK albasheer_a_Page_055.tif
9f266ae3739e0f83e91d5b4e2db5c9a6
6b07cc9b33466a6e3e528f6054433f0212d95177
F20101222_AABEGW albasheer_a_Page_039.tif
0b864cbf67f109267dc4af1e51c18eff
71d1c13013f09c6934d033badd0fb67484d0c855
F20101222_AABEIA albasheer_a_Page_072.tif
356ba9fc9f96f11ff6c9d25842ad65e1
5af4b893a958eb607c1b9fe195903a09ee67e6d3
F20101222_AABEHL albasheer_a_Page_057.tif
62a227d44b41b68a2209a853e7b16c7a
466e303666b0b6e035a65368e4bfc181deeb0574
F20101222_AABEGX albasheer_a_Page_040.tif
832045b68258d162c52f1ed755ce2840
a107960ec90646816c83aa5e1a0456313e7cf891
F20101222_AABEIB albasheer_a_Page_073.tif
f71d765d8acd4f12aff9e944271fe51b
7df50e82fd9e59707c7638c1b240908e9e1952ea
F20101222_AABEHM albasheer_a_Page_058.tif
145626c22f5755d9ff80a7196e01424a
34988fefe98ec68f8a9b67d2c1bc71009ba1c2d9
F20101222_AABEGY albasheer_a_Page_041.tif
2484e6634f6c73f77e3ab0f4631248b5
46ea604d47a558cbd449a3e6e393bb3a9097cd6b
F20101222_AABEIC albasheer_a_Page_074.tif
d9fd8a2ca2136de036bbd19a8e8bafaa
272541ab6721543f17ee4036c6755c5fb2c21555
F20101222_AABEHN albasheer_a_Page_059.tif
12b261507495ba6aa7228f9268fecff0
3e74ed6bb930bcb90007f74bc5a90ee251ff9ffb
F20101222_AABEGZ albasheer_a_Page_043.tif
e708a8cdc3063c18675497c9012a6a8c
13b357d0b7cb773baecf07e6f6ee065082f18d3f
F20101222_AABEID albasheer_a_Page_075.tif
2a3870bd67ff81c05b5575add704e1c6
d8dd702247a014ba8e13a12b67b2801ed1827169
F20101222_AABEHO albasheer_a_Page_060.tif
8f7f56773ba4ef9da71561aa58edab7d
c989750323e207354bc8b92146707be42dd0c4fe
F20101222_AABEIE albasheer_a_Page_077.tif
72491a35896d0858cdd18de35e7fde4f
bde788e51407ff7fbc640761fe26c098d09abf10
F20101222_AABEHP albasheer_a_Page_061.tif
b22c844514d89ea5a0cd264923dff137
defb3a15e787fda3d02a8785b6a3777c25ae089d
F20101222_AABEHQ albasheer_a_Page_062.tif
ef0dd1be81b4932c1b4b1f7b48892b9a
5251fb051d0bac415b9a9d2f7bfca676ede032fb
F20101222_AABEIF albasheer_a_Page_078.tif
bd07bded0044629d515603baaef7e1bb
cf686e441ba9155f2802b284cabe1ab66de151b6
F20101222_AABEHR albasheer_a_Page_063.tif
bb519f7f00e5894f4375b6d3dbb0ec4c
6cc817ceb0cfcfc2f52b34192cd48dd31797541c
F20101222_AABEIG albasheer_a_Page_079.tif
24fa7895d26cf3989689e0e9599617fe
3aba3c4b302159ea55322f36b560aab679b88b63
F20101222_AABEHS albasheer_a_Page_064.tif
d95b6d9aa3560260d0560a579055ec20
92cd40bc4f14a9b9a0d51c6af23dcf1552da887b
F20101222_AABEIH albasheer_a_Page_080.tif
991bad5d1a5e886d4886f2a906ca3133
e7404257a12bdc32a917f1c13b01481e58e52510
F20101222_AABEHT albasheer_a_Page_065.tif
edc3bdcbe68a8ebe276505f084177018
7b844016d9d412e2c875dd7fceaac5dbc8403a2f
F20101222_AABEII albasheer_a_Page_081.tif
eac3df808111623c86088f65cf788539
a0dd1a9f2e1b772d8dd1b30898f747720065dbfb
F20101222_AABEHU albasheer_a_Page_066.tif
638aad0d506a75918177952f3c6aea61
c7819228065543013cd7be986d1f09412c67f3e4
F20101222_AABEIJ albasheer_a_Page_082.tif
7c432bc6e5ef2ecfb1c4a34ad2eb4857
25050c6b79099fdf2abe7c70911c996c0093b5f5
F20101222_AABEHV albasheer_a_Page_067.tif
7623292d5e976ed74e2bdcb77766f121
68d954ff926847fe00d5c9e9c276110fc8dd5f59
F20101222_AABEHW albasheer_a_Page_068.tif
5e054fdaab28c6ed82874851b7febaa8
f7349cf9121d344a3fb7264b8fc3aefc72086391
F20101222_AABEIK albasheer_a_Page_083.tif
7f73a0eb25470fd23f3c201c5ee21bba
d5b6ee48e0ce6c22ac811557a3666e011f26ccd2
F20101222_AABEHX albasheer_a_Page_069.tif
d1942cb13fbee3cd65baf2a2398fd1a7
80fd7f4d84fcf5fe957f277cfa7d4d45b33c1d3d
F20101222_AABEJA albasheer_a_Page_100.tif
c5625e3234e63984bd8f499f831ef1fc
e385178261ebd84a39606068cf85d6c417f9b0f6
F20101222_AABEIL albasheer_a_Page_084.tif
f69a24dc1a33498ddc27f2361c28c357
f49ab22ad7443e1f6510893ed08d2782ef152dff
F20101222_AABEHY albasheer_a_Page_070.tif
3efd299f5df4db7fdaaff1747db8b014
df56b68324cde445b64e8000dff1903136bdab4a
F20101222_AABEJB albasheer_a_Page_101.tif
049f8c1033b169dd768c9f372a0a4354
cabfeb2a1b55ebb137d302f7902e260806fca4bf
F20101222_AABEIM albasheer_a_Page_085.tif
ff8cea9b3c809632b5710443f22aa4e8
43e21ef03bfef5016db7a6b7164e9093c7874e57
F20101222_AABEHZ albasheer_a_Page_071.tif
cca8a8119a22845847b85e3f3dabd6c3
68a34d8fe9accec431f5d45c3a5fd5a3d425d034
F20101222_AABEJC albasheer_a_Page_102.tif
53a9b1e45ed87b4377834131a08687a7
0d2494f3df96e6243807bf2e0b1c9907d8242502
F20101222_AABEIN albasheer_a_Page_086.tif
c80365909aa20aa6299dbea62e988a98
d1014991c965de5e1a0b0674934d862ea919232b
F20101222_AABEJD albasheer_a_Page_103.tif
cabf5698a7d247b29aa836d652f81b79
63da6c292af09690f2f1d1c711f73b0bb48d8ab5
F20101222_AABEIO albasheer_a_Page_087.tif
784e696d010c0f9d863e6e3a88edee4a
f0fd75b5b199798abe7e94edd622ef81fcdf2268
F20101222_AABEJE albasheer_a_Page_104.tif
069332b0b3f4f314c59183b97bb29479
289e105f4b27dc6499f6d0c15c4cd8f40fe5442c
F20101222_AABEIP albasheer_a_Page_088.tif
d022d416b244d3001a1c3f0d96e26399
0558aef12e753fed88fc0a46a626e789a9f98bf2
F20101222_AABEJF albasheer_a_Page_105.tif
477c13df66c537a2ed286dd2d305e125
5e1b669b5d5e8539ca67fdecda07144ba2906202
F20101222_AABEIQ albasheer_a_Page_089.tif
072dc01268dec7c41ecae80dd4034751
301a0c4aab977ea3cf101095b5ef7f35064bfe5d
F20101222_AABEJG albasheer_a_Page_106.tif
bc652ec1732cc28249fa4245ad31533e
52b67c5d39c1a730a5691d092a107dd70f4b1fe5
F20101222_AABEIR albasheer_a_Page_091.tif
d349ceadea6a670e6badfc75fd9db100
45c28d34c08f32d6bd62593c8d088dea67277724
F20101222_AABEJH albasheer_a_Page_108.tif
2c5a8993e29b6b25de128b4e0458e58f
5cfe20704731eea93ebf7c2f9cc9f8e9d1b80ba7
F20101222_AABEIS albasheer_a_Page_092.tif
e145690f7e5349924fdaebd51fb45f0a
812f13e66a5fd98bfd56d46cbc2e97352b877734
F20101222_AABEJI albasheer_a_Page_109.tif
08ddc88f5f3a5c14c4b30eed33727307
135d84bcbf8c5de3251caf31636046ce58daa3dc
F20101222_AABEIT albasheer_a_Page_093.tif
1a10872eb63054704ff6c871e286632f
df25f270e901efd24122c38869bd3b5eef741399
F20101222_AABEJJ albasheer_a_Page_110.tif
69e62f45cc93d88c25a31fb0c23474fc
a4290bbe59f6be532efcce3451f9df255eed5f6e
F20101222_AABEIU albasheer_a_Page_094.tif
720cef13e2e912d22de9c60f22c2d5ce
ca53ed62e4f5a40c18a4b3b365666795eda3f4b6
F20101222_AABEJK albasheer_a_Page_111.tif
dd73e49d0f6f4d98d139ff6299e8d606
7fbf58ce02c42985604e9ce5531430050b6a34c4
F20101222_AABEIV albasheer_a_Page_095.tif
f740bb4ef9abcc0cb481c3c1ec55e96b
80ee956ecce0f6b5b71b16f3d87446c61496fdd9
F20101222_AABEIW albasheer_a_Page_096.tif
a8aca9d51cd251b7c071696e070abd0e
5dd5f62ab4f5430e5dfbd5638fd3185c08077251