A physical anthropomorphic phantom of a one-year-old child with real-time dosimetry

MISSING IMAGE

Material Information

Title:
A physical anthropomorphic phantom of a one-year-old child with real-time dosimetry
Physical Description:
xv, 207 leaves : ill. ; 29 cm.
Language:
English
Creator:
Bower, Mark William
Publication Date:

Subjects

Subjects / Keywords:
Nuclear and Radiological Engineering thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Nuclear and Radiological Engineering -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1997.
Bibliography:
Includes bibliographical references (leaves 201-206).
Statement of Responsibility:
by Mark William Bower.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 028623448
oclc - 38863616
System ID:
AA00014246:00001


This item is only available as the following downloads:


Full Text











A PHYSICAL ANTHROPOMORPHIC PHANTOM
OF A ONE-YEAR-OLD CHILD WITH
REAL-TIME DOSIMETRY












By

MARK WILLIAM BOWER


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































Copyright 1997

by

Mark W. Bower






























This dissertation is dedicated to my daughter Anastasia Marie Bower and my wife
Janet who have sacrificed and supported me in this effort and my entire career. I truly
could not have done it without them and their support.














ACKNOWLEDGMENTS


The author would like to express sincere gratitude to his major advisor, Dr. David

E. Hintenlang, for his confidence and enthusiasm throughout the entire graduate program

and particularly in this research. His friendship, confidence and mentoring style were

critical to the successful completion of this work and related technical papers.

The author would also like to thank his supervisory committee members, Dr.

Wesley E. Bolch, Dr. W. Emmett Bolch, Dr. William S. Properzio, and Dr. Jonathan L.

Williams, for their ample guidance and suggestions related to the research and resulting

technical papers which have been submitted for publication.

Special thanks go to the "Pediatric X-ray" group, consisting of Dr. Hintenlang, Dr.

Bolch, Dr. Bolch, Dr. Williams, Ricardo Reyes, Lionel Bouchet, Kathleen M. Hintenlang,

and Kennita Johnson, for their friendships as well as stimulating discussions and

suggestions.

The author wishes to thank his wife, Janet Bower, for her unconditional support

and devotion throughout this period. Finally, a special thanks to his daughter, Anastasia

Bower, who was forced to share quality time and the attention of her father with this

research. Without the love and support of these two individuals the author would not

have had the desire and stamina to finish this work.















TABLE OF CONTENTS


age


ACKNOW LEDGM ENTS .... .................................................. ............... iv

LIST OF TABLES .................................................................................................. viii

LIST OF FIGURES........................................................................... ......................... x

ABSTRACT......................................................................................................... xiv

CHAPTERS

1 INTRODUCTION AND BACKGROUND.......................... ...................... 1

General ............................................................. ............................... ..................... 1
Phantoms............................................... 4
Detectors........................................................ ............................................
Gas Filled Detectors............... .. .......... ..... ...................... 9
Thermoluminescent Detectors ......................................................... 10
Diode Detectors.................. .................. ........................ 10
Plastic Scintillators............................................................... .. 13
M OSFET Detectors................................................ .................... .. 14
Rationale .................................................... ...................................................... 15
Pediatric Focus........................ ..... ............................... ................... 15
Ease of Use..................................................................................... ............. 16
Age..................................................................................... .. ................. 17
Effective Dose............................................................................. .................. 17
Approach.................................................................................... ................... 18
Detectors ........................................ ...... ................. ................ 18
M material Selection ............................................................................................. 18
Phantom Construction.................................... .................... .. 19

2 M ATERIALS AND M ETHODS................................................. ...................21

General............................... ...................................................................... 21
M OSFET Dosimetry System ................................ ...........................22
Introduction ........................................ ................................................ 22









Physical C characteristics ...................................................................................... 25
Sensitivity ................. ............................... ......................... 27
Linearity............................ ........................................ 32
A ngular R response ........................................................ .. .................... 34
Post-exposure Response....................................................... 37
Measurement of Absorbed Dose...................... .... ......................... 42
Tissue Substitute Development ................................. ............................45
Tissue-equivalent Material Fabrication .................................................. 45
Verification of Tissue-equivalency......................................... 48
Soft Tissue Substitute ................................... .................................. 49
Lung Tissue Substitute........................... ........ .............................. 51
Bone Tissue Substitute....................................................... 53
Phantom Construction.............................................................. .................... 56
Phase 1 Cylindrical Phantoms........................... ........................ 56
Phase 2 Composite Trunk Phantom................................. .............. ............ 58
Monte Carlo Simulation Studies............................................... ....................64
Phase 3 Heterogeneous One-Year-Old Phantom...................... ..................... 66
Skeleton ............................................... ...... .................. 66
Lungs....................................................... 76
Head ........................................................79
Outside trunk and legs..........................................................................88
MOSFET locations in Phase 3 heterogeneous one-year-old phantom............91

3 RESULTS AND DISCUSSION OF MOSFET CHARACTERIZATION.................96

4 RESULTS AND DISCUSSION OF PHANTOM CONSTRUCTION.................... 100

Trunk Component Masses............................. ................................ 100
H ead C om ponent M asses...................................................................................... 102

5 RESULTS AND DISCUSSION ASSOCIATED WITH THE PHASE 1
P H A N T O M ........................................................................... .......................... 103

Soft Tissue Phase 1 Cylindrical Phantoms .............................. .......................... 103
ST E S 1 Substitute................................................................... .................... 103
STES2 Substitute......................................... .............................................. 105
Lung Tissue Phase 1 Cylindrical Phantoms.......................... .... ........ ...... .... 107
LTES Substitute.............................. ................................................... 107
LTES2 Substitute.................... ... ................................................ 109
B T E S Substitute .................................................................... .......................... 110
Summary Phase 1 Phantom Comments................................................................. 111

6 RESULTS AND DISCUSSION ASSOCIATED WITH THE PHASE 2
PH AN TOM ........................................ ........................................................... 112

G general ................ ... .................................... ........................................... 112









A P V iew ................................................................................... ..................... 1 13
P A V iew ............. ................................................... ........................................ 118
L ateral V iew ......................................................................... ........................... 122

7 RESULTS AND DISCUSSION ASSOCIATED WITH THE PHASE 3
P H A N T O M .......................................................................... ......................... 126

8 C O N C LU SIO N S .................................................................. ........................... 131

MOSFET Dosimetry System............................................. .... 131
Phantom C onstruction.................................................................. .............. 132
O rgan D oses.................................................. ................... ................... 133

APPENDICIES

A MATERIALS ................................................ 135

Detector Cost Comparison................................ .......... ... 135
Tissue Substitute Component Sources............................................... 136

B SO FTW A R E ..................... ........ ................................................................ 137

Softw are W edge ..................................................................... ......................... 137
X C O M ............................................................................... .... ...................... 140
XCOMP5R............................. ........................................ 144

C RADIATION COEFFICIENTS........................................ 145

Soft T issue.............................................................................. ... ................ 145
Total Photon Attenuation Coefficient ........................................ 145
Mass Energy Absorption Coefficients............................................................. 146
B one Tissue .............................................. ................................................. 147
Total Photon Attenuation Coefficient ......................................................... 147
Mass Energy Absorption Coefficients................... ... ........................ 148
Lung Tissue ......................................... ........................................... .............. 149
Total Photon Attenuation Coefficient .................................... .................. 149
Mass Energy Absorption Coefficients.......................................................... 150

D CROSS SECTIONAL SLICES OF THE PHASE 3 HETEROGENEOUS
ONE-YEAR-OLD PHANTOM ............................................ .......................... 150

E ORGAN AND EFFECTIVE DOSES UTILIZING THE PHASE 3
PH A N T O M ....................................................................... ...................... 165

LIST OF REFERENCES................. ........................................... ...................201

BIOGRAPHICAL SKETCH ........................................................... ................... 207
















LIST OF TABLES


Table W

1. Sensitivity of Therapy and High-sensitivity MOSFET Dosimeters at Diagnostic
Energy Levels.................................................... ....... ............ 30

2. Elemental Composition of the Soft Tissues and Soft Tissue Substitute, STESI,
by Percent M ass .................................. ..... ....... ................. ..................... 50

3. Elemental Composition of the Lung Tissue Substitutes LTES2 and LTES1 by
Percent M ass......................................... ................... ..... ....... .............. 52

4. Tissue Composition and Reference Masses (g) of Skeletal Tissues in the
One-year-old ................................ .... ........... ............... ........................... 54

5. Anatomic Compositions of Skeletal Tissue (Percent Mass)...................................... 55

6. Elemental Compositions of Skeletal Tissue (Percent Mass).................................... 56

7. Dimensions and Volumes of the Cristy & Eckerman Mathematical Phantom. ............61

8. Dimensions and Volumes of the Phase 2 Composite Trunk Phantom.........................61

9. Positioning of the MOSFET Dosimeters within the Phase 2 Composite Trunk
P hantom ....................................................................... ................... ......... 64

10. Effective Dose Tissue Weighting Factors (ICRP, 1991) ......................................... 91

11. Positioning of the MOSFET Dosimeters within the Phase 3 Heterogeneous
One-Year-Old Phantom ............................ ................ .................... 93

12. Active marrow in individual bones, parts of bones, or bone groups expressed as
the percentage of active marrow in the body (Cristy, 1981) .................................. 94

13. Percentage Active Bone Marrow Associated with the Bone MOSFET Locations.... 95

14. Sensitivity Calibration Factors (R/mV)...................................... .............. 97

15. Post Exposure Response................................. ............................98









16 Conversion Factor Converting Exposure to Absorbed Dose.................................... 99

17. Mass of the Individual Components for the Phase 3 Heterogeneous
One-Year-Old Phantom........................................ .................................... 101

18. Effective Doses for Male and Female One-Year-Olds .......................................... 130















LIST OF FIGURES


Figure page

1. Relative Angular Response of a Diode Detector Designed for Diagnostic X-ray
M monitoring .............................................. .................................................... 11

2. Response Comparison Between Diode Detectors and an MDH Ion Chamber............ 12

3. Thomson and Nielsen Electronics LTD. MOSFET Dosimetry System.....................26

4. Sensitivity of MOSFET System with Low Total Absorbed Dose................................ 31

5. Sensitivity of MOSFET System with Substantial Total Absorbed Dose...................... 31

6. Linearity of MOSFET System at 60 kVp .......................................................... 33

7. Linearity of MOSFET System at 90 kVp ................. ....... .............33

8. Linearity of MOSFET System at 120 kVp..................... ..................... ................. 34

9. Relative Angular Response of MOSFET Dosimeters, Free-in-air.............................. 35

10. Relative Angular Response of Therapy MOSFET Dosimeters within a Tissue
Equivalent Phase 1 Cylindrical Phantom...................... .............. ............... 37

11. Relative Angular Response of High-sensitivity MOSFET within a Tissue
Equivalent Phase 1 Cylindrical Phantom............................ ......................... 38

12. Long Term Post-exposure Response of a MOSFET Dosimeter................................ 40

13. Mid Term Post-exposure Response of a MOSFET Dosimeter................................41

14. Short Term "Realistic" Post-exposure Response of a MOSFET Dosimeter..............42

15. M multiple Query Response........................... ..... ................................ 43

16. Partially Completed Phase 2 Composite Trunk Phantom with Lung and Spine
Cylinders Exposed ..................... ................... ............. 59

17. Side View of the Phase 2 Composite Trunk Phantom............................................. 62









18. Top View of the Phase 2 Composite Trunk Phantom ....................... ............ 63

19. Physical Skeleton Model with 4 Ribs and a Wood and Plaster Cast of the Head. ......67

20. Arm Bones with a Plaster Type Mold............................... ........................... 70

21. Pelvis with M old........................ ................ ............... .................... 72

22. Clavicles with Partial Mold .................... ...................... .......73

23. Scapula and M old .................................................................... ................... 74

24. Ribs with Partial Mold Still Containing Some Ribs............................ ............ 76

25. Left Lung with Simple Mold and Wooden Spacer..................... ................... 77

26. Lungs and Lung Molds, Left Lung has a Larger Section Removed........................78

27. Bouchet-Bolch Head Mathematical Head Model............................ ............ 80

28. Lower Head Mold with Wood and Plaster Model of Head .................................... 81

29. Top Half of Cranium Mold in Two Parts.......................................................... 83

30. Upper Face Region with M old ........................................ ............................ 84

31. Teeth and Accompanying Mold............................... ........................... 85

32. Mandible with Associated Mold.................. ........................................................ 86

33. Front of Head Region with the Top Half of Cranium...................... ............ 87

34. Phase 3 Heterogeneous One-Year-Old Phantom Trunk, Partially Completed,
with Most Internal Features Shown............................. ................... 89

35. Legs of Phase 3 Heterogeneous One-Year-Old Phantom....................................... 90

36. Absorbed Dose per Unit Exposure Comparison Between MOSFET Measurement
and Monte Carlo Simulation for STESI Soft Tissue Substitute........................... 104

37. Absorbed Dose per Unit Exposure Comparison Between MOSFET Measurement
and the Original Monte Carlo Simulation for STES2 Soft Tissue Substitute.......... 106

38. Absorbed Dose per Unit Exposure Comparison Between MOSFET Measurement
and the Second Monte Carlo Simulation for STES2 Soft Tissue Substitute........... 107

39. Absorbed Dose per Unit Exposure Comparison Between MOSFET Measurement
and the Monte Carlo Simulation for LTES1 Lung Tissue Substitute................... 108









40. Absorbed Dose per Unit Exposure Comparison Between MOSFET Measurement
and the Monte Carlo Simulation for LTES2 Lung Tissue Substitute..................... 109

41. Absorbed Dose per Unit Exposure Comparison Between MOSFET Measurement
and the Monte Carlo Simulation for BTES Bone Tissue Substitute..................... 110

42. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 60 kVp Anterior to Posterior View ........ 115

43. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 90 kVp Anterior to Posterior View ........ 116

44. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 120 kVp Anterior to Posterior View....... 117

45. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 60 kVp Posterior to Anterior View ........ 119

46. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 90 kVp Posterior to Anterior View ........ 120

47. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 120 kVp Posterior to Anterior View....... 121

48. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 60 kVp Left to Right Lateral View......... 123

49. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 90 kVp Left to Right Lateral View......... 124

50. Comparison of Absorbed Dose per Unit Exposure Measurements in the Phase 2
Phantom to Monte Carlo Simulations at 120 kVp Left to Right Lateral View....... 125

D-l. Cross Sectional View of the Head at Z=44.06 cm, to Scale............................... 153

D-2. Cross Sectional View of the Upper Face Region, to Scale................................. 154

D-3. Cross Sectional View of the Teeth, to Scale...................................... .... 154

D-4. Cross Sectional View of Lower Head at Z = 32.91 with Spine and Mandible,
to Scale........................................ .................. .................. 155

D-5. Cross Sectional View of the Neck with Upper Spine, to Scale........................... 156

D-6. Cross Sectional Views of the Arm Bones, to Scale......................................... 157

D-7. Cross Sectional View of the Scapulae with Cutting Planes, to Scale.................. 158









D-8. Clavicles Axis, the Center of the Taurus is Shown Between the Two Cutting
Planes, to Scale....................................................... ................................. 159

D-9. Cross Sectional View of the Lungs Illustrating the Removed Section Locations
and the Spine, at Z = 26 cm, to Scale................................. .............. 160

D-10. Cross Sectional View of the Pelvis with Cutting Planes, to Scale..................... 161

D- 1. Cross Sectional View of the Trunk, at Z = 30.26, Illustrating the Top of the
Arm Bones, Ribs, and Spine, to Scale.......................................... ................... 162

D-12. Cross Sectional View of the Trunk, at Z = 0, Illustrating the Bottom of the
Arm Bones and Pelvis, to Scale.......................................................... 163

D-13. Cross Sectional View of the Legs and Leg Bones, at Z = 0, to Scale............... 164














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

A PHYSICAL ANTHROPOMORPHIC PHANTOM OF A ONE-YEAR-OLD CHILD
WITH REAL-TIME DOSIMETRY

By

Mark W. Bower

December 1997

Chairman: David E. Hintenlang
Major Department: Nuclear and Radiological Engineering

A physical heterogeneous phantom has been created with epoxy resin based tissue

substitutes. The phantom is based on the Cristy and Eckerman mathematical phantom

which in turn is a modification of the Medical Internal Radiation Dose (MIRD) model of a

one-year-old child as presented by the Society of Nuclear Medicine. The Cristy and

Eckerman mathematical phantom, and the physical phantom, are comprised of three

different tissue types: bone, lung tissue and soft tissue. The bone tissue substitute is a

homogenous mixture of bone tissues: active marrow, inactive marrow, trabecular bone,

and cortical bone. Soft tissue organs are represented by a homogeneous soft tissue

substitute at a particular location.

Point doses were measured within the phantom with a Metal Oxide Semiconductor

Field Effect Transistor (MOSFET) based Patient Dose Verification System modified

from the original radiotherapy application. The system features multiple dosimeters that









are used to monitor entrance or exit skin doses and intracavity doses in the phantom in

real-time. Two different MOSFET device were evaluated: the typical therapy MOSFET

and a developmental MOSFET devices that has an oxide layer twice as thick as the

therapy MOSFET thus making it of higher sensitivity. The average sensitivity (free-in-air,

including backscatter) of the "high-sensitivity" MOSFET dosimeters ranged from 1.15 x

105 mV per C kg-1 (29.7 mV/R) to 1.38 x 105 mV per C kg-1 (35.7 mV/R) depending

on the energy of the x-ray field.

The integrated physical phantom was utilized to obtain point measurements of the

absorbed dose from diagnostic x-ray examinations. Organ doses were calculated based on

these point dose measurements. The phantom dosimetry system functioned well providing

real-time measurement of the dose to particular organs. The system was less reliable at

low doses where the main contribution to the dose was from scattered radiation. The

system also was of limited utility for determining the absorbed dose in larger systems such

as the skeleton. The point dose method of estimating the organ dose to large disperse

organs such as this are of questionable accuracy since only a limited number of points are

measured in a field with potentially large exposure variations. The MOSFET system was

simple to use and considerably faster than traditional thermoluminescent dosimetry. The

one-year-old simulated phantom with the real-time MOSFET dosimeters provides a

method to easily evaluate the risk to a previously understudied population from diagnostic

radiographic procedures.














CHAPTER 1
INTRODUCTION AND BACKGROUND

General


The two most common quantities used in the assessment of the risks from

exposure to ionizing radiation are (1) the individual tissue absorbed doses or (2) the

effective dose as introduced by the International Council on Radiation Protection (ICRP)

(ICRP 1995). Effective dose is a measure of the sum of the absorbed dose to specific

tissues multiplied by their respective tissue weighting factors. According to the ICRP, this

quantity is proportional to the radiation detriment to the individual. The tissue weighting

factors are based on the mortality risk from cancer, the risk of severe hereditary effects to

subsequent generations, and the risk of nonfatal cancer. The absorbed dose to a specific

tissue is difficult and time consuming to measure; consequently, the effective dose and

overall estimated risk is also difficult to calculate directly. The problem is compounded in

pediatric radiology, where the patient is particularly sensitive to radiation effects and the

overall body and relative organ sizes differ greatly from the adult Reference Man which is

commonly used for determining effective dose (ICRP 1975).

The current state-of-the-art technique to physically determine effective dose from

external exposures consists of four stages: 1) determine the radiation exposure rate, free-

in-air, at a calibration point, 2) deduce the surface absorbed dose rate, 3) calculate the

absorbed dose rate at any point of interest relative to the surface absorbed dose rate and 4)









after considering allowances for the size, shape and composition of the patient and for

differences between the patient and the phantom, make the final effective dose calculation

(ICRU 1973). The complexity associated with of these calculations are addressed in

International Commission on Radiation Units and Measurements (ICRU) Report 20: "For

external radiation, the absorbed dose at a point of interest in tissue is the product of the

absorbed dose measured at the surface of the body and a factor relating the absorbed dose

at the depth of interest to that of the surface. This factor is a complex function of the area

of the field irradiated, the depth of the point of interest, the type and energy of the

radiation, and the direction of incidence of the radiation on the body." (ICRU 1971).

Tables of conversion factors are commonly used to obtain an estimates of organ doses

from radiation exposure measurements for a particular set of circumstances, radiographic

examination, tube potential, inherent filtration of the x-ray system, for example. The

conversion factors are determined by measuring absorbed doses in physical phantoms for

the particular set of circumstances, or by mathematical simulations benchmarked against

physical measurements.

Monte Carlo mathematical routines are used to simulate radiation exposures and

calculate organ and effective doses (Cristy 1980; Cristy and Eckerman 1987; Hwang et al.

1976b; Kodimer 1995; Kramer et al. 1982; Rosenstein 1988; Rosenstein et al. 1992; Wall

and Jones 1985). The Monte Carlo routines all use various improvements to a

mathematical model originally presented by Fisher and Snyder (Fisher and Snyder 1967)

for organ doses and effective doses resulting from the internal uptake of radionuclides.

The mathematical phantoms utilize simple geometries to describe the various organs and

structures of the human body, based on reference measurements (ICRP 1975). There









have been many refinements of the original models (Bouchet et al. 1996; Cristy 1980;

Cristy and Eckerman 1987; Hwang et al. 1976a) but the principle of using relatively

simple shapes such as ellipsoids, cylinders, and spheres, to represent organs and other

body components has remained consistent. A physical phantom replicating the original

Fisher and Snyder mathematical model of an adult male has been created in the past to

physically verify certain internal and external exposures (Garry 1973; ICRU 1993). The

phantom, named "Mr. Adam," was constructed of various solid and liquid tissue

substitutes to represent soft tissue, lung and skeletal tissues. A phantom consisting of a

physical representation of the Monte Carlo computational model for a five-year-old child

was also designed, built and utilized in 1978 (Chen et al. 1978). Chen's phantom utilized

thermoluminescent dosimeters to investigate the absorbed doses to the bone marrow from

sixteen typical radiographic examinations of pediatric patients, and to verify the existing

mathematical models.

The desire to develop a direct and simple method of measuring organ and effective

doses in a pediatric phantom and verify mathematical modeling was the basis for this

research. An epoxy resin based phantom, replicating the current generation of the Fisher-

Snyder type phantom of a one-year-old child (Bouchet et al. 1996; Cristy and Eckerman

1987) has been constructed. Real-time Metal Oxide Semiconductor Field Effect

Transistor (MOSFET) dosimeters were inserted into various locations to obtain point

dose estimates of the organ absorbed dose and effective dose measurements in real-time.

The methods developed in this research can be applied in the construction of direct

reading subsequent phantoms for adults, infants and children of various ages.









Phantoms


Report 48 of the ICRU defines a phantom as "structure that contains one or more

tissue substitutes and is used to simulate radiation interactions in the body" (ICRU 1993).

Phantoms are used extensively in diagnostic radiology, radiotherapy, radiation protection,

radiation biology, nuclear medicine and radiation accident analysis. They are commonly

used to investigate dosimetric and imaging abnormalities and in radiographic equipment

calibrations. The complexity of phantoms vary from very simple geometries with single

tissue substitutes, such as a slab water simulating soft tissue or a sheet of copper

simulating a chest, to very complex geometries with multiple realistic tissue substitutes.

These realistic phantoms are referred to as anthropomorphic phantoms and simulate the

size, shape and composition of a human body to provide the same attenuation and

scattering characteristics.

Anthropomorphic phantoms, usually of the adult male, have been designed and

built to simulate the attenuation and scatter characteristics of typical patients and to

estimate the anticipated absorbed doses to the whole body and individual organs (ICRU

1993). Dosimetric phantoms are generally designed to be useful for either internal or

external radiation exposures but rarely for both. Water or solid water substitutes

(Constantinou et al. 1982) are sometimes used for dosimetric phantoms to simulate the

body as homogenous soft tissue; however, there are major nonhomogeneities found in the

body which should be modeled in an anthropomorphic phantom, such as lungs and bones

(ICRU 1976) these nonhomogeneities become increasingly important at lower photon

energies such as those used in diagnostic radiology.









Anthropomorphic phantoms, used to determine organ doses, are typically utilized

in conjunction with either multiple thermoluminescent dosimeters (TLDs) or a single

direct reading detector which can be placed in a variety of locations within the phantom

(Garry 1973; ICRU 1993). The length of time required to obtain a single effective dose

estimate with any of these methods greatly restricts the use of effective dose for radiation

risk analysis in diagnostic radiology. The procedures presented here demonstrate a

method to effectively eliminate this time delay, facilitating a more wide spread use of

effective dose calculations as a measure of risks in diagnostic radiology.

Due to the complexity of the anthropomorphic phantom and the potential for

widespread commercialization, most anthropomorphic dosimetric phantoms are built to

represent the adult man (Reference Man). Scaling factors are usually used to convert

results from the adult man anthropomorphic phantom to obtain those for women and

children. A scaled-down adult phantom is inappropriate to simulate a child since a child is

not just a reduced version of an adult. Organ growth is a complex process and can not be

described by simply scaling of the anatomy of an adult. The anatomical geometry of a

child is different from an adult. For example, the weight of the head with respect to total

body weight is greater for a child than an adult, the trunk of a child is more cylindrical

than an adult trunk which is more elliptical, and some internal organs, such as the thymus

gland, are larger with respect to other major organs in the child (Hwang et al. 1976a).

The percentage of extracellular water in children is larger and represents a larger

percentage of the total body weight than in adults and the concentration of certain

minerals is lower in the skeleton of children (Haschke et al. 1981). The analysis of the

dose incurred in pediatric radiology is of interest since growing tissue is generally









considered to be more sensitive to radiation and in most examinations a larger portion of

the child's body is included in the primary beam (NCRP 1989).

Recent studies of pediatric fluoroscopy (Nicholson et al. 1995) have shown that

increased filtration combined with the removal of the antiscatter grid can reduce the total

dose to pediatric patients by a factor of more than four without the loss of diagnostic

quality. Fluoroscopy units and image intensifiers are typically designed for adult patients.

According to Nicholson, a pediatric patient has insufficient mass to cause the proper

selection of appropriately penetrating x-rays in a fluoroscopic system operated in the

automatic exposure mode (AEC). The smaller patient also produces less scatter and

therefore makes an antiscatter grid unnecessary. Nicholson contends that properly

penetrating x-rays without an antiscatter grid reduce the entrance skin doses

approximately seven fold and reduce the effective dose by more than four fold. The use of

a physical adult anthropomorphic phantom would preclude the investigation of

Nicholson's contentions while a one-year-old anthropomorphic phantom would facilitate

this investigation.

Current commercial anthropomorphic phantoms are usually of an adult size and

are designed specifically for either internal or external sources. The phantoms designed

for internal irradiation studies are tissue equivalent (for adults) and some of the organs can

be replaced with identical organs which have been loaded with radioactive material. One

example of this type of phantom is the Lawrence Livermore Chest Phantom, developed by

Alderson Phantoms (ICRU 1993). The absorbed dose can be determined at certain points

by loading TLDs into the phantom and inserting an organ which has been loaded with a









radionuclide. The phantom could also be used to a certain extent for external radiation

studies.

Most of the phantom materials are tissue equivalent over a certain range of

interest; some phantoms however, are built with actual human bones and animal lungs.

The 3-M chest phantom is composed of adult human bones embedded in plastic. The

Humanoid System is composed of a human skeleton embedded in tissue simulating plastic

and includes a pair of dog lungs permanently fixed in an inflated state (Conway et al.

1984). While actual human skeletons have the exact elemental composition of human

bone, including all trace elements, and if obtained wet would have the appropriate human

bone density, they are difficult to obtain and may not be representative of the population in

general. A homogenous bone tissue substitute was developed for this research, rather

than attempting to locate an appropriate human skeleton, for humanitarian reasons as well

as availability and cost considerations. The synthetic homogenous skeleton is also more

easily modeled mathematically. Some anthropomorphic phantoms are designed to provide

radiographs which are similar to those of an actual patient. These phantoms may or may

not physically look like an actual patient. Christensen developed a modular chest phantom

with several drawers of one cm thick Lucite, each drawer contained one component, for

example a heart made of wax (Constantinou et al. 1986). The drawers could be changed

to simulate different abnormalities. Another type of anthropomorphic phantom has been

developed in which an actual radiograph is taken, digitized and a block of Lucite is milled

to produce a similar image. Due to orientation restrictions and the difficulty in locating

organs correctly, neither of these types of phantoms are appropriate for this research.









Appropriate tissue equivalent materials can be constructed by adding a particulate

filler with a relatively high effective atomic number, such as calcium carbonate, to an

unfilled liquid resin with a low effective atomic number. By adjusting the concentrations

of the various components, the theoretical interaction coefficients can be matched between

the tissue and the tissue equivalent material over the energy range of interest

The rapid advances in computer technology permits the use of high-end personal

computers for conducting computational simulations of diagnostic radiographic

procedures which are relatively fast, very reproducible and much less expensive than

simulations run on main frame computers. Mathematical phantoms based on the Fisher-

Snyder type model are being utilized to investigate the doses from external irradiation. A

current research project being conducted at the University of Florida is applying the

Electron Gamma Shower 4 (EGS4) Monte Carlo radiation transport code to five

mathematical phantoms representing newborn and one-, five-, ten- and fifteen-year-old

children in order to generate organ absorbed doses for typical pediatric examinations.

NCRP Report No. 117, "Research Needs for Radiation Protection" states "There is a need

to verify the mathematical models used in the calculation of energy deposition

distributions and dose in situations which cannot be directly measured ... It is also

important to substantiate the models for exposure to external radiation which take into

account age and sex differences, tissue heterogeneities and complex internal structures"

(NCRP 1993). The absorbed dose per unit exposure measured within tissue equivalent

cylinders, a simple trunk phantom and a final heterogeneous one-year-old phantom were

compared to the EGS4 generated results from plain film mathematical simulations. The









comparisons were useful in interpreting the MOSFET dosimeter response and in the

coding of the EGS4 mathematical phantoms.

Detectors


A variety of detectors have been used to measure radiation exposure or absorbed

dose. These include gas filled detectors, radiographic film (Olsen and Hansen 1990),

semiconductors (Butson et al. 1996; Orlic et al. 1989; Schroder et al. 1994),

thermoluminescent detectors (Olko et al. 1994), plastic scintillation detectors (Meger-

Wells et al. 1994), Metal Oxide Semiconductor Field Effect Transistor (MOSFET),

dosimeter (Butson et al. 1996; Gladstone et al. 1994; Hughes et al. 1988; Mackay 1996;

Soubra et al. 1994) --even alanine dosimeters have been used (Olsen, Hansen 1990).

Gas Filled Detectors

Gas filled detectors are sensitive to mechanical shocks, aging of the gas mixture,

temperature and humidity changes. These restrictions and their size made them unlikely

choices for the detectors for this project; however they were used in the earliest phantoms

and have even been used as recently as 1973 (Garry 1973). Ionization chambers are

normally used to determine the free-in-air exposure and the entrance skin exposure for

diagnostic radiology and then conversion factors are used to determine the dose

equivalent. A recent research project which typifies the use of gas-filled detectors used an

ionization chamber to determine the entrance dose, film to determine the transmitted

radiation, a solid polystyrene phantom and a Monte Carlo routine to determine the dose

equivalence in a chest radiograph (Petrone et al. 1996).









Thermoluminescent Detectors

An anthropomorphic phantom utilizing TLDs could probably be considered

standard equipment in regard to determining organ doses and ultimately effective doses.

The system requires initial calibration of the TLDs, disassembly of the phantom, placing

the TLD in the desired location, reassembly of the phantom, exposing the phantom,

disassembly of the phantom, removal of the TLDs and processing of the TLDs. Proper

processing of the TLDs requires a delay period of approximately four hours to allow

transient effects to dissipate. The TLD must be heated to a high temperature in a special

oven where the light given off is measured and correlated to the dose. The TLDs must

also be annealed prior to reuse. This lengthy process effectively limits the use of the

phantom to a single exposure per day for each set of TLDs. The phantoms utilizing TLDs

allow the determination of effective dose from a single exposure by simultaneously taking

measurements in multiple organs, as opposed to a single direct reading detector which

requires multiple exposures. A single direct reading detector system requires the phantom

assembled, irradiate, disassembled and the detector moved to a different location several

times to obtain a complete analysis.

Diode Detectors

Diodes have been used successfully in radiation therapy to measure both entrance

and exit doses on actual patients. The detectors are traditionally used in the megavolt

range. In the 6 to 12 megavolt range, the detectors have an excellent angular response

rate according to product sales literature. A relatively recent study found at therapy

energy ranges diodes agreed with ion chamber readings within 3%, had less than 0.5%

nonlinearity up to 4 Gy, and demonstrated less than 1.7% angular dose anisotropy (Lee et









al. 1994). Sun Nuclear CorporationTM is currently developing a diode system for use in

the diagnostic energy range for use with fluoroscopic systems. The system was evaluated

for use with this project. The energy response was excellent; however, it showed a

dramatic angular response. The relative angular response is shown in Figure 1. The

response is uniform when the device is tilted 45 degree from the designed orientation

but the response quickly deteriorates until there is almost no response at 90 degrees. The

response from the back of the device is also quite low. The angular response could


Figure 1. Relative Angular Response of a Diode Detector Designed for Diagnostic X-ray
Monitoring










possibly be accounted for with the analysis software or measures could be taken to always

orient the phantom towards the x-ray source; nevertheless, the use of the phantom would

be limited to certain geometries. The size of the device effectively precludes its use in the

phantom since it could not always be oriented toward the source especially with dynamic

studies where the source is in motion.

The energy response of the Sun Nuclear diodes are shown in Figure 2. The device

was calibrated at 90 kVp and demonstrated a response which was very similar to the

response from the ion chamber. At the lower tube potentials, from 30 to 100 kVp, the

response was nearly identical. Above 110 kVp the diode response was slightly lower than

the response from the RadcalTM Corporation model 1015C ion chamber with a model

10X5-6 chamber (commonly referred to as an MDH).

Another disadvantage of the current diodes are the high atomic numbered

materials used in their construction and in their associated construction of the high voltage








700
E 500 -----------i

300. -ode
u 200_
100

0 20 40 60 80 100 120 140
Tube Potential (kVp)



Figure 2. Response Comparison Between Diode Detectors and an MDH Ion Chamber









cables. It is possible for the diode detector to be used in conjunction with a detector with

better angular response to obtain a more accurate estimate of the entrance skin exposure.

Plastic Scintillators

A new plastic scintillation detector has been developed (Meger-Wells et al. 1994)

which has a small sensitive volume and is near water equivalent. A commercially available

plastic scintillator, BC-400, is manufactured by the Bicron Corporation Premium Plastic

and Liquid Scintillators ofNewbury, Ohio. The detector system with optical fiber light

guides and remote photomultiplier tubes (PMT) is custom built and has some interesting

characteristics (Beddar et al. 1992b). The scintillator is similar to the NE-102

manufactured by Nuclear Enterprises in Monmouth Junction, New Jersey. In Beddar's

detector system, one of the optical fibers acts as the "signal" light guide and it is optically

coupled to the plastic scintillator to collect and transmit the light output generated to a

PMT. The other fiber is light-shielded at the tip and is coupled to a second PMT at the

other end. The output from this fiber acts as a background signal due to the radiation

induced light generated in the fibers themselves. The background output is subtracted

from the signal output to obtain the final signal.

The plastic scintillation detector is very nearly water equivalent. The detector

(polyvinyltoluene), the wall (polystyrene) and water are closely matched for mass energy-

absorption coefficients (Beddar et al. 1992a). Beddar states "Actually, this is the best

combination of materials for matching purposes among all the known dosimeters with the

exception of the Fricke dosimeters. (However, Fricke dosimeters are not sensitive to the

absorbed doses commonly encountered in radiotherapy, and do not give real-time dose

measurements.)" He used the Bragg rule for compounds and mixtures to compare the









mass energy absorption coefficients to demonstrate the water equivalence of the

scintillators in the energy rangy of 0.1 to 20 MeV.

Temperature variations in the range of 50C to 500C did not significantly affect the

response of the a detector system, and within 50 of normal room temperature (220C),

the response was constant. Beddar demonstrated that the plastic scintillator has less

radiation damage and Meger-Wells demonstrated that the angular dependence was less

than that of a diode detector.

Similar to the diode detectors presented previously, the plastic scintillation

detectors were designed to be used at therapy energy levels. However, according to

Beddar "Theory suggests that it may be possible to calibrate these detectors in terms of

absorbed dose in water and use them in any x-ray or electron radiation field to measure

dose without the need of any further corrections. The stability of light detection systems

such as PMTs may preclude this possibility, however." The lack of a commercial "off the

shelf" plastic scintillation diagnostic system and the high cost of the customized systems,

driven by the cost of the PMT, indicate this may not be an economical choice of a detector

system for this research.

MOSFET Detectors

The MOSFET dosimeter is an extremely small (0.04 mm2 active area) detector

designed for use in measuring radiotherapy x-ray and electron beam surface doses (Butson

et al. 1996). The detector is 1000 times smaller than ion chambers and 100 times smaller

than diodes. The dosimetry system has been shown to be very effective for patient dose

measurements in fluoroscopic procedures (Mackay 1996). Mackay demonstrated the

dual-bias, dual-device, direct reading MOSFET dosimetry system had an excellent energy









response and accuracy in the radiotherapy range. However, relatively little information

has been published utilizing MOSFET dosimeters in diagnostic energy levels. The system

is manufactured by Thomson & Nielsen Electronics, LTD, (25E Northside Road, Napean

Ontario, Canada K2H 851) and can be expanded to a total of 20 dosimeters, each of

which can be operated independently. The system is self contained; however, the

capability exists for interfacing with a personal computer. The unique aspects of the

MOSFET dosimetry system and the cost analysis shown in Appendix A made it the

detector of choice for this research. It is possible that the ideal setup would be to use the

MOSFET detectors in conjunction with other detectors such as diodes or plastic

scintillators but that is beyond the scope of this research.

Rationale


Pediatric Focus

There is a strong rationale for the pediatric focus for this research. The major

impetuous for selecting a physical pediatric phantom in general and a one-year-old in

particular is the opportunity to complement current mathematical modeling research being

conducted at the University of Florida on the organ doses from typical pediatric

examinations.

A physical one-year-old phantom has a mass of approximately 10 kg (21 lb.). This

size is easily manageable. The size and weight of an adult phantom would make it

cumbersome and difficult to manufacture as the initial phantom. There also appears to be

a lack of pediatric phantoms on the commercial market. An adult phantom would also

have to be designed to be either male or female to account for size differences and the









presence or lack of breasts. The Cristy and Eckerman mathematical phantoms use the 15

year old (male) phantom as the female and the larger adult phantom as the male. Breast

absorbed dose is a significant contribution to the effective dose in females but it is not in

males. The size of some of an infant's bones are small enough to make molding a

challenge if the bones are to be removed without breaking.

The majority of pediatric radiology examinations are of children in their first few

years of life; therefore, a phantom of a one-year-old is fairly representative of a large class

of examinations. The doses from diagnostic radiology to pediatric patients have been

studied less than those given to adults, in spite of statements from other researchers. For

example, Chapple et al. claims "Pediatric patients are considered to be at particular risk for

a variety of reasons, including their higher risk factors for certain types of cancer, the

increased opportunity for the expression of induced cancers, their likely non-cooperation,

and the high frequency of some examinations during childhood" (Chapple et al. 1991;

Chapple et al. 1993).

Ease of Use

A simple to use direct-reading pediatric anthropomorphic phantom/detector system

is a benefit not only in and of itself, but for the assistance it provides to clinical

applications, and to other research in diagnostic pediatric radiology. The technology can

easily be expanded to other phantoms and a full series of phantoms could be constructed

to assist many different aspects of radiation protection.

Currently, in typical quality assurance tests, radiation protection surveys and

research is performed using ionization chambers or TLDs and simple phantoms to

determine entrance skin exposures or absorbed doses at a certain points. Mathematical









calculations are then performed separately to determine either the organ absorbed dose or

the effective dose. The direct reading system utilized is simpler to operate, more time

efficient, provides more extensive information and has additional flexibility. For instance

the effective dose from a fluoroscopy procedure with extensive spot films projected at

oblique angles could be easily evaluated or the effect of using different filtration could be

evaluated on automatic exposure control x-ray systems. The phantom/detector system

provides almost immediate results from multiple exposures taken over a short time period.

The rapid response time and amount of data collected optimizes the availability of the x-

ray system for patient studies or for other research interests.

Age

Age appropriate phantoms are required for accurate analysis of pediatric

dosimetry. An adult phantom or even a scaled down version of an adult phantom is not

appropriate for studying pediatric examinations or for comparing the results to those from

mathematical modeling. It is foreseen that this is the first of a series of phantoms which

will eventually contain an adult male, an adult female, an infant, and pediatric phantoms

representing several different ages.

Effective Dose


Effective dose (E) as presented by the ICRP in 1991 is the most current and

comprehensive measure of the radiation detriment. Effective dose equivalent (HE) as

presented by the ICRP in 1976 is the term currently used in federal guidance and in

implementing regulations. The effective dose evaluates the absorbed dose from more

tissues and applies tissue weighting factors which are numerically different from the









weighting factors used with the effective dose equivalent. The ports available for the

MOSFET dosimeters in the physical phantom allow sufficient point measurements to be

used to calculate organ doses for subsequent calculation of either effective dose or

effective dose equivalent.

Aoproach


Detectors


A Patient Dose Verification System based on the MOSFET detectors was obtained

with a high-sensitivity bias supply and five therapy detectors. At a later date, two
developmental high-sensitivity detectors were provided by Thomson and Nielsen, LTD for

testing and evaluation. The system was evaluated for free-in-air exposures, exposures

with a backscatter plate, and within tissue equivalent cylinders. The MOSFET system is a

modular system where each module accommodates up to five dosimeters. Due to

financial considerations, the system was not expanded to its full 20 dosimeter capacity. A
detailed section on the characterization of the dosimetry system can be found in Chapter 2,

Materials and Methods. The dosimetry system is interfaced to a personal computer system

with the computer program Software WedgeM.


Material Selection

The design of an anthropomorphic phantom requires the careful selection of tissue

substitute materials. The materials must closely match the volume, density and chemical

composition characteristics of the represented tissue for a proper radiological response at

the energy of interest. The materials chosen must be commercially available, relatively

simple to fabricate and maintainable for a long period of time. The epoxy resin materials

originally developed by White and others (Constantinou et al. 1982; Herman et al. 1986;









Herman et al. 1985; White 1977; White et al. 1986; White et al. 1977) met these

requirements and were used with variations. Three different materials were needed for the

manufacture of the phantom. The materials modeled consisted of a bone substitute (a

homogeneous mixture simulating trabecular bone, cortical bone, active marrow, inactive

marrow cartilage and miscellaneous material), a soft tissue substitute (a common

assumption is made that all soft tissue organ can be simulated by a single substitute

material) and a lung substitute (a heterogeneous mixture of lung tissue and air) One of

the difficulties expressed in the literature is determining what the "standard" tissue is for

pediatric patients, since the patients are rapidly growing and changing and there is a

limited data set for healthy pediatric patients. The elemental composition of lung and soft

tissue have been presented by Cristy and Eckerman (1987) based on ICRP Publication 23

data (ICRP 1975) for use in Monte Carlo calculations for all ages of phantoms and were

used for modeling purposes for this research. The ICRP Publication 70 (ICRP 1995), an

update to the skeleton section of ICRP Publication 23, was used as the guide for creating

a skeletal composition which more closely represents a one-year-old than the adult

composition used in the Cristy and Eckerman models. Cristy and Eckerman recognized

that an infant's skeletal system has a different elemental composition than adults and they

provided a unique infant skeletal composition. However, they used the adult skeletal

composition for all other ages, including their one-year-old phantom.

Phantom Construction

The tissue substitute required a combination of molding and milling, although

milling was kept to a minimum when possible by applying careful molding techniques.

The larger phantoms were constructed in successive layers as White et al. reports charring









when the thickness exceeds certain thickness, from 1 to 10 centimeters, based on the

hardeners used with the resin. Guides (small holes) were installed to allow the placement

of the detectors in the desired organ locations, while still allowing access to the detectors

for calibration and replacement. Each detector operates independently allowing the

determination of individual organ doses as well as an overall effective dose. The detectors

are generally located at the centroid of the sites of the tissues for which there are tissue

weighting factors, namely gonads, colon, lungs, stomach, bladder, breast, esophagus, liver,

thyroid, and several skeletal locations.

The detectors are interfaced through a personal computer system through the use

of a serial input/output software program Software WedgeT. The Software WedgeT

program provides real-time input and output to any Windows-based application. The

software is presently set to utilize the spreadsheet program Excelm by Microsoft;

however, any spreadsheet, word processor or text editor could be utilized for this

purpose. Appendix B contains the programming steps necessary to interface the

dosimetry system to the spreadsheet.














CHAPTER 2
MATERIALS AND METHODS

General


Three rather distinct sections are presented in this chapter. In the first section, the

MOSFET dosimetry system is characterized. The second gives a description of the

modeling of the bone, lung, and soft tissue substitutes. Finally, a section on phantom

construction is presented, progressing from Phase 1 simple (one tissue) cylindrical

phantoms, to a Phase 2 composite phantom of the trunk, to the Phase 3 heterogeneous

one-year-old phantom.

A commercial Patient Dose Verification System manufactured by Thomson &

Nielsen Electronics Ltd., 25E Northside Road, Nepean, Ontario, Canada, K2H 8S1,

utilizing miniature non-invasive Metal Oxide Semiconductor Field Effect Transistor

(MOSFET) dosimeters originally designed for radiotherapy applications was evaluated

and utilized for this research at diagnostic energy levels. The system features multiple

dosimeters which may be used to monitor entrance or exit skin dose and intracavity doses

in phantoms in real-time. Both the standard MOSFET dosimeter designed for

radiotherapy and a newly developed "high-sensitivity" MOSFET dosimeter designed for

lower dose measurements were characterized. The sensitivity, linearity, angular response,

post-exposure response and physical characteristics were evaluated.









MOSFET Dosimetry System


Introduction

Recently documented cases of erythema caused by radiation overexposure from

diagnostic x-ray systems has established a desire to monitor absorbed doses from these

procedures (Wagner et al. 1994). Two established methods of monitoring absorbed organ

doses include (1) measuring the free-in-air exposure at the location where the x-rays are

expected to enter the patient, known as entrance skin exposure (ESE), and then using a

table of conversion factors to convert this quantity to an estimated absorbed organ dose

[BRH, 1979 #2341] or (2) using thermoluminescence dosimeters (TLD) on actual patients

to measure the ESE directly and again use a table of conversion factors to convert this

quantity to an estimated absorbed organ dose. The ESE measured directly with the TLD

contains a backscatter component which is not present in the free-in-air measurement of

the ESE, care must be taken to use the appropriate conversion tables. The method of

measuring the ESE and utilizing generic patient conversion factors is not patient specific

and is not likely to account for abnormalities which could result in higher doses such as

extremely large or uncooperative patients. A TLD system requires a lengthy processing

time, limiting the number of patients which can be monitored and preventing the

possibility of intervention before the completion of the examination. An ideal system for

monitoring absorbed doses during diagnostic radiographic procedures would have the

following characteristics: (1) present an extremely small dosimeter area to prevent

interference with the quality of the diagnostic image, (2) provide real-time results, (3) be









simple to use, (4) maintain a level of precision equivalent to an ion chamber, and (5)

demonstrate no angular dependence.

A typical ion chamber easily meets conditions 2, 4 and 5, but the relatively large

size of the x-ray probe, ion chamber, and associated high voltage cabling present an

unacceptably large inhomogeneity within the phantom. A very large ion chamber or an

ion chamber outside of the primary beam could be used to obtain an indication of the

absorbed dose for ESE measurements, but precise determinations of absorbed dose in

organs would be difficult if not impossible to obtain with this device. The size of a typical

ion chamber and its high voltage requirements make it undesirable for use in this research.

Thermoluminescent dosimeters present acceptably small perturbations in the x-ray field,

but their lengthy processing time prevents the use of the absorbed dose information during

the measurement and TLD systems do not allow intermediate dose monitoring. An

alternate dosimetry system based on MOSFET technology was evaluated against these

established criteria of small size, real-time analysis, simplicity, limited angular dependence

and acceptable level of precision.

The MOSFET is a very common microelectronic device. MOSFET devices are

commonly used in space applications as radiation detectors (Sharp and Pater 1996) and

are seeing increasing applications in radiotherapy (Butson et al. 1996; Soubra et al. 1994).

The extremely small size of the dosimeter has been exploited for in vivo measurements in

both animals (Gladstone and Chin 1995) and in humans (Hughes et al. 1988). It is a

layered device consisting of a p-type semiconductor separated from a metal gate by an

insulating oxide layer. Ionizing radiation forms electron-hole pairs in the oxide insulating

layer of the MOSFET. The applied bias causes these electrons to travel to the gate while









holes migrate to the silicon, silicon-oxide interface where they become trapped. The

trapped positive charges produce a negative shift in the voltage required to allow current

to pass through the MOSFET. The shift in voltage is proportional to the radiation

absorbed dose deposited and thus allowing the MOSFET to be used as a dosimeter

(Gladstone and Chin 1991; Hughes et al. 1988; Vettese et al. 1996). The sensitivity of the

MOSFET can be increased by applying a positive gate bias during irradiation which

increases hole trapping efficiency by suppressing electron-hole recombination within the

oxide layer and by increasing the oxide thickness thus increasing the number of electron-

hole pairs generated. The oxide layer of the Patient Dose Verification System's high-

sensitivity MOSFETs are twice as thick (1 ipm) as the therapy MOSFETs (0.5 Irm). It is

possible to further increase the sensitivity of a single MOSFET device. Some authors

(Gessinn and Sarrabayrouse 1993) have estimated sensitivities as high as 9 V Gy"i (90

mV rad') with a large applied bias and very thick oxide levels. The cost associated with

the long growth times and quality control procedures required for manufacturing the very

thick oxide layers and the high applied bias voltages are restrictions to obtaining an

economical, operational, single MOSFET device with such a high sensitivity. The

expected lifetime of such a device would also be very short since the higher the sensitivity,

the sooner the device reaches saturation voltage. The MOSFET dosimeters utilized for

this work are considered to be saturated when they have received an accumulated dose

which exceeds the reading capability of the reader, 20 V. Other authors (Kelleher et al.

1995; O'Connell et al. 1996) have developed a method of stacking MOSFET devices for

increased sensitivity. Utilizing unbiased devices with a single MOSFET sensitivity of

0.036 V Gy (0.36 mV rad'"), they have measured a sensitivity of 0.262 V Gy"' (2.62 mV









rad"') for a stack of four MOSFETs. They have calculated a theoretical sensitivity of

8.260 V Gy-1 (82.6 mV rad"~) for a dosimeter with 40 MOSFET devices stacked in series.

The high-sensitivity dosimeters evaluated in this work have a resolution adequate for most

applications involving diagnostic radiology, but greater sensitivity may be obtained using

one of the methods listed above.

Physical Characteristics

The Thomson and Nielsen Patient Dose Verification System, Model TN-RD-50,

utilizes dual bias, dual MOSFET device, dosimeters. The system was originally designed

for measuring verification doses in radiation oncology; however this work demonstrates

they are adaptable to studies in diagnostic radiology. The basic system consists of a

reader, a bias supply, power supply, MOSFET dosimeters, and an optional printer. A fully

expanded system consists of twenty dosimeters with a separate bias supply for every five

dosimeters.

A monitoring module consists of the bias supply and up to five dosimeters. The

monitoring module must be disconnected from the reader prior to placing the MOSFET

dosimeters on a patient in order to protect the patient from any possible electrical hazard

from the integrated system. The monitoring module may remain connected to the reader

during physics work when a live patient is not involved.

Each dosimeter of the Patient Dose Verification System consists of two MOSFET

devices, mounted together, operated at different bias voltages. Each MOSFET device has

an active area of 0.04 mm2, and the pair are mounted under a 1 mm thick layer of black

epoxy to a 20 cm long, thin, semi-opaque polyamide laminate cable encasing two gold

wires. The extremely thin and flexible laminate cable is attached to a sturdy 1.4 m cable









which is connected to the dual bias supply. The entire dosimetry system is shown in

Figure 3 including; the reader, power supply, bias supply, a MOSFET dosimeter alongside

a 15 cm (6 inch) ruler for reference, a Phase 1 soft tissue-equivalent cylindrical phantom

with a MOSFET dosimeter inserted and the associated cabling. The large cylinder to the

right is the Phase 2 composite trunk phantom.
























Figure 3. Thomson and Nielsen Electronics LTD. MOSFET Dosimetry System


The TN-RD 50 Patient Dose Monitoring System measures the difference in

response between the two MOSFET devices comprising each dosimeter. A single

MOSFET device is sensitive to temperature and also exhibits a decreasing sensitivity as a

function of accumulated radiation dose. These effects are nearly eliminated by measuring

the response difference between a matched pair of MOSFET devices formed on the same









silicon chip and operated at two different positive gate bias. The resultant temperature

coefficient from a dual-bias, dual MOSFET dosimeter has been measured to be less than

0.015 mV*C"' (Soubra et al. 1994).

To facilitate monitoring of the lower absorbed doses associated with diagnostic

x-ray systems, a model TN-RD-19 high-sensitivity bias supply was used in this research.

The high-sensitivity bias supply provides a voltage of 15 V and 0.5 V to the two

MOSFET devices in each dosimeter. Two different types of MOSFET dosimeters were

evaluated for use in this research: the standard MOSFET designed for radiotherapy

applications, and a newly developed high-sensitivity MOSFET designed for lower dose

applications.

Sensitivity

The sensitivity, defined as the response of the system (mV) per unit of exposure (C

kg" or R), was found to depend on several factors including the oxide thickness, the

energy of the x-ray field, the presence of a backscatter medium, and the dosimeter

orientation for surface doses. The Patient Dose Verification System utilizes two separate

algorithms for the voltage measurement depending on the total accumulated dose, before

and after the accumulation of 4,000 mV; therefore, the sensitivity must be reevaluated

periodically and at a minimum after the dosimeters reach an accumulated dose associated

with a measurement of 4,000 mV.

The sensitivity of the MOSFET dosimeters was measured over the x-ray tube

potential range of 40 to 145 kVp at intervals of 5 kVp for the therapy dosimeters and 10

kVp for the high-sensitivity dosimeters. The MOSFET dosimeter is a consumable device

in which the charge continues to accumulate until it exceeds the voltage range of the









reader, since the high-sensitivity dosimeters were three times more sensitive than the

therapy dosimeters they are consumed three times as fast. In many of the analyses

performed the measurement interval was extended for the high-sensitivity dosimeters to

compensate for this higher consumption rate. The MOSFET dosimeters were placed on an

acrylic calibration plate to provide a backscattering media during the sensitivity

measurements. The plate was supplied by the manufacturer to simulate the backscatter

expected from a patient. The center line of the ion chamber was placed in the x-ray field

at the same distance as the MOSFET dosimeters to obtain simultaneous exposure and

dose measurements. During the initial sensitivity calibration measurements of the therapy

MOSFET dosimeters, efforts were made to expose the dosimeters to roughly the same

dose at each interval, with exposures ranging from 5.7 x 10'4 C kg (2.2 R) to 1.1 x 10" C

kg'1 (4.4 R). The dose was held relatively constant by adjusting the techniques of the x-

ray system and the number of x-ray exposures taken. For example, at a tube potential of

40 kVp, six x-ray exposures were taken at 400 mAs per exposure. At a tube potential of

70 kVp, one x-ray exposure was taken at 375 mAs. At a tube potential of 145 kVp, one

exposure was taken at 150 mAs. Later sensitivity measurements of the high-sensitivity

and therapy dosimeters used a single exposure at a constant mAs setting for each

sensitivity measurement.

The epoxy which secures the MOSFET devices to the polyamide laminate cable is

applied as a roughly hemispherical drop approximately 1 mm thick and creates a raised

"bubble side" and a "flat side" of the dosimeter. The MOSFET is in direct contact with

the cable which is much thinner than the epoxy bubble. There is a marked difference in the

surface sensitivity responses between the two sides; the average sensitivity with radiation









directed toward the flat side of the dosimeter is only 75% of the average sensitivity when

radiation is directed toward the bubble side when no backscatter plate is used. The

sensitivity does not show an angular dependence when the MOSFET dosimeter is

irradiated within a tissue equivalent Phase 1 cylindrical phantom. The average sensitivity

of the radiotherapy MOSFET dosimeters, including backscatter, with the bubble side

oriented toward the x-ray source ranged from 4.06 x 104 mV per C kg"1 (10.5 mV/R) to

5.23 x 104 mV per C kg' (13.5 mV/R). The sensitivity of the high-sensitivity MOSFET

dosimeters, including backscatter, with the bubble side oriented toward the x-ray source

ranged from 1.13 x 105 mV per C kg" (29.0 mV/R) to 1.36 x 105 mV per C kg (35.2

mV/R). The sensitivity of both types of MOSFET dosimeters over a range of energies for

a variety of orientations, are given in Table 1.

The sensitivities as a function of tube potential are shown in the Figure 4 for both

types of MOSFET dosimeters early in their exposure history. An MDH ion chamber was

used to simultaneously measure exposure while obtaining absorbed dose (mV)

measurements with the Patient Dose Verification System. A linear fit to the data was used

to obtain calibration factors as a function of tube potential. After the dosimeters had been

in use for a period of time the sensitivity calibration was repeated and the data, shown in

Figure 5, appeared to fit quadratic functions better than the linear functions which were

shown in Figure 4.

The dosimetry system allows the operator to enter a single correction factor for

each MOSFET dosimeter to convert the mV measurement to a dose in units of either gray

or rad (or any other desired unit, however the only available labels on the display unit are

gray, rad or mV). The output of the system can also be directed to a personal computer,









Table 1. Sensitivity of Therapy and High-sensitivity MOSFET Dosimeters at Diagnostic
Energy Levels


Tube
Potential
(kVp)


Therapy
Dosimeter
Epoxy Side
no backscatter
mV/R

10.3

11.8

10.7

11.9

10.7

10.0

10.2

10.1

10.0

9.5

9.7


Therapy Dosimeter
Epoxy Side
with backscatter
mV/R


Therapy
Dosimeter
Flat Side
with backscatter
mV/R

7.9

8.3

8.7

9.8

9.8

9.8

9.4

9.3

10.1

9.5

9.5


High-sensitivity
Dosimeter -
Epoxy Side with
Backscatter
mV/R

32.3

30.8

35.2

31.4

33.6

35.0

31.7

31.5

32.6

34.7

29.0


through a separate computer program such as Software WedgeTM, where the energy

dependent correction factors can be applied.

The reader only displays whole numbers; however, it contains an extra unit in its

memory which can be accessed by using a calibration factor of 0.10, the lowest possible

setting. The setting permits the system to respond to a minimum exposure of 7.7 x 10.7 C

kg" (3 mR) for the high-sensitivity dosimeters and 2.5 x 106 C kg"' (9.5 mR) for the

therapy dosimeters. Other authors (Tarr et al. 1996) have used a similar dosimetry system

to measure a minimum resolvable dose of 300 pGy (30 mrad) from a 37Cs (0.66 MeV

gamma) source. Their system had a sensitivity of 0.2 V GyW' (2 mV rad"') at this energy








31







40




30

E

20 Hh Snaavty Dos*r
10


I h --. I -- H.----- .
10




40 50 60 70 80 90 100 110 120 130 140 150
Tube Potential (kVp)





Figure 4. Sensitivity of MOSFET System with Low Total Absorbed Dose







40





I Thf30Do


0 | igh Sn MtHy DOim T |aer
20 T.-iStb



10 -




40 50 0O 70 80 90 100 110 120 130 140
Tube Potential (kVp)





Figure 5. Sensitivity ofMOSFET System with Substantial Total Absorbed Dose









and they attributed the resolution limits to electronic noise in the dosimeter itself The

manufacturer claims that while the analog-to-digital converter has a resolution of 0.006

mV, the offset drift of the reader component resolution is approximately 0.1 mV and the

dosimeter itself contributes another 0.5 mV for a total reliable resolution of

approximately 0.6 mV. This resolution corresponds to an exposure of 4.4 x 10"' C kg"'

(17 mR) for the high-sensitivity dosimeter and 1.2 x 10"5 C kg"' (48 mR) for the therapy

dosimeter.



The linearity analysis was performed at three tube potential settings (60, 90 and

120 kVp) to cover the general range of techniques used in diagnostic radiology. A single

exposure time (1 second) was utilized and the current was varied over several settings (25,

40, 60, 100, 150, 250, 400 mA). The dosimeters were arranged as follows: (1) a high-

sensitivity dosimeter was oriented with the epoxy bubble portion facing the x-ray tube,

free-in-air, (2) a high-sensitivity dosimeter was oriented with the flat polyamide cable

facing the x-ray tube, free-in-air (3) a therapy dosimeter was place on the top of a 4.2 cm

diameter horizontal Phase 1 cylindrical phantom of tissue equivalent material with the

epoxy facing the x-ray tube, (4) a therapy dosimeter was placed at the center of the Phase

1 cylindrical phantom, and (5) a therapy dosimeter was placed on the bottom of the Phase

1 cylindrical phantom. The dosimeters were arranged to examine the linearity of both sides

of the dosimeters and to examine the effect of different scattering geometries.

The response of the detection system was found to be very linear with dose, as

shown in Figs. 6-8 for tube potentials of 60, 90 and 120 kVp. At 60 kVp, the high-

sensitivity dosimeters show a slight degree of nonlinearity with one point outside the one




















0 Soahl k lltly DOaalartr 000050 TOV.0lr SOar
+ 050h 0.0.I50l0 Oaaalar FraI Tlolrd 500r.0
To T.lr,.y flO0*la To0 o T-1d1r
T0lrlpy 0.1*a1r hlat Caar 1Cyl0dOr
Thirpy Boat. tla, at orttom o tCylhtdo


02 04 0.0 0.8 1.0 1.2 1.4
Exposure (R)


18 18 2.0 22


Figure 6. Linearity ofMOSFET System at 60 kVp


S---- fl lnS ilt Ocalmat.r bb.le. T.owrd Sor,.
0 -I --- Sigh $rafPlvly 000,0ltr0 Filt Toward Source

S Thrapy Doshfier It 1B*"n, of Cybller


















0 1 2 3 4 5
Exposure (R)


Figure 7. Linearity of MOSFET System at 90 kVp


IIII

















0 1 2 3 4 B 6 7 S

100






Exposure (R)



Figure 8. Linearity of MOSFET System at 120 kVp


sigma error bars, while at 90 and 120 kVp, excellent linearity is demonstrated. The

therapy MOSFET dosimeters have excellent linearity throughout this range of tube

potentials.

Angular Response

The angular response was evaluated for two configurations: one free-in-air and

one within a tissue equivalent Phase 1 cylindrical phantom. The dosimeters were

irradiated in 15 degree increments for a full 360 degree rotation, at 90 kVp and 300 mAs,

for the free-in-air measurements. The soft-tissue equivalent Phase 1 cylindrical phantom

used to measure the angular response is 8.4 cm long and has a diameter of 5.2 cm with a

small hole drilled in one end to allow insertion of the MOSFET dosimeter. The guide hole

allowed the placement of the MOSFET 3.3 cm down the central axis. The Phase I

cylindrical phantom was placed horizontally and rotated about its axis in 15 degree









increments for a full 360 degree rotation, at 90 kVp and 150 mAs. The dosimeter was

removed, rotated 180 degrees and replaced in the Phase 1 phantom, and the procedure

was repeated. The results of the two series of measurements were averaged to minimize

the possibility of introducing an angular dependence due to any undetected inhomogeneity

of the Phase 1 cylindrical phantom.

Angular dependence of the MOSFET dosimeter system irradiated free-in-air is

shown in Figure 9 as a composite of the angular response from five different therapy

dosimeters, normalized to the response with the radiation directed towards the top of the

epoxy surface of the dosimeter. The individual dosimeters all demonstrated an angular



90









180 0
180 I --- --- : --- I --: -- -- \ -- -----


Figure 9. Relative Angular Response ofMOSFET Dosimeters, Free-in-air









response similar to that shown in Figure 9, by averaging the five therapy dosimeters the

overall trend is more clearly evident. The reference orientation (zero degrees) is with the

flat portion of the dosimeter facing the x-ray tube; at 180 degrees the epoxy bubble side

facing the x-ray tube. The response is relatively constant for orientations with the radiation

entering from 90 to 270 degrees. The response is also relatively uniform for the flat side

of the dosimeter; however, it is only about 60% of the response from the epoxy side. The

packaging of the MOSFET dosimeter appears to be the main source of this angular

dependence. Other authors (Gladstone et al. 1994) have noted a directional dependence

of a comparable magnitude in the axial direction and a slightly greater dependence along a

transverse angle, from single device MOSFETs used to measure radiation from 9Sr. They

attributed the dependence to the packaging and did not observe any directional

dependence in megavoltage photon or electron beams when the probe was placed in a

spherical phantom under conditions of electronic equilibrium.

The angular response was very symmetrical for both the therapy and the high-

sensitivity MOSFET when irradiated within tissue equivalent cylinders. Figures 10 and 11

illustrate the excellent angular response possible when the MOSFET device is placed

within the Phase 1 cylindrical phantom. The lines in Figures 10 and 11 were created by

first creating a best linear fit to the data in rectangular coordinates, second predict values

at the same angles as the original data, and finally transferring the calculated points to the

polar plot. They are provided to demonstrate the general trend of the data and should not

necessarily be used to predict future responses.
























180 a
S 075 050 o05 025 0o0 075 .0










270




Figure 10. Relative Angular Response of Therapy MOSFET Dosimeters within a Tissue
Equivalent Phase 1 Cylindrical Phantom





Post-exposure Response

The post-exposure drift response was evaluated by irradiating the therapy MOSFET

dosimeter with a dose of 142.3 mV (approximately 3 x 10-' C kg"' or 11.5 R). The

dosimeter was read several times within the first 15 minute period, then approximately

every 15 minutes for the next 8 hours and finally one to three times a day for the next 21

days.















90










180 0
1 075 0.50 025 0.25 050 075 00










270

Figure 11. Relative Angular Response of High-sensitivity MOSFET within a Tissue
Equivalent Phase 1 Cylindrical Phantom


The dosimeter which was irradiated and used in the post-exposure drift response

measurement experiment for over 500 hours was used to demonstrate the effect of

multiple querying of the system. The system was queried for a response approximately

every 30 seconds for 36 individual measurements.

The post-exposure drift response of the MOSFET dosimeter is a complicated function of

time. The drift in single MOSFET devices has been attributed to a number of mechanisms

including: positive ion diffusion, slow interface states introduced by radiation exposure,

slow interface states introduced by hot-carrier injection from the silicon during stressing,









relaxation of charges residing in deep traps in the oxide, capacitance-voltage hysteresis,

enhanced 1/f noise, compensation of trapped holes, increased thermally stimulated current

in MOS capacitors and both thermal and nonthermal annealing of trapped charges

(Fleetwood 1996; Holmes-Siedle and Adams 1983). Mathematical models to describe the

drift effects and deconvolution methods have been proposed and employed for use in

applications requiring delayed readings (Gladstone and Chin 1995). Empirical modeling

of the data would be possible if reading the system was delayed. The post-exposure drift

response for the dual-bias, dual MOSFET device system utilizing therapy dosimeters is

shown in Figure 12. A dose was applied to the dosimeter and the post-exposure response

was examined without any additional radiation exposure. A relatively rapid rise of the

response occurred for approximately 26 hours, and a slower steady decline was observed

thereafter. Measurements were made over a 500 hour period with no leveling off

observed. Long-term delays in reading the dosimeters of this system are not anticipated

in normal applications since near-real-time measurements are one of the main

advantages/applications of the system. The mid-term response, consisting of the first full

workday (8 hours), of the post-exposure drift response is shown in Figure 13. An

empirical function could simply be fit to the data and used to correct dose measurements

made within the same day. The manufacturer indicates the system is designed to be read

within 15 minutes to ensure a post-exposure drift response of less than 2% for an

absorbed dose of 200 mV. The short-term response is illustrated in Figure 14 covering

this first fifteen minute period.

The dosimeter system measures the absorbed dose by measuring the voltage

required to pass a given threshold of current through the MOSFET dosimeter. This














170




ISO-



150
*



150









140
1X .





0 100 200 300 400 500

Time After Dose (hour)


Figure 12. Long Term Post-exposure Response of a MOSFET Dosimeter



process also affects the post-exposure response. The effect is minimal for a single query

of the system, but multiple queries of the same exposure will artificially increase the

indicated response. The expected post-exposure drift response from the dosimeter as

illustrated in Figure 12 should be a continual decline, as it had been for the previous 500

hours; however, when the dosimeter was queried multiple times in a short period of time,

the response increased as shown in Figure 15. Each data point in Figure 15 is a separate

query of the system with queries made approximately every 30 seconds for 20 minutes.














170











1. *
140



130



020
1[20 -' ----, ---, ---, i --, ---- --,- --,- ----- --,--
0 1 2 3 4 5 6 7 8 9
Time After Exposure (hour)



Figure 13. Mid Term Post-exposure Response of a MOSFET Dosimeter



The initial query in Figure 15 was made approximately 500 hours post exposure. The

effect can also be seen in Figure 12 by observing the difference in the response between

readings which were taken relatively close together as compared to readings separated

over longer time periods. The post-exposure response and the multiple query response

were not major concerns for this research since the data was obtained within a few

seconds of the exposure and the data was captured to a spread sheet precluding the need

for multiple queries.













170



10D



,150



140



130




0 2 4 6 8 10 12 14 16
Time Ater Expoure (minute)



Figure 14. Short Term "Realistic" Post-exposure Response of a MOSFET Dosimeter





Measurement of Absorbed Dose

The sensitivity has been defined as the MOSFET response per unit of exposure

(mV per C kg" or mV/R) at a given tube potential; consequently the absorbed dose (rad)

in the tissue may be derived as follows. The ion chamber measures exposure in air in units

of roentgen (R); therefore, this unit was used in measuring the sensitivity. The exposure

in air, as measured by the MOSFET dosimetry system is therefore the MOSFET dose

measurement (mV) divided by the sensitivity (mV/R). Expressed symbolically in Equation

1, where X is the exposure, M is the MOSFET dose measurement and S is the sensitivity.















128

128

124


122 -


118

116

114 .
112 ,-- --, --, ---, ---, ---, ---, -,
0 2 4 6 8 10 12 14 16 18 20

Time Aer Fint Query (min)
Each Data Point Represens an Addiional Query




Figure 15. Multiple Query Response






X= M* (Eq. 1)


One roentgen produces an absorbed dose of 0.876 rads in air; therefore,

D, = X*0.876 (Eq. 2)

The absorbed dose in tissue may be then be calculated with the following equation:


D,, = .. D, ) (Eq. 3)
(-Z/L,









Where p, / p is the average mass energy-absorption coefficient associated with

the mean energy of the x-ray spectrum. Equations 1-3 may be combined into one useful

equation as follows.


D,,, = 0.876. *~ ( M (Eq. 4)


The value of p.. I p for the oxide could be removed from the equation without

error; however, its inclusion indicates the absorbed dose in the MOSFET device is an

integral part of the corresponding dose value in tissue. The average mass energy-

absorption coefficient ratios are effectively 1.0 for soft tissue, but increase to a value of

3.5 for skeletal tissue at 30 keV. The sensitivity of each dosimeter as a function of tube

potential has been determined and was shown in Figures 4 and 5. Periodically the

sensitivity of the MOSFET dosimeters must be reevaluated.

The mean energy of the x-ray spectrum was estimated by utilizing the computer

program XCOM5R (Nowotny and Hifer 1985). The program is a spectrum generating

program which accounts for items such as the tube potential, the inherent filtration, the

target materials, etc. The program provides information other than the x-ray spectrum,

one of these items is the mean photon energy of the spectrum. More information on this

software is found in Appendix B. The determination of the exact mean spectrum energy is

not critical for the soft tissue or the lungs since the ratio of mass energy absorption

coefficients is relatively constant and nearly 1.0 over the diagnostic energy range. The

determination is more critical for the bones since that ratio is a rapidly changing with

photon energy and peaks around 30 keV.









Tissue Substitute Development


Tissue-equivalent Material Fabrication

The construction of physical phantoms required the development of processing

techniques for three tissue-equivalent materials: soft tissue, skeletal tissue (considered a

homogenized mixture of bone matrix and bone marrow), and lung tissue (considered a

homogenized mixture of soft tissue and air). The Phase 1 cylindrical phantoms consisted

of single homogenous cylinders of each of the tissue substitutes. The Phase 2 composite

trunk phantom utilized all three tissue substitute materials in simple cylinders representing

the lungs, spine and the soft tissue in the trunk of a one-year-old. The Phase 3

heterogeneous one-year-old physical phantom consists of three primary regions. The first

two regions are the trunk region and leg region where both are physical representations of

the one-year-old mathematical trunk and legs defined in Cristy and Eckerman (Cristy and

Eckerman 1987). The trunk region is an elliptical cylinder of soft tissue substitute which

encloses the lungs (constructed of lung tissue substitute) and the trunk skeleton (arm

bones, pelvis, spine, ribs, scapula, and clavicles all constructed of bone tissue substitute).

The leg region consists of the frustums of two circular cones of soft tissue substitute each

enclosing a leg bone (bone substitute). The third region is the head region. This region is

a physical representation of the Bouchet-Bolch head model developed for the Society of

Nuclear Medicine's Medical Internal Radiation Dose (MIRD) Committee. The adult

version of this model is described in (Bouchet et al. 1996) and MIRD Pamphlet Number

15 (Bouchet et al. 1997). Explicit reference to the one-year-old mathematical model is in

preparation (Bouchet, To be submitted Mar 1998). This region consists of a head and









neck region of soft tissue substitute enclosing the skeletal regions of the head (mandible,

teeth, cranium, and upper facial region).

A method of creating tissue substitutes originally developed by White (White

1977) and further developed by White et al. (White et al. 1986; White et al. 1977) and

Herman et al. (Herman et al. 1986; Herman et al. 1985) was modified for the purposes of

this research. The production of a tissue-equivalent material requires the addition of a

particulate filler of a moderately high effective atomic number, such as magnesium oxide

or calcium carbonate, to an unfilled liquid resin of a low effective atomic number. The

mass density of the substitute is adjusted by adding low density microballoons (hollow

gas-filled spheres). A foaming agent is also added to the lung substitute to provide a lung-

equivalent material with the proper mass density (White et al. 1986).

One of the significant differences between the method developed in this research

for creating soft tissue-equivalent materials and that utilized by White and colleagues is the

lack of dependence on a vacuum system. White et al. (1986) utilized a vacuum system in

which they could mechanically mix the material under the vacuum system to remove any

air bubbles trapped in the mixture. A process to mix the tissue substitutes under a vacuum

was not readily available and it was desirable to ultimately mold rather large and awkward

components which do not readily lend themselves to curing under a vacuum. Therefore a

vacuumless method was developed. Original attempts to recreate White's compounds by

placing a pre-mixed batch of material under a vacuum to cure and dry resulted in the

material expanding undesirably with small air pockets throughout, thus giving it an

unacceptable low mass density of 0.86 g/cm3. The vacuumless method developed for the

soft tissue utilizes an initial mixing ofepoxy, hardener, and magnesium oxide powder.









This mixture tends to foam immediately after the dry magnesium oxide powder is

suspended in the epoxy-hardener mixture with air bubbles rising to the surface for a few

minutes. The trapped air is encouraged to escape through mechanical shock and normal

migration. Since the epoxy-hardener/magnesium oxide material has a higher viscosity than

the full soft tissue equivalent material, the trapped air escapes more readily. The air also

escapes more readily from this mixture since the epoxy resin has not begun to harden as it

does latter in the curing process. The other powdered materials are then added to the

degassed mixture. A very small amount of air is introduced during this mechanical mixing

but it escapes readily as well. The first few millimeters of the finished hardened material

may contain air bubbles if the degassing was not performed with the proper diligence but it

was discovered that this layer could easily be removed and a very homogenous material

could be readily produced. As discussed below, good results are found with this method

without the volume restrictions associated with mixing materials under vacuum.

One disadvantage of the epoxy resin based tissue substitute method noted by

previous researchers (White 1977) is the production of heat during the chemical reaction

which may cause charring. White limited the maximum thickness which may be poured at

any one time to approximately ten centimeters. The disadvantage is overcome by pouring

several successive layers and allowing each to harden before pouring the next. The

overheating was not observed in this work but most of the layers poured were less than

ten centimeters.

The bone tissue substitute material utilizes calcium carbonate as the higher atomic

number filler. The calcium carbonate does not present a foaming problem such as that

presented with the magnesium oxide. The minimal amount of air induced in the









mechanical mixing process is quickly dissipated. Since air removal is not a major concern

with the bone tissue substitute, all the ingredients can be mixed concurrently, without the

two step process used for the soft tissue.

The lungs tissue substitute also utilized a modification of a method originally

developed by White (White et al. 1986). White utilized a foaming agent and a surfactant

to obtain the proper lung density. Attempting to reproduce White's work resulted in a

lung tissue substitute which was too dense. The density was not reproducible for different

mixtures and it was discovered that the mixture was actually over-foaming, partially

collapsing upon itself, and then foaming a second time. The method developed in this

work utilizes only half the amount of foaming agent which White utilized, with minor

adjustments made to the other components. White warned of a possibility of the lung

substitute collapsing due to mechanical shocks or to a "high grade" of magnesium oxide.

White noted that molds constructed of silicon rubber, PTFE "Vinamould" and

waxed perspex have been found to work well with these tissue substitute materials (White

et al. 1977). Extensive use of flexible 0.05 inch thick polyethylene sheeting was used in

our molds with good results; waxed PVC piping was also used effectively. These

alternate molding materials were inexpensive, readily available and the sheeting was easily

manipulated to the proper shape. The resin base tissue equivalent materials were also

milled after curing if necessary. Most of the individual bones, the lungs and the final

external surfaces were milled to obtain the proper size and shapes.

Verification of Tissue-equivalency

In the design of tissue substitutes for use in diagnostic radiology, one must match

the mass attenuation and mass energy-absorption coefficients of the proposed materials









with those of some standard reference tissue media over the photon energy range 1.5 to

150 keV. Since the goal was to construct a full physical representation of the Cristy and

Eckerman pediatric model series, data from Cristy and Eckerman (1987) and ICRP

Publication 23 were utilized for this purpose.

For each proposed tissue substitute, the computer program XCOM written by

Berger and Hubbell at the Center for Radiation Research, National Bureau of Standards,

(Hubbell 1982) was used iteratively to match the total mass attenuation coefficients

between the proposed material composition and the reference tissue. The XCOM

program is a Microsoft DOSTM based program which provides partial interaction

coefficients and total attenuation coefficients for elements, compounds or mixtures. The

operator enters the chemical symbol and fraction by weight for each component and either

accepts the standard energy grid or supplies a file of the desired energies. Appendix B

contains a listing of the inputs and an example output for the program.

Final specifications for all three tissue substitutes were found to give photon

interaction coefficients which agreed with those of Cristy and Eckerman (1987) within

1-3% over the full energy range of diagnostic interest. The photon interaction coefficients

of the soft tissue substitute also compare favorably to those given for individual tissues in

ICRU Publication 44 (ICRU 1989). Once a promising tissue substitute was developed, its

density was measured and adjusted accordingly by varying the concentration of phenolic

microballoons within the mixture.

Soft Tissue Substitute

The primary soft tissue-equivalent substitute developed, STESI, consists of the

following materials by percent mass: 47.34% Araldite GY60-10, 18.93% Jeffamine T-403









(a hardener), 15.65% polyethylene powder (medium density), 17.43% Magnesium Oxide

powder and 0.65% Phenolic microballoons. The materials used in STES1 are the same,

with different concentrations, as one of the soft tissue substitutes originally developed by

White. White modeled his material to be used over a wide energy range, 10 keV to 100

MeV. The concentrations developed here were adjusted to more closely model the

diagnostic energy range, 1.5 keV to 150 keV and still provide the appropriate soft tissue

density of 1.04 g/cm3.. Table 2 gives a comparison of the elemental composition of this

mixture, the soft tissue composition given in ICRP Publication 23, and soft tissue

composition given by Cristy and Eckerman (1987). As with most attempts to match soft

tissue reference composition in phantom construction, a necessary substitution of


Table 2. Elemental Composition of the Soft Tissues and Soft Tissue Substitute, STES1,
by Percent Mass.


Cristy & Eckerman ICRP Publication 23
(1987)
10.454 10.448
22.663 23.218
2.49 2.488
63.353 63.021
0.112 0.113
0.013 0.013
0.03
0.134 0.1327
0.204 0.199
0.133 0.134
0.134 0.199
0.024 0.023
0.005 0.005
0.003 0.003
0.001 0.0007795
0.0002819
0.001 0.0006966
0.0001824


STESI
Soft Tissue Substitute
7.54
59.23
1.97
20.64

10.51



0.11


Element

Hydrogen
Carbon
Nitrogen
Oxygen
Sodium
Magnesium
Silicon
Phosphorus
Sulfur
Chlorine
Potassium
Calcium
Iron
Zinc
Rubidium
Bromine
Zirconium
Niobium









carbon for oxygen is required. Appendix C contains a point-by-point comparison of the

mass energy absorption coefficients for the soft tissue substitute developed here, the

composite soft tissue composition used by Cristy & Eckerman in their mathematical

simulations and the composite soft tissue composition give by the ICRP in the Publication

23 on Reference Man.

Lung Tissue Substitute

The epoxy resin based lung tissue-equivalent substitutes were produced using a

foaming agent and a surfactant in a manner similar to that presented by White (White et al.

1986). A considerable amount of difficulty was experienced attempting to develop a lung

tissue substitute of the proper density. A lung tissue-equivalent substitute, LTES2, was

initially developed for the Phase 2 composite trunk phantom to allow preliminary testing

to begin. The LTES2 lung tissue substitute was made of 42.1% Araldite GY60-10, 16.9%

Jeffamine T-403 (a hardener), 15.0% polyethylene, 14.0% Magnesium Oxide, 10%

Phenolic micro balloons, 1% DC 1107 (a foaming agent) and 1% DC200/50 (a surfactant)

by weight. The density of the LTES2 material was 0.36 g/cm3 which is slightly higher

than the density of 0.296 g/cm3 used in the Cristy and Eckerman model. After many

efforts to decrease the density by various methods, it was discovered that the material

was actually over-foaming and collapsing back upon itself. By reducing the amount of the

DC1107 foaming agent used in the LTES2 lung tissue substitute a material of

approximately the proper density was obtained.

An improved lung tissue substitute, LTES1, was developed for the Phase 3

heterogeneous one-year-old phantom. The LTESI substitute utilizes all the materials

originally used by White, at slightly different concentrations. The LTES2 lung tissue









substitute utilized in the Phase 2 trunk phantom contained a different hardener than that

used in the LTES1 substitute. The LTES1 lung tissue-equivalent substitute is composed

of 34.1% Araldite GY60-10, 17.0% Araldite DY025 (a hardener), 9.5% HY596 (another

hardener), 15.0% polyethylene, 18.5% Magnesium Oxide, 4.4% Phenolic micro balloons,

0.5% DC1107 foaming agent, and 1% DC200/50 surfactant by weight. Table 3 compares

the elemental composition of the LTES2 and LTES1 lung tissue-equivalent materials used

in the physical phantoms with the elemental composition of lung tissue used in the Cristy

and Eckerman computational studies.


Table 3. Elemental Composition of the Lung Tissue Substitutes LTES2 and LTES1 by
Percent Mass.


Element Cristy & Eckerman
(1987)


Hydrogen
Carbon
Nitrogen
Oxygen
Sodium
Magnesium
Aluminum
Silicon
Phosphorus
Sulfur
Chlorine
Potassium
Calcium
Iron
Zinc
Rubidium


10.135
10.237
2.866
75.755
0.184
0.00727
0.00123

0.08
0.225
0.266
0.195
0.00891
0.037
0.0013
0.000942


Lung Tissue Substitute
LTES2
Phase 1 & Phase 2
Phantoms
7.76
61.56
1.76
19.55

8.44

0.84


0.10


Lung Tissue Substitute
LTES1
Phase 1 & Phase 3
Phantoms
8.40
60.77
1.69
17.22

11.16

0.61


0.15









Appendix C contains a point by point comparison of the mass energy absorption

coefficients for the two lung tissue substitutes developed here (LTES 1 with the

appropriate density and LTES2 with the higher-than-optimal density and slightly different

mixing formula), the lung tissue composition used by Cristy & Eckerman in their

mathematical simulations and the lung tissue composition given by the ICRP in the Report

23 on Reference Man.

Bone Tissue Substitute

Finally, a bone-tissue substitute, BTES, approximating the skeletal tissue of a one-

year-old was manufactured. In the 1987 document describing the Cristy and Eckerman

models, the authors made the following observations concerning the skeleton of the

newborn:

"The skeleton of the newborn contains more water, less fat, and less mineral than

the adult skeleton. Furthermore, the distinction of two bone types, cortical and

trabecular bone, is not evident in the newborn skeleton, and the marrow of the

skeleton is all active. Thus it is clear the elemental composition of the adult

skeleton cannot be used when evaluating radiation transport in the newborn."

Consequently, the authors proposed a different skeletal tissue medium for their

newborn phantom, but utilized the adult skeletal tissue medium for all other pediatric

phantoms including the one-year-old. New data on average skeletal atomic composition

as a function of age is now available in ICRP Publication 70 (ICRP 1995). Table 4

highlights mass differences between Cristy and Eckerman's one-year-old skeletal model

and the newer model given in ICRP Publication 70. While some of the differences are

attributable to sorting tissues such as cartilage and teeth into different categories, other









differences are found in the masses of bone mineral and total skeleton. The mass of the

bone mineral is 499 grams in the one-year-old according Cristy and Eckerman, with 200

grams being trabecular bone and 299 grams being cortical bone. This bone mineral mass

is 590 according to the new data in ICRP publication 70. The new data allows for both

the manufacture of age specific skeletal-equivalent materials, as well as the incorporation


Table 4. Tissue Composition and Reference Masses (g) of Skeletal Tissues in the
One-year-old.
Cristy & Eckerman (1987) ICRP Publication 70
(g) (g)
Bone mineral 499 590
Trabecular bone (200)
Cortical bone (299)
Active marrow 150 150
Inactive marrow 20 20
Cartilage -306
Miscellaneous" 469 50

Total skeletal mass 1140 1170
"Miscellaneous tissues include cartilage for the Cristy & Eckerman model while this
category includes teeth, periosteum, and blood vessels for the ICRP Publication 70
model.
b'CRP Publication 70 skeletal tissue dose not include periarticular tissue or blood.


of age specific skeletal media into the Cristy and Eckerman pediatric models to be used in

the Monte Carlo concurrent research.

The skeleton composition (in terms of water, protein, mineral and fat by percent

mass) can be used to calculate an appropriate elemental composition of the one-year-old

skeleton. The mineral percent by weight for a one-year-old skeleton is provided explicitly

in ICRP Publication 70. The water and fat percentages can be obtained from the

document's data tables and the percentage of protein can be obtained by subtraction.









Table 5 compares the skeletal compositions of the Cristy and Eckerman models (newborn

and adult, where the latter is assumed for all ages other than the newborn) to the one-year-

old composition derived from the data of ICRP Publication 70.


Table 5. Anatomic Compositions of Skeletal Tissue (Percent Mass).

Material Newborn One-year-old Adult
Cristy & Eckerman ICRP 70 Cristy & Eckerman
Water 61 52' 33
Protein 17 17.2 19
Mineral 21 23 28
Fat 1 7.8' 19
'Dickerson (1962)


Table 6 then compares the corresponding elemental composition of the Cristy and

Eckerman newborn and adult skeletons, the one-year-old skeletal composition derived

from the data of ICRP Publication 70, and finally the designed composition of the bone

tissue substitute used in this research. Cristy and Eckerman used an 18-element model for

their adult skeleton and a 12-element model for their newborn skeleton. As can be seen in

Table 6 the concentration of the six additional trace elements in the adult skeleton are very

low; therefore, on the basis of the low trace element concentrations in the adult and the

precedent of the 12-element model for the newborn, the 12-element model was utilized for

the one-year-old skeleton analysis. Again, all of these skeletal compositions represent a

homogenized mixture bone matrix and bone marrow. No attempt is made to explicitly

model active and inactive marrow regions within trabecular and cortical bone skeletal

sites. The final one-year-old bone tissue substitute, BTES, contains approximately the

same concentration of calcium as the one-year-old model and has the necessary









substitution of carbon for oxygen required in most physical phantom modeling. The

density of this mixture is 1.18 g/cm3.


Table 6. Elemental Compositions of Skeletal Tissue (Percent Mass).


Element Cristy & Eckerman
Newborn
Hydrogen 7.995
Carbon 9.708
Nitrogen 2.712
Oxygen 66.812
Sodium 0.314
Magnesium 0.143
Phosphorus 3.712
Sulfur 0.314
Chlorine 0.140
Potassium 0.148
Calcium 7.995
Iron 0.008

Additional trace elements
for adults

Fluorine
Silicon
Zinc
Rubidium
Strontium
Lead


Cristy & Eckerman
Adult
7.337
25.475
3.057
47.893
0.326
0.112
5.095
0.173
0.143
0.153
0.190
0.008


ICRP 70
One-Year-Old
7.840
14.973
2.798
60.459
0.346
0.161
4.211
0.346
0.154
0.163
8.541
0.009


0.025
0.002
0.005
0.002
0.003
0.001

Phantom Constniction


Phase 1 Cylindrical Phantoms

Simple cylinders of epoxy resin-based tissue-equivalent were constructed for a

proof-of-concept benchmark study between physical measurement of radiation absorbed

dose and Monte Carlo simulations. Five Phase 1 cylindrical phantoms each simulating one

of three different tissue types were used; the bone tissue substitute discussed previously,


Bone Tissue
Substitute
5.86
52.98
2.03
24.54




5.79

8.81









the two lung tissue substitutes of differing densities, and two versions of the soft tissue

substitute. The first soft tissue substitute version, STESI, has been discussed previously

and utilizes magnesium oxide as the main filler. The STESI substitute was used in both

the Phase 2 composite trunk phantom and the Phase 3 heterogeneous one-year-old

phantom. The second soft tissue substitute version, STES2, was a preliminary version

referenced in several funding application documents and used in Chapter 5, "Results and

Discussion Associated with the Phase ljhantom," to highlight certain utilities of the

Phase 1 cylindrical phantoms. The STES2 soft tissue substitute utilizes aluminum oxide as

the filler material and had a density of only 0.889 g/cm3 as compared to a density of 1.04

g/cm3 given in Cristy and Eckerman (1987) for their soft tissue media. Nevertheless, the

mass energy absorption coefficients were still within the desired 3% limit. The STES2

Phase 1 soft tissue-equivalent cylindrical phantom had a lower than desired density and

utilized aluminum oxide as a filler material rather than magnesium oxide simply because it

was manufactured during the early phases of the research before the STES1 version of

soft tissue substitute was developed. The utilization of a slightly lower density phantom is

acceptable for comparing physical and mathematical estimates of dose in the initial

benchmark study as long as the same density is utilized in the corresponding Monte Carlo

simulation model. The STES2 Phase I cylindrical phantom consisted of a mixture of

Araldite GY60-10 (46.28%), Jeffamine T-403 (18.51%), aluminum oxide (15.40%),

medium density polyelythylene powder (15.00%), and phenolic microballoons (4.81%).

The final product had a density of 0.889 g/cm3 and an elemental composition of carbon

(60.91%), oxygen (21.25%), aluminum (8.15%), hydrogen (7.65%), nitrogen (1.9307%)

and chlorine (0.11%).









A 5.2 cm (2.04 inch) diameter PVC pipe split lengthwise into three sections was

used along with an end cap as the mold for the Phase 1 cylindrical phantoms. The PVC

was waxed between each use to prevent the epoxy from adhering to the wall of the PVC.

The cured Phase 1 phantoms each had a diameter of 5.2 cm with a range of lengths from

5.0 cm for the LTES1 lung substitute to 8.4 cm for the STES2 soft tissue substitute. The

length was rather arbitrary depending on how much material was mixed and what other

molds were being filled at the time the Phase 1 phantoms were created. A small hole was

drilled in the center of one end of each of the Phase 1 phantoms to allow the insertion of

MOSFET dosimeters.

The five Phase 1 cylindrical phantoms were then placed in horizontal positions

during x-ray irradiation with the central axis of the cylinders positioned 90 cm from the

estimated focal spot. A general purpose three-phase x-ray system with 2.8 mm Al

inherent filtration (Philips Maximus C850) was used to produced the x-ray field. The

maximum voltage and current available were 150 kVp and 800 mA, respectively. A

Radcal Corporation radiation monitor, model 1015C, utilizing a 6 cm3 ion chamber,

model 10X5-6, was used to measure the exposure produced by the x-ray system. The

Phase 1 cylindrical phantoms were irradiated at tube potentials ranging from 40 to 140

kVp at 10 kVp intervals. The tube current and time settings were held constant during the

irradiation.

Phase 2 Composite Trunk Phantom

The Phase 2 composite trunk phantom was created by manufacturing individual

cylinders to represent the lungs and spine. These were then positioned in a cylindrical

mold representing the trunk of a one-year-old child and the intervening space was filled









with soft tissue-equivalent substitute material. The lung and spine cylinders are clearly

shown in the partially completed phantom, Figure 16. The soft tissue material was

poured in layers approximately 7 to 10 cm thick. White et al. (1977) noted the potential

of a build up of heat from the chemical curing process which could lead to charring of the


Figure 16. Partially Completed Phase 2 Composite Trunk Phantom with Lung and Spine
Cylinders Exposed









interior of the tissue substitute. He recommended using layers less than 10 cm thick to

preclude this potential problem. For this work it was discovered that layers 7 to 10 cm

thick contained an easily manageable amount of material which did not begin to harden

before the air bubbles had been removed. Charring or overheating was not noted;

however, on the thicker layers a very slight increase in temperature could be felt on the

outside of the mold. The method of layering also allowed a soft tissue substitute layer to

be used to secure the interior components rather easily, usually a single interior component

could be isolated by insertion into phantom in this manner.

The physical dimensions of the Phase 2 composite phantom were very generally

modeled after the specification of the one-year trunk given in Cristy & Eckerman (1987).

The lengths of the lungs, spine and trunk were consistent with the Cristy & Eckerman

model; however, simple right circular cylinders were used in the Phase 2 phantom rather

than the elliptical cylinders and more complicated lung shape of the Cristy & Eckerman

model to ease initial manufacturing and minimize analytical complications in this

preliminary phantom. The choice of the radius of the lungs and spine were limited by the

available sizes of PVC pipe which was used as a molding material. PVC pipe was used as

a molding material because it was readily available and had been used in the Phase 1

cylindrical phantoms. Tables 7 and 8 compare the dimensions and volumes of the Cristy

and Eckerman model with those used in the composite trunk phantom. Schematic

diagrams of the Phase 2 composite trunk phantom illustrating the positions of these

individual regions are shown in Figures 17 (front view) and 18 (top view).









Table 7. Dimensions and Volumes of the Cristy & Eckerman Mathematical Phantom.

Cristy & Eckerman (1987)
Model Elliptical Cylinders

Region Left to Right Semi-Axis Anterior to Posterior Length Volume
(cm) Semi-Axis (cm) (cm3)
(cm)
Lungs 2.68 4.88 10.53 288
Spine 0.88 1.63 28.36 128
Trunk 8.8 6.5 30.7 5517



Table 8. Dimensions and Volumes of the Phase 2 Composite Trunk Phantom.
Region Radius Length Volume
(cm) (cm) (cm3)

Lungs 2.54 10.5 213

Spine 1.15 28 116

Trunk 7.62 30 5472



Following completion of the composite one-year trunk phantom, four 0.36 cm

(9/64 inch) diameter, holes were drilled to allow placement of the MOSFET dosimeters

from the top of the phantom. The MOSFET guide holes were positioned along the central

axis of each lung, along the central axis of the spine and in the soft tissue approximately at

the center of a triangle form between the lung and spine. Five 0.36 cm (9/64 inch)

diameter guide holes were also drilled in the bottom of the phantom. One of the guide

holes was located along the central axis of the spine in the bottom and three others were

located in the soft tissue along the same axes described previously. The final guide hole

was located in the soft tissue opposite the bone.







2.35 cm
I I


30 cm


Lung


5.2 cm


Figure 17. Side View of the Phase 2 Composite Trunk Phantom


Lung

--5.2 cm
S5.2 cm


19 cm


- --









Spine


Figure 18. Top View of the Phase 2 Composite Trunk Phantom


Table 9 gives a description of the location of each dosimeter and their respective

depth in the phantom. A total of five dosimeters, two high-sensitivity MOSFETs and

three therapy MOSFETs, were available for use for any single exposure. Consequently,

for each combination of incident beam kVp and projection view, two exposures were

performed one for the top set of four MOSFETs, and one for the bottom set of five

MOSFETs. Phantom exposures were performed for AP, PA, and Left Lateral views at

tube potentials of 60, 90, and 120 kVp using the same Phillips Maximus C850 machine

utilized with the Phase 1 cylindrical phantom irradiations.









Table 9. Positioning of the MOSFET Dosimeters within the Phase 2 Composite Trunk
Phantom.

Detector Location within the Distance from Distance from
Designation Phase 2 Phantom phantom's top phantom's
bottom

1 Center of right lung region 6 cm
2 Center of upper spine region 6 cm
3 Center of left lung region 6 cm
4 Soft tissue region central to lung and 8 cm
spine regions
A Soft tissue region below right lung 10 cm
B Center of lower spine region 15 cm
C Soft tissue region below left lung 8 cm
D Soft tissue region central to below 16 cm
lungs and spine regions
E Soft tissue region opposite lower spine 6 cm


Monte Carlo Simulation Studies

In this phase of the study, computational models of the Phase 1 cylindrical

phantoms, the Phase 2 composite trunk phantom and the Phase 3 heterogeneous one-year-

old phantom were created for use with the Electron Gamma Shower (EGS4) Monte Carlo

radiation transport code. EGS4 is a general-purpose code which allows one to simulate

the coupled transport of photon and their secondary electrons within any user-defined

compound or mixture (Nelson and Hirayama 1985). Considering the low energy of

diagnostic x-rays, secondary electron transport was considered unnecessary in these

simulations. Lionel Bouchet, University of Florida doctoral candidate, performed all the

EGS4 code modeling and manipulations associated with this research. Bouchet is

conducting Monte Carlo research with some assistance by Ricardo Reyes, University of









Florida Graduate Student, under the supervision of Dr. Wesley Bolch at the University of

Florida.

For benchmarking the proof-of-concept phantoms, mathematical models of the

5.2-cm diameter Phase 1 cylindrical phantoms were encoded within EGS4 using photon

cross sections based upon the tissue composition given above. A 1-cm diameter sphere

was located on the central axis at the MOSFET location in which the absorbed dose was

assessed for each irradiation simulation. Since exact spectra were not available for the

system, the computer program XCOM5R (Nowotny, Hifer 1985) was employed to

calculate the x-ray spectra of the Philips Maximus C850 machine. More information on

XCOMSR can be found in Appendix B. The program provides the spectra for constant

potential x-ray systems with various parameters (kVp, anode angle, distance, and various

absorbing materials) from tungsten anodes. Estimates of tissue dose per unit entrance

exposure (dose conversion coefficient) were determined for a variety of tube potentials for

subsequent comparison with the MOSFET measurements.

For benchmarking the Phase 2 composite trunk phantom, a mathematical model

was constructed within the EGS4 system based upon the specifications outlined in Figures

16 and 17, as well as in Table 9. As with the earlier Phase 1 cylindrical phantoms, small

spheres were placed within the mathematical model at locations mimicking the positions of

the MOSFETs within the physical phantom. In this case, sphere sizes of both 0.5-cm

diameter and 1.0-cm diameter were created, with no discernible difference in the results.

Energy deposition as well as tissue dose were determined for each sphere location and

size, and for the entire soft-tissue, bone, and lung regions within the phantom. Ten million

particles were used for each simulation.









Phase 3 Heterogeneous One-Year-Old Phantom

The construction of the Phase 3 heterogeneous one-year-old phantom was a

multiple step process. First molds were created for the individual skeleton components

and lungs. Second, the molds were filled with the liquid resin tissue substitutes discussed

previously and allow to cure for 48 hours. Third, the cured components were milled to

final shape. Once the internal components were completed, molds were created for the

external parts, the trunk, legs and head. The soft tissue substitute material was then

poured into the external molds in the layered method discussed for the Phase 2 composite

trunk phantom. The layered method facilitated the placement of the skeletal and lung

components in a manageable fashion. By adjusting the thickness of the layers, the

components were usually placed in the appropriate location on the solid cured layer of soft

tissue. Another layer of soft tissue substitute material was then poured into the external

mold around the components firmly fixing them in place. Finally, the external components

were milled to the proper size.

Skeleton

The body skeleton consisting of two leg bones, two arm bones, a pelvis, a spine, 4

of the 12 ribs, two scapula, and two clavicles can be seen in Figure 19. The manufacturing

process for each bone type is outlined below. The head model shown in Figure 19 is not

bone; it is a wood and plaster cast shown here for reference only. The head bones were

not completed prior to the insertion of the body skeleton into the external mold; therefore,

they were not available for photographing at the time the body skeleton was

photographed. When the head bones were available the body skeleton was already

installed in the trunk of the phantom. The head is modeled after the Bouchet-Bolch








mathematical head model described in "Revised dosimetric models of the pediatric head
and brain" (Bouchet et al. 1997) which was developed at the University of Florida. It is
discussed separately in a following section but it contains the following skeletal structures:
a cranium, an upper face region, teeth and a mandible.


Figure 19. Physical Skeleton Model with 4 Ribs and a Wood and Plaster Cast of the Head.


Legs bones
The mathematical equations for the legs bones given by Cristy and Eckerman are:

r 4. 142 (1.54 +0.03 and -26.42 0 (Eq. 6)
x44+261 +y <[1.54+0.03z2 and 26.42z 0 (Eq. 5&6)
1 26.42 ad-6,2_z









The equations are for a frustum of a circular cone with the plus sign for the right leg and

the minus sign for the left leg. In the physical construction of the leg bones the circular

cone is cut to the angle of the bottom of the phantom and is therefore not a true frustum.

This slight modification is not expected to produce a noticeable effect on the measurement

of absorbed dose.

A circular cone wooden leg bone was first created on a lath and this wooden leg

bone was used as the male component to create the female mold. A variety of products

were evaluated for an appropriate mold material. A desired material would have good

release characteristics and would be readily available at a reasonably low cost. It was also

hoped that a reusable mold could be found. Fiberglass was used initially by wrapping the

wooden leg with cellophane and attempting to coat the outside with fiberglass. The

fiberglass failed to collapse around the wooden leg properly, the wall thickness was not

sufficient to allow splitting and rejoining the mold for removal of the finished leg, and the

process was awkward to work with; therefore, this process was abandoned prior to

pouring any actual leg bones. The second material evaluated was expandable foam

insulation. A rigid mold was made by utilizing a section of PVC pipe split lengthwise and

the foam insulation was sprayed around the wooden leg bone again covered with

cellophane. Apparently the foam requires direct access to the air as the material at the

bottom of the mold did not expand and dry as expected until the mold was opened. The

foam method did allow the creation of useable molds. One leg and one arm were created

with this type of mold. The epoxy resin had an affinity for finding breaks in the surface of

the insulation and filled a large portion of the air pockets which were outside the desired

bone location with tissue substitute material. The resulting bones required a large amount









of milling and the foam method was eventually abandoned for an even simpler method.

The final method used to mold the leg bone utilized paraffin wax as the filling material,

melted around the wooden leg within the PVC pipe. This method produced acceptable leg

bones; however, the mold was partially destroyed in extracting the bone.


Arms
The mathematical equations for the arm bones given by Cristy and Eckerman are:

(0.0102 (z- 30.26) +(x 8.)Y (y 2 (60.52 +(z -30.26)
.62 ) 1.76) 60.52 )

and 0 z s 30.26 (Eq. 8)

Equation 7 is incorrect, however, and should contain a "" prior to the 0.0102 term. The

equation as given by Cristy and Eckerman creates a slight asymmetry in the positioning of

the arm bones. The error was detected prior to setting the arm bones such that the bottom

of the arm bones are in the correct position, i.e. symmetrical. An unrelated error placed

the top of the physical arm bones approximately 0.3 cm closer to the center of the

phantom than the mathematical bones are positioned; the misalignment is consistent

between the left and right arm bone and therefore symmetry is maintained. It is not

expected that this slight misalignment will affect the final results.

The arms were created in a manner similar to that described for the leg bones.

First, a wooden model was created and used as the male component in the molding

process. Second, a female component was constructed using a filler material. Third, the

bone tissue-equivalent material was poured into the female component of the mold and

allowed to cure. Finally, the arm bone was removed and any necessary milling was

performed.









An additional molding method was examined for one of the arm bones. Plaster

was used as the filler material in the PVC type of mold previously described. The method

worked well with the flatter arm bone, as the plaster was only needed on the bottom and

sides. A rubberized molding material was used to create a boundary between the plaster

and the epoxy resin tissue substitute and a commercial mold release material was used

prior to pouring the arm bone. Unfortunately the mold was destroyed in the removal

process. Figure 20 shows the arm bones prior to being installed in the Phase 3

heterogeneous one-year-old phantom along with the wax mold used to create one of the

arm bones.


Figure 20. Arm Bones with a Plaster Type Mold


The mathematical equations for the spine given by Cristy and Eckerman (1981)


are:









]2 + 3 2 1 and 9.65Sz< 38.01 (Eq. 9& 10)

The spine was created with the paraffin wax method described in the leg bone section.

The spine was created as one continuous bone for the physical phantom. The head model

of Bouchet and Bolch was created separately from the Cristy and Eckerman model.

Consequently, there exists a slight discontinuity between the spine within the neck and the

spine within the trunk. In a effort to make the phantom more transportable the head and

neck model was made separate from the trunk. The neck is attached to the trunk but the

spine is not permanently affixed in the neck. The head model is quickly and easily aligned

utilizing space in the neck for the spine and it is easily removable for transport.

Pelvis
The mathematical equations for the pelvis given by Cristy and Eckerman are:


(52 (Yl9 2 11 for the outside and (Eq. 11)


(x + 2.47 1 2 fortheinside (Eq. 12)

y > -1.95 0
New mold construction methods were needed for the remaining components of the

phantom. Figure 21 shows the pelvis with the pelvis mold. This mold utilizes 0.05 inch

thick polyethylene sheet which was originally sold at a local home improvement store as a

device for holding yard waste bags. A wooden base was first cut out on a scroll saw

according to the dimensions of the equation given above (with allowances made for the

thickness of the polyethylene). The polyethylene was then cut to fit and attached to the




























Figure 21. Pelvis with Mold


wooden base. After many trials it was discovered a hot glue gun was the best method to

attach the polyethylene to the wood. The polyethylene had excellent release properties

which was desirable for removing the finished product but was a nuisance when

attempting to attach the polyethylene to wood. The hot glue was a quick and efficient,

nonpermanent method. When the final product was due to be removed, the polyethylene

could be separated from the wood with only a moderate amount of force and the mold did

not leak during the curing process. The second curve was cut into the scrap wood and

another piece of polyethylene was cut to fit. A bottom was attached and the mold filled.

After the bone tissue substitute had cured, it was removed and the bottom section was

removed from the pelvis. A minor amount of milling was performed using a disk sander.









Claicle
According to Cristy and Eckerman, the clavicles are two portions of a torus which

lie along a circular arc:

x + (y 1.38)2 = 50.9796 (Eq. 16)

at z = 29.93 cm with a radius of 0.393 cm. The clavicles lie between the planes

1.38-y= lx*5.68 and 1.38-y= x *0.43 (Eq. 17)

The mold for the clavicles seen in Figure 22 was created by simply routing out a

grove along the given circular arc. The completed clavicles are also shown in Figure 22.

A piece of tubing was used as the actual mold by filling the tube with bone tissue material

and clamping it to the grove for curing. The mold worked quite well; however, the curing

time was increased with no direct air access and the clavicles were removed prematurely.


Figure 22. Clavicles with Partial Mold








The clavicles then cured slightly out of the desired shape. It was discovered that by

heating the material for a few seconds in a microwave oven, it became pliable enough to

bend it back into shape. The cooled material was fully rigid and no adverse effects were

noted.

Scapula
The mathematical equations for the scapula given by Cristy and Eckerman are:


(i2 1 oude and + >1 (Eq. 18& 19)

The scapula are contained within the volume between two concentric elliptical cylinders.

For each scapula, the volume is bounded by the planes z = 22.32, z = 29.52, y = 0.37*lxl,

and y = 1.18* x|. The mold for the scapula, Figure 23, was created as described


Figure 23. Scapula and Mold








previously for the pelvis. The scapula was poured before the advantages of hot glue were

discovered. The polyethylene was attached with quick drying epoxy as well as contact

cement. As can be seen in the view of the mold in Figure 23, the mold did not maintain its

seal well during the curing process. The scapula which was in left mold when the seal

broke contained a much lower density, even though the outside dimensions were similar,

than the scapula from the right side which maintained its seal. The low density scapula

had a considerable amount of visible air bubbles and had to be recast.

Ribs
The mathematical equations for the ribs given by Cristy and Eckerman are:


(I 2 + 6. 1 ) for the outside and (Eq. 20)


i + 6 2 l I for the inside (Eq. 21)

The mold for the ribs consisted of an elliptical wooden base with polyethylene

attached around the circumference, a polyethylene spacer was then placed at the bottom of

the mold and a final outer layer of polyethylene was added around the circumference of

the spacer. All twelve ribs were poured and allowed to cure in one piece, then the

individual ribs were cut off the mold using a hand held rotary power tool with a cutting

wheel attached. Figure 24 illustrates part of the rib mold with some of the individual ribs

which have already been removed from the mold.

The inside polyethylene layer was destroyed by the rotary power tool since the ribs

were so thin and had to be cut completely to prevent breakage; therefore, it was simplest

to cut into the polyethylene.

























Figure 24. Ribs with Partial Mold Still Containing Some Ribs


Lungs

The mathematical equations for the right lung given by Cristy and Eckerman are:

(x+3.74) 2 y V (z-19.08S2
2.68 / 103 1
if 20.10 z 5 24.60 and y < 0.70; then x < -2.90 (Eq. 24, 25 & 26)

for the left lung;

x -3.74 y z z19.082
2.68J .88) 10.53 1 and z19.08 (Eq. 27&28)

if 19.08 < z < 24.80 and y < 0.40; then x 2 3.90 (Eq. 29, 30, 31)

The above equations describe a half ellipsoid with a section removed from each

lung. The section removed from the left lung is larger than the section removed from the

right lung. The shape of the section removed can be seen in Figure 25 for the left lung,









the shape was similar for the right lung except it did not protrude as far into the lung and

does not extent to the base of the lung. The physical lungs were created by cutting a pair

two dimensional elliptical wooden bases using the equations with z = 19.08. A hole was

cut into the center of the top base to allow pouring of the lung tissue substitute. The

polyethylene sheet described previously was cut to fit forming an elliptical form. Wooden

spacers as shown in Figure 25, were cut and covered with the polyethylene sheeting to

provide the removed sections. The spacer were attached in the proper location, as

described by the cutting planes above, to the inside surface of the polyethylene mold. The

lung tissue substitute LTES 1 was poured into the form and allowed to expand and cure.




















Figure 25. Left Lung with Simple Mold and Wooden Spacer


The lungs were removed from the form after 48 hours and finish milled to the

appropriate size. Figure 25 shows a completed lung with the simple mold and the wooden

spacer. The heart is slanted to the point where a larger section is removed from the left









lung as can be clearly seen. The milling method produced lungs of high quality; however,

air pockets are exposed in this method. The exposed air pockets are filled with soft tissue

substitute when the lungs are installed. Since some of the air pockets are relatively large,

an attempt was made to make a plaster casting of one of the finished lungs and use the

plaster casting to pour a lung with a closed membrane on the surface. Unfortunately, the

casting was not perfect and the final lung required milling as well.

The milling was performed by cutting thin cardboard templates representing

approximately 1 cm steps in the z direction of each lung and utilizing these templates as

guides to obtain the proper shape. Completed lungs and lung molds are seen in Figure 26.

The difference in the size of the cut out portions between the lungs are clearly shown.


Figure 26. Lungs and Lung Molds, Left Lung has a Larger Section Removed









Head

The head was the most challenging aspect of the Phase 3 heterogeneous one-year-

old phantom. The Bouchet-Bolch head model is more complicated than the Cristy and

Eckerman head model and the three dimensional curvature of the cranium precluded the

use of the polyethylene sheeting which had been used so advantageously on the previous

components. The complexity of the head model can be seen in Figure 27. The major

portions modeled for this phantom are the upper and lower portions of the cranium, the

upper face region, the teeth, and the mandible and the overall external surface of the head

and neck. The spine region has been discussed previously under the spine subsection of

the skeleton section.

A wood and plaster model of the head was created to assist in the development of

the physical manifestation of the Bouchet-Bolch head model. The model was created by

first printing full scale templates of the head equations. See Appendix D for an example of

one of these templates. Wooden layers inch thick were then cut out utilizing the

template guides. The wooden layers were assembled, creating a stair step model with one

edge of the wooden layer the correct circumference and the other edge somewhat less.

Plaster of Paris was used as a filler to obtain the required curved surfaces. The plaster is

easily sculpted when relatively wet and can be sanded when dry. The completed wood

and plaster head model is shown in Figure 28. The mold utilized for the lower head is also

shown in Figure 28. The upper-face region, mandible, teeth and lower section of the

cranium were all created independently, attached and placed in the lower head mold where

soft tissue substitute material was added to represent the brain and other soft tissue in the

head.











Cranium


Upper-face Region


S Spine










Teeth Mandible


Figure 27. Bouchet-Bolch Head Mathematical Head Model


Thyroid




























Figure 28. Lower Head Mold with Wood and Plaster Model of Head


The mathematical equation for the ellipsoid top of the Bouchet-Bolch head model

is:


4+ + 6 -1 for z>2 44.06 (Eq. 32)

For the section below the crown of the head, 32.91 < z < 44.06, a cylinder is used

to describe the front with a cutting cone describing the back and two back planes joining

the two. The equations given for the cylinder and cutting cone are:


1 for y < 3.925 cylinder (Eq. 33)


+2 J 2 0.043(z -44.06)+1 for y > 3.925 cutting cone (Eq. 34)









The lower portion of the head is relatively uniform with a cylinder and a cutting

cone. A wood and polyethylene mode was created using the dimensions given in the

equations above and the process described previously. This external mold for the bottom

portion of the head is shown in Figure 28 with the wood and plaster head model.

The head was created by manufacturing the cranium in separate top and bottom

halves. Other individual components manufactured are the upper face region, teeth and

mandible. The top of the cranium was filled with soft tissue substitute creating a

hemispherical section of the top of the head. The upper face region, teeth and mandible

were attached to each other with a small portion of soft tissue substitute. The lower

hemisphere of the cranium and the mandible-teeth-face section were placed in the head

mold and clamped into place, a moderate amount of soft tissue substitute was poured into

the cranium until a sufficient amount attached to the front section and the material was

allowed to cure. A bottom portion of the head mold was created from a polyethylene

sheet with a hole cut for the insertion of the spine. The liquid material filled the voids

under the cranium, around the front section and over the cranium. The soft tissue

substitute was allowed to cure and the top section was attached again using soft tissue

substitute as the adhesive medium.


Cranium
The top half of the cranium was created by using the wood and plaster head model

described previously as the male section to create a female mold from plaster. The rubber

molding material discussed in the "arms section" was used to coat both the surface of the

head model and the female mold.









A spacer 0.21 cm thick, the desired cranium thickness, was placed between the

two portions. A commercial mold release was utilized in an attempt to save the mold but

with little success. Both sections of the mold were destroyed in the removal process. The

cranium thickness is approximately 0.21 cm thick and no effort was made to add the 0.08

cm thick skin of the Bouchet-Bolch model.


Figure 29. Top Half of Cranium Mold in Two Parts


A new method was required for bottom section of the cranium since the wood and

plaster model was destroyed creating the top portion. A thin, approximately 0.21 cm, flat

disk was poured on a cellophane sheet placed in a large round form. Before it had

hardened completely it was removed and formed around the existing top half of the

cranium. A triangular shaped piece of material had to be removed to allow the disk to be

shaped around the cranium properly. The method was very successful and considerably

easier than the method described previously. The bottom portion of the cranium was









milled to fit the upper face region after it had cured. Two small sections were removed at

the top of the connection between the cranium and the upper face region to provide access

for the soft tissue to reach the lower portions of the head.


Upper Face Region
The mathematical equations for the upper face region given by Bouchet et al.

(1997) are:


( 2 )2 <1; 37.31 z 38.85-0.67y and y5-3.21 (Eq. 34&35)


The upper face region and the mold use are shown in Figure 30. The upper face

region was poured in a single hemispherical section and the angled portion was milled on a

disk sander after the material had cured.


Figure 30. Upper Face Region with Mold








Teeth
The mathematical equations for the teeth given by Bouchet et al. (1997) are:
( x y + 3.21 y + 3.21
12 <1 and +( > 1 (Eq. 37 & 38)

This simple mold was fashioned similar to that described for the pelvis and other

bones. The teeth and teeth mold is shown in Figure 31.


Figure 31. Teeth and Accompanying Mold


Mandible
The mathematical equations for the mandible given by Bouchet et al. are:

Front: d2 +y 3.21) -<1 and 2 + 3.2 1 (Eq.
S -3.21 a 0.325 337 la z (Eq.

y -3.21 and 0.0625y + 33.27 z 35.29 (Eq.


39 & 40)

41 & 42)


Back: 321) and 2 -(-3.21)21 (Eq. 4344)
-(-(3-)- 31221) 2 1 (Eq. 43 & 44)
Back: and 4. 2E 6.18 -




Full Text
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 EONPBPW61_K87FCK INGEST_TIME 2013-09-27T22:16:21Z PACKAGE AA00014246_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES