BREAST DOSIMETRY IN CLINICAL MAMMOGRAPHY By LUIS ALBERTO DO REGO BENEVIDES A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2005
Copyright 2005 by Luis Alberto do Rego Benevides
I dedicate my dissertation to my ch ampion, companion and most beloved spouse Sandra Julie (Villanueva) Benevides; my l oving parents, Luis Alberto Gomes Machado Benevides, and Noemia Fernades (Borge do Rgo) Benevides; and in loving memory of one taken so young Mary Cruz (Villanueva) Contreras. I have stood on shoulders of giants but no one greater than my wife. Sh e has given me the strength, encouragement, and assistance and has been my island of so lace in the sea of despair throughout this journey. My fledgling dream would not ha ve taken flight without her constant unconditional love and dedicati on. I dedicate this life's achi evement to her with all my enduring love. Aos meus pais, esta tese foi feita gra as ao vosso sacrifcio e dedicao aos filhos. Agora sou doutor de doutores. Ela dedicada com grande agradecimento e amor.
iv ACKNOWLEDGMENTS The academic achievement that this dissert ation represents is the result of a long journey that nears an end. At journey's end, I reflect on the trail of dedicated professionals that have shared their knowledge and experience to fan my inspirations. I would like to express my sincerest gratitude to my committee chairman, Dr. David E. Hintenlang, for his friendship, unrelenting confidence, enthusiasm and mentoring throughout my stay at the University of Flor ida. I would also like to thank my cochairman, Dr. Libby F. Brateman, for her mentoring, friendship, and thought provoking conversations. I would like to thank my committee members, Dr. Wesley E. Bolch, Dr. Manuel M. Arreola, and Dr. Di etmar W. Siemann for their encouragement and invaluable guidance. I would also like to thank Dr. Albert F. Eble, Trent on State College, Trenton, NJ; Dr. Jack Fresco, Princeton University, Pr inceton, NJ; Dr. William F. Blakely, Armed Forces Radiobiology Research Institute, Beth esda, MD; and CAPT (ret.) Paul K. Blake, MSC, USN for the instrumental parts they played at various crossroads along this journey. I would like to acknowledge th e contributions of Thomson and Nielsen Electronics Ltd., Ottawa, Canada, for providing the MOSFET models and very-high sensitivity bias supply used in this study. Specifically would like to thank Dr. Addelbasset Hallil, Thomson and Nielsen Electronics Ltd., for his technical advice in th e utilization of the dosimeters.
v I would also like to acknowledge the cont ributions of Naval Research Laboratory, Washington, District of Colu mbia, for providing the protot ype dosimeters and software used in this study. Specifical ly would like to thank Dr. Alan L. Huston for his technical advice in the utilization of the dosimeters. I would like to thank my colleagues fr om the Department of Nuclear and Radiological Engineering and U.S. Naval Hosp ital Jacksonville from their contributions in this research. There are many others i nvolved in this endeavor, too numerous to mention, but each has contributed a great more than I can ever repay. I would be remiss not to thank the taxpayers, United States Government, Department of Defense, and Department of the Navy for giving me the uniqu e privilege to complete one of my life's achievements. The retrospective population study was re viewed and approved by the U.S. Navy Institutional Review Board (B04LHOOOOO-052) and the Hea lth Center Institutional Review Board (220-2004) at the University of Florida. The population study was conducted in accordance with Health Insura nce Portability and Accountability Act of 1996 standards. The dissertation was part ially supported by a grant from the U.S. Department of Energy, Nuclear Engineer ing Education Grant Program, DOE-DE-PS0702ID14200. The views expressed in this disser tation are those of the author and do not reflect the official policy or position of the Department of the Navy, Department of Defense or the U.S. Government.
vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................xi LIST OF FIGURES.........................................................................................................xiv ABSTRACT.....................................................................................................................xi x CHAPTERS 1 INTRODUCTION........................................................................................................1 2 BACKGROUND..........................................................................................................7 Breast Anatomy and Physiology..................................................................................7 Breast Carcinogenesis...................................................................................................9 Historical Perspective: Efficacy to Oversight.............................................................10 Risk and Mammography.............................................................................................14 Efficacy of Mammography..................................................................................14 Risk Associated with Mammography Exposures................................................15 Lifetime Risk from Mammography.....................................................................16 Mammography............................................................................................................18 Photon-Tissue Interactions..................................................................................18 Mammography Unit Characteristics....................................................................18 Breast Tissue Density.................................................................................................22 Density Measurement..........................................................................................23 Tissue Compression.............................................................................................24 Tissue Characteristics..........................................................................................25 Breast Dosimetry Reviews.........................................................................................25 Hammerstein et al., 1979.....................................................................................26 Stanton et al., 1984..............................................................................................28 Dance, 1990.........................................................................................................29 Wu et al., 1991, 1994...........................................................................................29 Sobol and Wu, 1997............................................................................................30 Dance et al., 1999................................................................................................31 Simulating Breast Tissue............................................................................................31 Population Demographics...................................................................................31
vii Geise and Palchevsky, 1996.........................................................................32 Dance et al., 1999.........................................................................................33 Maskarinec and Meng, 2000........................................................................33 Rosenberg et al., 2001..................................................................................33 Kruger and Schuler, 2001.............................................................................34 Other researchers ........................................................................................34 Volumetric Breast Measurements: Fife, 1990.....................................................34 Protocol Tools.............................................................................................................35 Tissue Equivalent Material..................................................................................35 Dosimetry Devices..............................................................................................36 Metal oxide semiconductor field efffect transistor ...................................36 Fiber optic-coupled dosimeter ...................................................................37 Monte Carlo Method...........................................................................................38 3 CHARACTERIZATION OF MOSFET DOSIMETERS FOR APPLICATION IN CLINICAL MAMMOGRAPHY................................................................................55 Introduction.................................................................................................................55 Material and Methods.................................................................................................58 MOSFET and Patient Dose Verification System................................................58 Diagnostic Radiography Units............................................................................59 General Electric Senographe DMR..............................................................59 Picker Condenser Discharge Mobile Radiography Unit..............................60 Exposure Measurements......................................................................................60 Dosimeter Angular Response..............................................................................60 Dosimeter Sensitivity..........................................................................................61 Dosimeter Linearity.............................................................................................62 Dosimeter Reproducibility..................................................................................62 Results and Discussion...............................................................................................62 Dosimeter Angular Response..............................................................................62 Dosimeter Sensitivity..........................................................................................63 Dosimeter Linearity.............................................................................................64 Dosimeter Reproducibility..................................................................................64 Conclusions.................................................................................................................64 4 CHARACTERIZATION OF FIBE R-OPTIC-COUPLE D DETECTOR FOR DOSIMETRY IN CLINICAL MAMMOGRAPHY..................................................86 Introduction.................................................................................................................86 Materials and Methods...............................................................................................90 Fiber-Optic-Coupled Dosimeters (FOCD)..........................................................90 X-Ray Field.........................................................................................................91 Exposure Measurement.......................................................................................91 Dosimeter Reproducibility..................................................................................91 Dosimeter Angular Response..............................................................................92 Dosimeter Sensitivity..........................................................................................92 Dosimeter Linearity.............................................................................................93
viii Results and Discussion...............................................................................................93 Dosimeter Angular Response..............................................................................93 Dosimeter Sensitivity..........................................................................................94 Dosimeter Linearity.............................................................................................94 Dosimeter Reproducibility..................................................................................94 Conclusions.................................................................................................................94 5 A BREAST TISSUE EQUIVALENT PH ANTOM SERIES FOR USE IN CLINICAL MAMMOGRAPHY..............................................................................127 Introduction...............................................................................................................127 Materials and Methods.............................................................................................130 Phantom Manufacturing....................................................................................130 Results and Discussion.............................................................................................135 Conclusions...............................................................................................................138 6 ANTHROPOMETRIC VARIATIONS IN MAMMOGRAPHY.............................155 Introduction...............................................................................................................155 Materials and Methods.............................................................................................156 Results and Discussion.............................................................................................162 Conclusions...............................................................................................................169 7 ESTIMATING BREAST GLANDULARITY.........................................................196 Introduction...............................................................................................................196 ACR BI-RADS Method....................................................................................197 Planimetry Method............................................................................................199 Histogram Threshold Method............................................................................199 Tube Loading Method.......................................................................................200 Volumetric Method...........................................................................................201 Method Precepts.......................................................................................................201 Materials and Methods.............................................................................................203 Study Population...............................................................................................203 Population Study Selection Criteria..................................................................203 Mammography Imaging....................................................................................204 Mammography Facility and Film Processing....................................................205 Film Digitizer....................................................................................................205 Image Segmentation..........................................................................................206 Monte Carlo Modeling......................................................................................207 Breast Tissue Composition Estimating Methods..............................................208 ACR BI-RADS method ............................................................................208 Planimetry method ....................................................................................209 Histogram threshold method. ...................................................................209 Tube loading method ................................................................................209 BRTES-MOD tissue-equivalent threshold method ................................210 Results and Discussion.............................................................................................213
ix Monte Carlo Modeling......................................................................................213 Breast Tissue Composition Estimating Methods..............................................214 ACR BI-RADS Method.............................................................................214 Planimetry Method.....................................................................................215 Histogram Thresholding Method...............................................................216 Tube Loading Method................................................................................216 BRTES-MOD Tissue Equivalent Thresholding Method...........................217 Conclusions...............................................................................................................218 8 AVERAGE GLANDULAR DOSE BA SED ON HOMOGENEOUS PHANTOM244 Introduction...............................................................................................................244 Method Precepts.......................................................................................................246 General Concepts...............................................................................................246 Phantom Factor..................................................................................................248 Volumetric Factor..............................................................................................249 Anatomical Factor.............................................................................................250 Materials and Methods.............................................................................................250 Monte Carlo Simulations...................................................................................250 Phantom Factor..........................................................................................251 Volumetric Factor......................................................................................251 Anatomical Factor......................................................................................252 Patient-specific Monte Carlo Simulations..................................................253 Monte Carlo Model Elemental Composition.............................................253 Monte Carlo Spectrum...............................................................................253 Study Population and Selection Criteria............................................................254 Typical Mammographic Examination...............................................................255 Mammography Imaging....................................................................................255 Mammography Facility and Film Processing....................................................256 Digitization and Image Segmentation...............................................................256 Free-in-air Entrance Skin Exposure..................................................................257 Average Glandular Dose...................................................................................258 Results and Discussion.............................................................................................259 Monte Carlo Simulation HVL...........................................................................259 Heel Effect Impact of Dg...................................................................................259 Phantom Factor..................................................................................................259 Volumetric Factor..............................................................................................261 Anatomical Factor.............................................................................................262 Free-in-air Entrance Skin Exposure..................................................................262 Average Glandular Dose...................................................................................263 Conclusions...............................................................................................................264 9 CONCLUSIONS AND FUTURE WORK...............................................................297 Conclusions...............................................................................................................297 Future Work..............................................................................................................303 Dosimetry..........................................................................................................303
x Phantom Manufacturing....................................................................................304 Population Demographic and Anthropometric Studies.....................................304 Breast Tissue Fibroglandular Content...............................................................305 Individual Average Glandular Dose..................................................................305 Dissertation Derived Peer Re viewed Journal Articles......................................305 APPENDIX MCNP-5 INPUT FILES ...........................................................................................307 Free in Air Measurement MCNP-5 Input File..........................................................308 Free in Air Measurement with Aluminum MCNP-5 Input File...............................309 Phantom Factor MCNP-5 Input for Wu and BRTES-MOD phantoms....................310 Volumetric Factor MCNP-5 Input............................................................................312 Anatomical Factor MCNP-5 Input...........................................................................314 Energy Spectra Histogram for MCNP-5 Input........................................................316 Material Cards for MCNP-5 Input...........................................................................349 LIST OF REFERENCES ..........................................................................................355 BIOGRAPHICAL SKETCH ....................................................................................366
xi LIST OF TABLES Table page 2-1. Anatomical distribution of breast cancer in either breast.........................................51 2-2. Radiographic features suggestive of malignancy as describe in Kopans. 12............52 2-3. X-ray Spectra examined by Dance in developing DgN tables.26...............................53 2-4. The elemental constituents of adipose, breast glandular tissue................................54 3-1. MOSFET dosimeters, model TN-1002 RDI, sensitivity..........................................80 3-2. MOSFET dosimeters, model TN-502 RDS, sensitivity...........................................81 3-3. MOSFET dosimeters, model TN-1002 RDS, sensitivity.........................................82 3-4. MOSFET dosimeters, model TN-1002 RDM, sensitivity........................................83 3-5. MOSFET dosimeters, model TN-1002 RD, sensitivity...........................................84 3-6. Reproducibility of MOSFET dosimeters................................................................85 4-1. Gated fiber-optic-coupled dosimeter sizes.............................................................121 4-2. GFOC dosimeter, 1.1 mm model, sensitivity conversion factors..........................122 4-3. GFOC dosimeter, 4.0 mm model, sensitivity conversion factors..........................123 4-4. GFOC 1.9 mm dosimeter mode l sensitivity conversion factors............................124 4-5. GFOC dosimeter, 1.6 mm model, sensitivity conversion factors..........................125 4-6. Reproducibility of GFOC dosimeters....................................................................126 5-1. Commercial suppliers used in this study for epoxy-resin matrix...........................151 5-2. Breast tissue equiva lent modified series (BRTES-MOD) material........................152 5-3. Elemental composition, effec tive atomic number and the mass............................153 5-4. Elemental composition, e ffective atomic number,. ...............................................154
xii 6-1. Data collected in the retrospective study................................................................184 6-2. BI-RADS density categories used for population analysis....................................185 6-3. Interpreting radi ologist assigned BI-RADSÂ® density categories...........................186 6-4. Hormone replacement therapy BI-RADS density category...................................187 6-5. Mean BI-RADS density as a function of compression thickness..........................188 6-6. Mean BI-RADS density as a function of age group...............................................189 6-7. Technologist applied compression pressure...........................................................190 6-8. Technologist applied compression pressure...........................................................191 6-9. Mammography unit settings...................................................................................192 6-10. Study population CC breast measurements............................................................193 6-11. Study population MLO breast measurements........................................................194 6-12. The CC and MLO views PNL ratio......................................................................195 7-1. Data collected in the retrospective study................................................................237 7-2. Spectra parameters used in MCNP5 simulations...................................................238 7-3. The final exit doses from three patterns of tissue layers........................................239 7-4. Study population descriptive statistics breast glandularity....................................240 7-5. ACR BI-RADS assessment category 1 or 2...........................................................241 7-6. The mean glandularity calculated for LCC and RCC............................................242 7-7. Tube loading fitting parameters.............................................................................243 8-1. Spectra used in population ev aluation and MCNP-5 simulations..........................291 8-2. Data collected in the retrospective study................................................................292 8-3. Population demographics of breast volumetric parameters...................................293 8-4. Fitting parameters for entrance skin exposure reference measurement.................294 8-5. Study population descriptive statistics of Dg.........................................................295
xiii 8-6. The table lists the Pa rameters and Dg calculation..................................................296 A-1. Spectra used in population ev aluation and MCNP-5 simulations..........................316
xiv LIST OF FIGURES Figure page 2-1. Anatomical features of the human breast with microscopic detail..........................40 2-2. American College of Ra diology accreditation phantom..........................................41 2-3. Mass attenuation comparison betwee n glandular and adipose breast tissue............42 2-4 Raw x-ray tube target spectra at 28 kVp..................................................................43 2-5. Illustration of typical mammogra phy unit and typical tube geometries...................45 2-6. Illustration of focal spot and heel effect...................................................................46 2-7. A cross-sectional view of the comp ressed-breast edge and uniform region........... 47 2-8. Hammerstein et al. simplified model with an 0.5 cm layer of adipose tissue..........48 2-9. The graph shows the relationship between breast tissue glandularity.....................49 2-10. A MOSFET dosimeter functional sche matic.Adapted from Thompson Nielsen....50 3-1. A schematic of the typical clinical orie ntation of the dosimeter to the x-ray beam.66 3-2. The schematic illustrate s the experimental setup.....................................................67 3-3. The graph shows the angular res ponse of the dosimeter model TN-502RDS........68 3-4. The graph shows the angular res ponse of the dosimeter model TN-1002RD..........69 3-5. The graph shows the angular res ponse of the dosimeter model TN-1002RDM......70 3-6. Sensitivity response in MOSFET dosimeter model TN-1002RDI...........................71 3-7. Sensitivity response in MOSFET dosimeter model TN-502RDS............................72 3-8. Sensitivity response in MOSFET dosimeter model TN-1002RDS..........................73 3-9. Sensitivity response in MOSFET dosimeter model TN-1002RDM........................74 3-10. Sensitivity response in MOSFET dosimeter model TN-1002RD...........................75
xv 3-11. The graph shows the linearity of MOSFET dosimeter model TN-1002RDI...........76 3-12. The graph shows the linearity of MOSFET dosimeter model TN-502RDS............77 3-13. The graph shows the linearity of MOSFET dosimeter model TN-1002RD............78 3-14. The graph shows the linearity of MOSFET dosimeter model TN-1002RDM.........79 4-1 The schematic illustrates the fiber-optic-coupled dosimeter system........................96 4-2 The fiber-optic-coupled dosimeter computer interface software display.................97 4-3. A schematic illustrates the axial-angular evaluation geometry................................98 4-4. The schematic illustrates the experimental setup used for FOCD...........................99 4-5. The graph shows the axial-angula r response of the 4.0 mm dosimeter model......100 4-6. The graph shows the axial-angular response of the 1.9 mm dosimeter model......101 4-7. The graph shows the axial-angula r response of the 1.6 mm dosimeter model......102 4-8. The graph shows the axial-angula r response of the 1.1 mm dosimeter model......103 4-9. The normal-to-axis angular response of the 4.0 mm dosimeter model..................104 4-10. The normal-to-axis angular response of the 1.9 mm dosimeter model.................105 4-11. The normal-to-axis angular response of the 1.6 mm dosimeter model.................106 4-12. The normal-to-axis angular response of the 1.1 mm dosimeter model.................107 4-13. The normal-to-axis angular response of the 4.0 mm dosimeter model.................108 4-14. The normal-to-axis angular response of the 1.9 mm dosimeter model.................109 4-15. The normal-to-axis angular response of the 1.6 mm dosimeter model.................110 4-16. The normal-to-axis angular response of the 1.1 mm dosimeter model.................111 4-17. The graph shows the sensitivity response of the 4.0 mm dosimeter model..........112 4-18 The graph shows the sensitivity re sponse of the 1.9 mm dosimeter model...........113 4-19. The graph shows the sensitivity response of the 1.6 mm dosimeter model...........114 4-20. The graph shows the sensitivity response of the 1.1 mm dosimeter model ..........115 4-21. The mean sensitivity response plo tted as a function of dosimeter volume...........116
xvi 4-22. The linearity of the dosimeter re sponse of the 4.0 mm dosimeter model.............117 4-23. The linearity of the dosimeter re sponse of the 1.9 mm dosimeter model ............118 4-24. The linearity of the dosimeter re sponse of the 1.6 mm dosimeter model.............119 4-25. The linearity of the dosimeter re sponse of the 1.1 mm dosimeter model.............120 5-1. ACR phantom.........................................................................................................140 5-2. Glandularity as a function of compressed breast thickness....................................141 5-3. The BRTES-MOD phantoms.................................................................................142 5-4. The BRTES-MOD step phantom dimensions........................................................143 5-5. The fractional contribution to the total attenuation coefficient..............................144 5-6. The effective atomic number (Zeff) and the Zeff PE..................................................145 5-7. Mass density of BRTES-MOD phantom series.....................................................146 5-8. Tube-current time product (mAs) as a function of phantom composition.............147 5-9. Mass attenuation coefficient ( 25 keV) of BRTES-MOD phantom series..............148 5-10. Ratio of Âµ -1 for BRTES-MOD phantom series....................................................149 5-11. Ratio of Âµen -1 for BRTES-MOD phantom series.................................................150 6-1 CC and MLO segmentation regions and lengths used in this study.......................171 6-2. Breast edge region illustration in the geometry of a clinical mammography unit.172 6-3. Breast geometry model to determine CC and MLO whole area............................173 6-4. Breast edge width model to esti mate the adipose region thickness ......................174 6-5. Distribution of age groups de termined from the study population........................175 6-6. BI-RADS density categor y distribution for our study...........................................176 6-7 CC and MLO compression pressure used on patients during imaging..................177 6-8. Patient compressed breast thickne ss for CC and MLO views under compression.178 6-9. Study population tube potential distribution..........................................................179 6-10. Distribution of t ube current-time product..............................................................180
xvii 6-11. The measured CC view whole area........................................................................181 6.12. The measured MLO view whole area....................................................................182 6-13. Distribution of the calcula ted breast-edge region width........................................183 7-1. BRTES-MOD step phantom dimensions...............................................................220 7-2. Segmentation regions used in LCC and RCC mammographic views....................221 7-3. MCNP-5 geometry used in Monte Carlo (MC) simulations..................................222 7-4. BRTES-MOD TET method illustrated flowchart..................................................223 7-5. The five glandularity-estimating me thods using a 10% increasing increment......224 7-6. The five glandularity-estimating me thods using the ACR BI-RADS density.......225 7-7 Glandularity-estimating methods...........................................................................226 7-8. Glandularity-estimating methods as a function of age group................................227 7-9. A graph of the tube loading method model of GE-800T (Mo/Mo)........................228 7-10. A graph of the tube loading method model of GE-800T (Mo/Rh)........................229 7-11. A graph of the tube loading method model of GE-DMR (Mo/Mo).......................230 7-12. A graph of the tube loading method model of GE-DMR (Mo/Rh)........................231 7-13. The tube loading method model of GE-800T (Mo/Mo)........................................232 7-14. The tube loading method model of GE-800T (Mo/Rh).........................................233 7-15. The tube loading method model of GE-DMR (Mo/Mo)........................................234 7-16. The tube loading method model of GE-DMR (Mo/Rh).........................................235 7-17. A graph of the predicted gla ndularity as a function of BRTES-MOD...................236 8-1. Wu DM breast model showing the adipose and fibroglandular regions................266 8-2. Average glandular dose conversion factors DgN....................................................267 8-3. BRTES-MOD step phantom measurements..........................................................268 8-4. Flowchart rationale for homogeneous phantom factor used with BRTES-MOD..269 8-5 MCNP-5 geometry used in the simulations...........................................................270
xviii 8-6. The clinically measured half-val ue layer as function of MCNP-5 HVL...............271 8-7. The percent difference of average glandular dose, Dg...........................................272 8-8. Adipose and glandular region doses of of 2 cm.....................................................273 8-9. Adipose and glandular region doses of 3.3 cm......................................................274 8-10. Adipose and glandular region doses of 4 cm.........................................................275 8-11. Adipose and glandular region doses of 6 cm.........................................................276 8-12. Adipose and glandular region doses of 8 cm.........................................................277 8-13. Phantom factor for 2, 3.3, 4, 6, and 8 cm compressed breast phantoms................278 8-14. Phantom factor, BR 12, as a func tion of compressed breast thickness..................279 8-15. Adipose and glandular region doses of 2 cm.........................................................280 8-16. Adipose and glandular region doses of 4 cm.........................................................281 8-17. Adipose and glandular region doses of 6 cm.........................................................282 8-18. Adipose and glandular region doses of 8 cm.........................................................283 8-19. Volumetric factor as a function of breast mass for a compressed breast of 2 cm..284 8-20. Volumetric factor as a function of breast mass for a compressed breast of 4 cm..285 8-21. Volumetric factor as a function of breast mass for a compressed breast of 6 cm..286 8-22. Volumetric factor as a function of breast mass for a compressed breast of 8 cm..287 8-23. Anatomic factor fa as a function of percent x-ray field misalignment...................288 8-24. Histogram distribution of average glandular dose Dg...........................................289 8-25. Average glandular dose tre nds of average glandular dose.....................................290 A-1. Free-in-air geometry for HVL measurements........................................................352 A-2. Phantom factor geometry for Wu and BRTES-MOD phantoms...........................353 A-3. Anatomical factor ge ometry for Wu phantom.......................................................354
xix Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy BREAST DOSIMETRY IN CLINICAL MAMMOGRAPHY By Luis Alberto do Rego Benevides August 2005 Chair: David E. Hintenlang Cochair: Libby F. Brateman Major Department: Nuclear and Radiological Engineering The objective of this study was show that a clinical dosimetry pr otocol that utilizes a dosimetric breast phantom series based on population anthropometric measurements can reliably predict the averag e glandular dose (AGD) impart ed to the patient during a routine screening mammogram. In the study, AGD was calculated usi ng entrance skin exposure and dose conversion factors based on fibroglandular content, compressed breast thickness, mammography unit parameters and modifying parameters for homogeneous phantom (phantom factor), compressed breast lateral dimensions (volume factor) and anatomical features (anatomical factor). The protocol proposes the use of a fiber-optic coupled (FOCD) or Metal Oxide Semiconductor Field E ffect Transistor (MOSFET) dosimeter to measure the entrance skin exposure at the time of the mammogram without interfering with diagnostic information of the mamm ogram. The study showed that FOCD had sensitivity with less than 7% energy dependen ce, linear in all t ube current-time product
xx stations, and was reproducible within 2%. FOCD was superior to MOSFET dosimeter in sensitivity, reusability, and reproducibility. The patient fibroglandular content was evaluated using a calibrated modified breas t tissue equivalent homogeneous phantom series (BRTES-MOD) designed from anthr opomorphic measurements of a screening mammography population and whose elemen tal composition was referenced to International Commission on Radiation Units and Measurements Report 44 tissues. The patient fibroglandular content, compressed breast thickness along w ith unit parameters and spectrum half-value layer were used to derive the cu rrently used dose conversion factor (DgN). The study showed that the use of a homogeneous phantom, patient compressed breast lateral dimensions and pati ent anatomical featur es can affect AGD by as much as 12%, 3% and 1%, respectively. The protocol was found to be superior to existing methodologies. In addition, the st udy population anthropometric measurements enabled the development of analytical equa tions to calculate the whole breast area, estimate for the skin layer thickness and optim al location for automatic exposure control ionization chamber. The clinical dosimetry pr otocol developed in th is study can reliably predict the AGD imparted to an indivi dual patient during a routine screening mammogram.
1 CHAPTER 1 INTRODUCTION Worldwide, according to the World Health Organization,1 approximately one million women and men are diagnosed with breast cancer annually. Predominately a disease of high-income developed countri es, breast cancer accounts for 22% of all cancers in women worldwide. In the United States (US), breast cancer is the second leading cause of mortality in women.2 One in eight (~13%) women in the US are expected to be diagnosed with breast cance r during their lifetime. Breast cancer is a primary disease of women. The American Ca ncer Society (ACS) estimates in 2005 an estimated 212,930 new cases of breast cancer will be diagnosed in the US, of which 1,690 will be male and 211,240 will be female.3 The ACS also estimated 40,870 deaths (460 male and 40,410 female) from breast cancer in 2005.3 Mortality rates in all women have decreased 2.3 % per year from 1990 to 2001.3 The ACS attributed the decrease in mortality rates to increased awareness, early detection, sc reening and improved treatment regimes. In the United States, screening mammograms are the primary tool of choice to diagnose the early stages of breast cancer. Th e U.S. Preventive Services Task Force and American Cancer Society (ACS) recommends that women over the age of 40 have an annual clinical breas t examination that includes a mammogram.2,4 ACS also recommends that asymptomatic women under the age of 40 have a clinical breast evaluation every three years. Unfortunatel y, these recommendations are followed by only
2 62.4% of American women, according to the 20 00 Behavioral Risk Factor Surveillance System.5 Even though the incidence of breast cancer continues to be of great concern to women and their families and friend, mamm ography screening has come under scrutiny in the past few years in several studies for not significantly contribu ting to the reduction of mortality rates.4-6 Since these initial studi es, subsequent studies have shown that there is indeed a reduction in mortality resulting from screening mammography.5 Many risk factors have been identified for breast cancer. One of these is ionizing radiation, which has been shown to be a carcinogen for breast tissue.6-9 Mammography uses ionizing radiation as the energy sour ce for imaging breast tissues for neoplastic changes and other conditions. All medica l procedures performed on a patient by a medical entity are fastidiously documente d with the exception of routine diagnostic radiological procedures, where only the captu red image along with patient identification and diagnosis are documented. The individual patient dose resulting from an exposure to ionizing radiation is seldom documented as a routine matter. Currently, there are no federal or state mandates that require r ecording individual patient doses for any diagnostic radiological proce dure. In addition, there ar e no standardized clinical protocols to quantify the indi vidualÂ’s breast tissue dose a nd doses are very different among individual patients. Therefor e, it is not possible to evaluate accurately breast dose for a population. In order to study populat ion dose for mammography, a standardized clinical protocol involving dose would al low accurate documentat ion for individual patients. Such population dose data could provide data to impr ove statistical uncertainties such as those from the National Research Council, Committee on the Biological Effects
3 of Ionizing Radiations: Health effects of exposure to low levels of ionizing radiation (BEIR V).7 Clinically, such a protocol would prov ide radiologists and medical physicists with an accurate and reproducible methodology to estimate the risk associated with the exposure to ionizing radi ation use in mammography. This study aims to develop a dosimetry protoc ol that can be utilized in the clinical environment to estimate an individual' s average glandular dose resulting from mammography imaging with gr eater accuracy than is currently possible. Chapter 2 provides important background required to improve understanding of the dosimetry protocol. The remaining chapters have been laid out in a logical order to provide all the information needed to estimate the average glandular dose, Dg for an individual patient, equation 1-1. Dg is calculated with a pati ent's entrance skin exposure Xese; dose conversion factor DgN; phantom factor fp; volumetric factor fv; and a anatomical factor fa. DgN is a function of the patient's compressed breast thickness tb; mammography x-ray tube potential kVp; spectra l half-value layer HVL; and the patient's fibroglandular tissue content G%. a v p b gN ese gf f f G HVL kVp t D X D %, , , (1-1) Chapter 3 and Chapter 4 explore the possi ble use of two types of dosimeters to measure Xese for a patient during the mammography pr ocedure. Chapter 3 evaluates the potential for the use of a high-sensitivity Metal-Oxide Semiconductor Field-Effect Transistor (MOSFET) to measure Xese during the imaging session. In this chapter, several models of MOSFET dosimeters are characte rized for use in clinical mammography by evaluating calibration factors, linearity, angul ar dependence, and reproducibility. Chapter 4 considers use of a fiber-optic-coup led (FOCD) dosimeter to measure Xese. FOCD
4 prototypes were characteriz ed for use in a clinical mammography imaging session by evaluating each of four dosimeters as to their calibration fact ors, linearity, angular dependence, and reproducibility. Chapter 5 details the development and ma nufacture of a modified breast-tissueequivalent phantom series (BRTES-MOD) to represent, not only elemental composition, mass density, and radiological in teractions, but also the scattering characteristics related to the dimensions of the breast. Manu facturing techniques and phantom components were modified to improve the homogeneity of the phantom. The phantom material is utilized to manufacture a series of compresse d breasts over a range of thickness (2-8 cm) and fibroglandular tissue c ontent (16.2-68.7%). In a ddition, a step phantom is manufactured and used to estimate G% for an individual patient from their clinical mammogram. Chapter 6 details a prospective study of mammography patients to determine the anthropometric measurements of typical br easts under clinical ma mmography conditions. A three-month (Jan-Mar, 2005) retros pective demographic study compiled mammography data acquired from 253 patient s (1040 mammograms) seen at a U.S. Navy health care facility in Jack sonville, Florida. The mammogr ams were reviewed, digitized and segmented to measure craniocaudal and mediolateral oblique measurements to determine the study population average breast dimensions and radiological characteristics. Chapter 7 details an investigation to dete rmine how radiographic opacity of breast tissues in mammography clinical geometry may be affected by tissue layering and surrounding tissues using a simplified model in a Monte Carlo (MCNP-5) environment.10
5 In addition, chapter 7 also describes an inte rcomparison of glandular tissue quan tification using the clinical populati on studied in Chapter 6.11 These include the American College of Radiology (ACR) breast imaging re porting and data system BI-RADSÂ® method (ACR BI-RADS), planimetry method (PM), histog ram-thresholding method (HTM), tubeloading method (TLM) and modified breas t-tissue equivalent phantom series thresholding method (BRTES-MOD TET). Through the ACR BI-RADS method, the radiologist assigns a single ACR designated density category to the patient. In the planimetry method, the fibroglandular content is the percent area of fibroglandular tissue to total breast area. The user in HTM assigns a minimum pixel intens ity as fibroglandular tissue, and then, using the image histogram, calculates the percent area of fibroglandular tissue pixels to total breast area. BRTES-M OD TET uses a fibroglandular calibrated step phantom to establish a calibra tion curve of fibroglandular co ntent as a function of pixel intensity, which is then applied to the histogram of the image. Chapter 8 details the Monte Ca rlo modeling used to derive fp, fv, and fa using the phantom developed for the ACR Mammography Accreditation Program as well as the BRTES-MOD phantom model developed in Chapter 5.12 Chapter 8 then brings together the information gathered from all previous chapters to complete the methodology to estimate average glandular dose using a homogeneous phantom BRTES-MOD. The final section of Chapter 8 compares the Dg derived from BRTES-MOD TET and compares it to HTM, ACR BI-RADS and the standard ACR method that is currently in use by mammography medical physicist. Chapter 9 details the clinical dosimet ry protocol that will that can be utilized in the clinical e nvironment to estimate an individual's average glandular dose resulting from mammogra phy imaging. Chapter 9 concludes with a
6 synopsis of future work needed to expand the dosimetric techniques to mediolateral oblique view and to facilitate the migrat ion of the method to digital mammography.
7 CHAPTER 2 BACKGROUND Breast Anatomy and Physiology The primary function of the female human br east is to produce lactate (milk) for the feeding of the human offspring during the early stages of in fancy. Figure 2-1 illustrates the general anatomical features of the huma n breast discussed in this paragraph. The breast is a modified gland of th e epidermis (skin) that develo ps on the thorax anterior to the rib cage and the pectoralis major muscle.13 Typically, it is located between the clavicle and eighth rib. The breast can be f ound medial to the apex of the axilla and lateral to the sternum. Anterior to the pectoral is major muscle is a layer of retromammary adipose tissue followed by glandular tissue. Gl andular tissue is seque stered in 15 to 20 lobes. Each lobe is surrounded by adipose ti ssue and has a varying number of lobules and ducts. The glandular tissue within the l obules is surrounded by collagenous connective tissue. Typically, each lobule consists of 30 terminal branches that constitute the parenchymal portion of the breast. The ducts leading anteriorly from the lobules are surrounded by loose mesenchymal tissue (develop s into connective tissue). The lobules are made up of terminal ducts terminating in lobulus units. Each terminal duct lobular unit (TDLU) excretes lactate into the extralobular terminal duct, which subsequently empties into the ductus lactifer. Just posterior to the areola, the ductus lactifer empties into the sinus lactifer. The sinus lactifer collects a small volume of lactate before it empties into the pars infundibularis in the mammilla (nipple) and exits the breast via the porus excretoris. Anterior to the glandular tissue is a la yer of subcutaneous adipose
8 tissue. Interweaving and compartmentalizing thes e structures are the inferiorly supporting structures of superficial fascia, deep fascia and CooperÂ’s ligaments. Lymphatic, vascular and enervation components are also found thr oughout these breast ti ssues. Histologically, each duct is made up of epith elial cells surrounded by myoe pithelial cells. The ducts extend into the TDLU and terminate in the glandular acini (grape-cluster cavity). Generally, breast tissue is subcategorized into glandular and adipose tissue. The glandular tissue is made up of the ducts, TD LUs and associated tissue that make up the lobes. The radiographic opacities of the breast typically visu alized in a mammogram are a composition of fibroglandular ti ssue, which, is composed of glandular tissue, fibrous tissue, ligaments and their associated vasculature and enervation. The content of fibroglandular tissue will determine the ra diographic opacity of the breast in a mammogram. The radiographic optical densitie s are also referred to as parenchyma or parenchymal patterns. Developmentally, glandular tissue in males remains underdeveloped, while in females it begins development during pubert y and continues to the age of thirty.13 During the developmental period, hormone levels pr omote development and increase the number of branches of the glandular structure in the female. The size of the breast is more dependent on the adipose content than the gl andular content of the breast. In general, breast size and the relate d breast tissue composition changes throughout the postpubescent life of the female.13 The single-most important f actor in changing breast tissue composition is reproductive hormones. It can be generally stated that a breast contains more adipose tissue in proporti on to glandular tissue as th e female ages after puberty.13 Pregnancy, being the apex of the hormone change in women, dramatically affects a
9 mammogram due to the active nature of the glandular tissue in pr eparation for the unborn child and mammography screening during pregnancy is typically avoided. Diagnostic mammography is performed fo r a number of medical conditions. Because of the focus of this study is the re lationship of mammography with the stochastic effect of breast cancer, the diagnosis of breast cancer is described in detail. Breast Carcinogenesis Cancer is the generic term used to describe the malignant neoplasia of cells that have lost their ability to regulate normal cellular functions.14 Neoplastic changes in cells have many root causes; however, of particular interest to th is research are those cancers that may be caused by ionizing radiation. Ne oplastic changes resulting from ionizing radiation have been shown in mammary tissu es of mice, rats, guinea pigs, and humans.7,9 The radiosensitive tissue of the human breast is the active glandular ti ssue, specifically in the terminal duct the TLDU.6 Regardless of the cause of breast cancer, seventy percent of all breast cancers develop in the periphery of the parenchyma.13 The periphery of the parenchyma consists of a one-centimeter surrounding layer of ad ipose tissue (retromammary adipose and subcutaneous adipose) in the breast.13 The distribution of carcinomas on breast (right or left) can be divided into five major quadrants and is shown in table 2-1.6 There are two general types of carcinomas: in-situ and invasive. In-situ cancers are contained within the anatomical structure. As an example, ductal carcinoma in-situ (DCIS) is found inside the duc ts leading from the TDLUs. Lobular carcinoma in-situ is located within the lobules. Eighty-five per cent of all in-situ ca rcinomas are DCIS.13 Invasive carcinomas develop from in-situ carcinomas and then spread to surrounding tissues. Invasive carcinomas can metastasize, traveling through circ ulatory or lymphatic
10 systems and develop secondary malignancies outside breast tissu es. Invasive ductal carcinoma is the most common fo rm of invasive carcinoma.5,13 It is currently believed that the genesis of the breast neoplasm is in the segment of terminal duct that extends into the lobulus units, figure 2-1.13 An initial biological indicator for these types of carcinomas is the fo rmation of clusters of micro-calcifications in the TDLU or associated ducts. Calcificat ions are composed of calcium hydroxyapatite and tricalcium phosphate and may contain heavy metals.13 Calcifications within themselves do not pose a risk and are typi cally a benign feature of normal breasts; however, if a cluster of five or more hete rogeneously-shaped micro-calcifications (150200 Âµm) is located within a 1 cc volume, caref ul evaluation may be warranted. The origin of the benign or malignant deposit of calcium in breast is not well understood.13 Another common biological indi cator of carcinoma is a mass. Masses can also be benign or malignant. A common feature of a ma lignant mass is the sp iculated nature of its borders. The origins of breast masses can be numerous, including a hard bruise, necrotic adipose tis sue and malignancy. 13 Even though calcifications and masses are the most common diagnostic markers for disease, a summary of findings that can be su spect for abnormalities is listed in table II.13 Historical Perspective: Efficacy to Oversight Soon after the discovery of x-rays by Roentgen in 1895, medical physicians used the newly discovered imaging modality to explore the human body. In 1913, Soloman, a German surgeon, reported visualiz ing breast cancer using x rays.15 Beginning in the early 50s, Gershon-Cohen and Ingleby developed x-ra y techniques and diagnostic criteria for diseases of the breast after having by ex amined over 1,600 women with general x ray imaging.16 Their efforts culminated in a paper called "Can Mass X-ray Surveys Be Used
11 in Early Detection of Early Cancer of the Breast?" in which they proposed the "Early detection of unsuspected neoplasms can be promoted by serial survey studies of the breast in women over 35 years of age" .16 In the late 50s, Robert L. Egan, a Houston radiologist, developed a imag ing technique using industrial film and fine-grain intensifying screen with high milliamperage-lo w voltage that provided clearer images that were easier to diagnose.15 Egan demonstrated the effi cacy of mammography in the late 1950s. Egan and his colleagues screen ed 1000 women for breas t cancer, of which 245 were diagnosed with breast cancer by bi opsy. Of the women that were diagnosed with breast cancer, 238 showed evidence of the malignancy in their mammograms.15 The efficacy of mammography gained momentum wh en the Health Insurance Plan of New York randomized study (HIP)(1963-1966) wa s conducted by Strax and Shapiro.17 Strax and Shapiro showed a reduction in mort ality rates as a result of mammography screening.18 As a result of the findings of the HIP study, the American Cancer Institute and the National Cancer Institute sponsored the Breast Cancer Detection Demonstration Project (BCDDP) from 1973 to 1978.15,17 In 1975, the United States Food and Drug Administration (FDA) began scrutiny of mammography facilities after a report by Henry Bicehouse in which he identified facilities in Pennsylvania with high patient doses.19 At that time, many facilities were not using dedicated mammography units. Soon th ereafter, the FDA initiated the Breast Exposure: Nationwide Trends (BENT) program to identify problems with dose and quality of mammography.17 In October of 1975, the National Cancer Institutes (NCI) developed a working group to revi ew the benefits of the HIP study.17 The NCI working
12 group determined that the benefit of screen ing was warranted for women over 50 years of age. Evaluation of the BCDDP data indicated th at two-thirds of the cancers detected could be identified only by mammography.17 The success of the BCDDP was attributed to the technological advances and us e of xeromammography and screen-film mammography in BCDDP centers.17 Ultimately the higher doses of xeromammography as compared with screen-film mammography resulted in the widespread use of screenfilm mammography, causing si gnificant reduction in dose, and risk for mammography screening. Consequently, the greatest reducti on in patient dose resulted from facilities migrating to film-screen mammography. In 1980, American Cancer Society (ACS) advocated that women receive a baseline mammogram between the ages of 35-39.15 In 1983, the ACS recommended that women between the ages of 40-49 have a screening mammogram every one to two years.15 The FDA performed a nationwide review of phantom image quality and radiation doses under the National Evaluation X-ray Trends (NEXT) program in 1985 and 1988.17 The NEXT-85 report identified numerous facilities with significant image quality probl ems. NEXT-88 reported the implementation of dedicated mammography units in facilities that had prev iously been using general purpose units. The report showed that ma mmography quality was a continuing issue when 13% of the facilities faile d the phantom image scores. In 1986, the National Council on Radiati on Protection and Measurements (NCRP) published radiation dose guidelines for mammography in Report No. 85.6 The ACS launched a nationwide campaign, the Breast Ca ncer Awareness Program, to encourage the use of screening mammography. The Ameri can College of Radiology (ACR), in view
13 of image quality issues, developed th e ACR Mammography Accreditation Program (MAP).17 In August 1987, the ACR MAP began accrediting mammography facilities.17 Even though the program was voluntary, the ACR accreditation soon became the standard of care for mammography facilities. However, it was not mandatory. The NCI and ACR joined ACS in recommending mamm ography for women between the ages of 40-49 in 1988. Public health policy in the United States progressed forward when in 1990, the Breast and Cervical Cancer Mortality Prevention Act (P.L. 101-354) was passed introducing screening mammogra phy as the nation's public hea lth policy for detecting the early stages of breast cancer. A cooperativ e agreement between the ACR and the Centers for Disease Control (CDC) st arted in September 1990 to de velop quality control and quality assurance curricula to improve mammography facilities.17 By 1994, 41 states had created regulatory oversight of their mammography facilities.17 In 1990, the U.S. Congress passed the Omnibus Budget Reconci liation Act that included federal funding for screening mammography. Soon thereafter, the Health Care Fina ncing Administration (HCFA) began reimbursing for screening mammography as a Medicare benefit. In late 1991 and early 1992, the national media brought unwanted attention to mammography when several exposÃ©s aire d on substandard mammography practices.17 A report by the U.S. Senate described the stat e of national mammography regulations as a "conflicting and overlapping patc hwork of federal, state, private, and voluntary standards".17 As a result of the media exposÃ©s and public pressure, federal legislation was drafted to codify qualit y assurance and quality cont rol measures in mammography facilities. The Mammography Quality Sta ndards Act (P.L. 102-359, MQSA) was enacted
14 into law in October 1992. The act intr oduced quality standard regulations for mammography diagnostic and screening faci lities, providers and equipment. Risk and Mammography Efficacy of Mammography The principle goal of a preventive medi cal procedure is to identify a potential disease at its early stages, thereby decreas ing morbidity and mortality. Under a national consensus that screening mammography was the standard of care to detect potential cancers, researchers began to look for the co st-benefit ratio of sc reening mammography. A cost-benefit analysis weighs the costs of providing the screening mammogram against the benefit of reducing breast cancer mortali ties. Starting in early 2000, the efficacy of screening mammography was criticized by two scientific studies, even though there had been numerous studies showing a benefit for mammography screening.5,8,18,20-22 Researchers in Sweden (Cochrane review ) concluded that there was no reliable evidence that screening mammogr aphy reduced mortality rates.21 As a result of the Cochrane review in 2000, the United States Se nate held a public hearing titled "Making Sense of the Mammography Controversy: Wh at Women Should Know" to clarify the findings of the Cochrane review study.20 As a result of the hearing, the United States Preventive Services Task Force, an indepe ndent panel of experts that provide policy guidance to the federal government, also revi ewed the studies and concluded that there was fair evidence that mammography could re duce breast cancer mortality up to 25% over ten years.23 The task force further stated that the evidence was strongest for women whose age was between 50 and 69, but conclude d that the benefit likely extended to women 40-49 years of age.23
15 Subsequent studies have again re-evaluated the cause and effect of cancer mortality rates and screening mammography. These have again concluded that there is a benefit derived from screening ma mmography. Two Swedish studies have shown long-term benefits of screening mammography.24,25 A study from the Netherlands in which 27,948 women participated showed a reduction of breast cancer mortality of 1-7%.26 In the United States, the ACS has reported that mo rtality rates have declined by 3.2% from 1995 to 2003 as a result of ma mmography screening programs.2 In 2004, reviews of the controversial studies have provided conclusi ve evidence that a substantial reduction in mortality was evident as a re sult of screening mammography.23,27 Risk Associated with Mammography Exposures Outside of risk factors for breast cancer a ssociated with individual health, family history, life-style and environment, a risk that has not been sc rutinized in recent times is the dose received by the glandular tissues of individual patients. For measuring dose, MQSA and the ACR MAP use the phantom that was initially developed for the ACR MAP, so that dose and image quality are assessed with the same test object. This phantom was designed to represent a 4.2 cm compressed breast composed of 50% glandular and 50% adipose tissues. The conceptual model for this phantom wa s first introduced by Hammerstein et al. in 1979.28 Hammerstein et al. stated that the m odel was based on an educated estimate of the average breast in women and not on population studies.28 The current ACR phantom that is used as the Â“average breastÂ” in th e United States is composed of a 4.2 cm thick acrylic block in which a wax insert is pl aced. The phantom was designed to mimic the automatic exposure response of a 4.2-comp ressed thickness with a 50% glandular composition, figure 2-2. The wax insert cont ains objects to evaluate image quality.
16 Because of variations in breast size and compositions among individuals, neither BR 12 nor the ACR phantom adequately re present the breast tissue composition of screening mammography populations.29,30 In order to quantify the glandular dose better for a population, a reproducible method of de termining absorbed dos e to the sensitive breast tissue is needed. As originally sugge sted by Hammerstein et al., mean glandular dose or average glandular dose is the current dosimetric quantity that best quantifies the energy deposited in the radio-sensi tive glandular tissue of the breast.7,9 Lifetime Risk from Mammography Currently, the preferred method in the U.S. to quantify the lifetim e risk associated with any radiation exposure to various organs and tissues as well as to individuals for incidence and mortality is the Health Effects of Exposure to Low Levels of Ionizing Radiation (BEIR V) report from the Committ ee on the Biological Effects of Ionizing Radiations.7 For breast cancer risk, the analysis pr esented in the BEIR V report is based on human data from two mortality series and three incidence series. The mortality series used were the Canadian Tuberculosis Fluor oscopy (CTF) Study (473 deaths) and the Life Span Study (LSS) of Atomic Bomb Survivors (151 deaths).7 The three incidence series were the LSS (367 cases), New York Acut e Postpartum Mastisis (NYAPM) Study (118 cases) and the Massachusetts Tubercul osis Fluoroscopy (MTF) Study (65 cases).7 The LSS and NYAPM groups reported having recei ved acute exposures, whereas the MTF and CTF groups received exposures fractionate d over a period of years. The model was developed with a five-year latency period; however, the data indi cated no excess cancer risk within 10 years.7 The BEIR V risk model for br east cancer mortality is given in equations for dose response 2-1, 2-2, 2-3a, and 2-3b, where ( d ) = individualÂ’s age-specific cancer risk,
17 0= age-specific background risk, f (d) = dose function, g ( ) = excess risk function, 1 = 1.22 Â± 0.610, d = mean glandular dose in Sievert (Sv), E = age of exposure, T = years after exposure.7 Estimated parameters for dose response equati ons 2-1, 2-2, 2-3a, and 2-3b, their standard errors and standard error percentage in parenthesis, are 1 = 1.385 Â± 0.554 (40%), 2 = -0.104 Â± 0.804 (773%), 3 = -2.212 Â± 1.376 (62%) and 4 = -0.0628 Â± 0.0321 (51%).7 ) ( ) ( 1 ) (0 g d f d (2-1) d d f1) ( (2-2) if E 15, ) 20 / ( ln ) 20 / ln( exp ) (2 3 2 1T T g (2-3a) if E > 15, ) 15 ( ) 20 / ( ln ) 20 / ln( exp ) (4 2 3 2 E T T g (2-3b) The BEIR V model indicates a higher risk for women 15-20 years of age and lower risk for women greater than 40 years of age. The BEIR V breast model has considerable sample variation due to the small number of cases used, which ther efore result in large statistical uncertainties in equation parameters.
18 Mammography Designed and optimized to take advant age of several unique radiological characteristics of breast tissues, mammography visualizes early biol ogical manifestations of breast cancer. Photon-Tissue Interactions Breast tissue is composed of various tissue types that have very similar radiological interactions. The mass density and atomic co mposition are so similar that it makes the tissues difficult to distinguish in radiographs in the normal diagnostic x-ray range of tube potentials (50-120 kVp). Dedi cated mammography units, desc ribed below, have been designed to take advantage of the difference that exists in the radiological interaction coefficients at low energies to enhance the differences between the glandular and adipose tissues, figure 2-3(a). The three primary phot on interactions encountered in the spectra from modern mammography units are photoele ctric effect, incoherent scattering and coherent scattering, figure 2-3(a). The dominan t contributor in diffe rentiating the tissues in mammography is the atomic number (Z3) dependence of the photoelectric effect mass attenuation coefficient. The differences be tween tissues become accentuated because of their density difference, as shown in thei r linear attenuation coefficient plot, figure 23(b). Mammography Unit Characteristics Mammography units have unique aspects, such as speciali zed target/filter combination, unit geometry and image record ing system. One unique characteristic of a mammography unit is its variety of specialized spectra. The majority of spectra used clinically today are in th e low-energy range (24-50 kVp; HVL, 0.23-0.60 mm Aluminum (Al)). The mammography x-ray tube target s are commonly made of molybdenum (Mo),
19 and units with more flow over target have a rhodium (Rh) or tungsten (W) target in addition. The target spectrum is typically filter ed with a Mo, Rh, or Al filter. The primary target-filter combination used in mammogra phy today is Mo/Mo. In the case of the Mo and Rh targets, the characteristic x-ray featur es of the filter employed plays a significant role in shaping the resultant spectra, figure 2-4. The characte ristic x rays of Mo are 17 and 19 keV, whereas for Rh they are located at 20.2 and 22.7 keV. The characteristic x rays emitted by the Mo or Rh target and filtered by Mo or Rh, respectively, are relatively unaffected by the filter of the same composition, because the characteristic x rays fall just below the K-absorpti on edge, figure 2-4, where the attenuation coefficient is reduced significantly. In the case where a Mo target is filtered by Rh filter, the higher Rh K absorption edge will cause a decrease in contribution from x rays below 19 keV and increases the contribution of x rays above 20 keV. The filtered spectrum, therefore, has an enhanced characte ristic x-ray distributi on lacking the low or high-energy photon components of the raw spectr a emitted by the target. The interaction between target and filter results in re moving low energy photons that contribute to increased patient dose and removing the hi gh-energy components that reduce image contrast. Target-filter combinations typical ly used clinically are Mo/Mo, Mo/Rh and Rh/Rh. A second important characteristic of a mammography unit is its beam geometry, figure 2-5. The geometry of the beam is designed to take advantage of the heel effect by having the highest intensity component of the beam impinging on the thickest part of the breast at the chest wall and the low intensit y component directed to the thinnest portion of the breast, the mammilla region, figure 2-6(a). The position of the anode and the
20 cathode improves the uniformity of the tr ansmitted x rays. In addition, the beam geometry also affects the resolution associated with the focal spot (FS) by reducing the projected FS size along the anode-cathode axis , figure 2-6(b). The unit's reference FS is typically measured at the axis that bisects the field (reference axis ), figure 2-6(b). The relationship between the FS length at the central axis or chest wall (Achest wall) and the FS length at the reference axis (Aref) is shown in equation (2-4), where is the angle between the central axis and the edge of the beam and is the angle between the central axis and the reference axis. tan / ) (tan( 1 ChestWall refA A (2-4) A third important characteristic of ma mmographic unit is the grid for reducing scattered radiation resulting in improved contrast. Today there are two types of grids used in mammography; linear, and crosshatch. In both instances, the grids are in constant motion during the exposure to avoid imaging the se pta. Radiation that is parallel to the septa is allowed to transverse the grid wher eas radiation that deviates from that path typically collides with septa and is absorbe d. The linear grids will ty pically have 31 lines per centimeter and a 5:1 ratio of lead to septa. The cross-hatch will typically have 23 lines per centimeter and a 4:1 ratio grid with copper septa. The image-capturing device is the last por tion of the imaging system described. Today there are two types of image-capturi ng devices in mammography, film screen and digital. Currently the most commonly used device is film screen, primarily because of cost.31 Screen-film mammography is unusual here, too, because it uses cassettes made of low-attenuating material and a singleemulsion film with one image-intensifying screen. The film-screen combination has been designed to provide adequate sp atial resolution of
21 the image at reasonable dose. The antiscatter grid reduces the scat tered radiation while allowing the primary beam to pass to the casse tte, enhancing subject image contrast. The x rays then travel through the cassette, film base and film emulsion before interacting with the phosphor screen to provide higher spatial resolution. The most commonly used phosphor in mammography is terbium-ac tivated gadolinium oxysulfide (Gd2O2S:Tb) The most probable interaction between the film and the phosphor occurs near the surface of the phosphor screen because of the hi gh effective Z of the phosphor screen. Digital systems have two types of imagecapturing devices, computed radiography (CR) or digital radiography (DR). Currently the only digi tal system that has been approved for use by the FDA is DR. CR is e xpected to be approved for mammography in 2005. Computed radiography (CR) uses a phot ostimulable phosphor detector (PSP) system. Most CR plates are made of 85% BaFBr and 15% BaFI doped with europium (Eu) commonly called barium flourohalide.31The flexible CR plate is used in a cassette that is similar to sc reen-film cassette without a screen. CR plates are exposed in the same manner as screen-film. The latent image is stor ed on the CR plate in the form of trapped electrons. The x ray energy interacts oxidizes Eu+2 to Eu+3 with a mobile high-energy electron. A portion of the mobile high-energy electron becomes trapped in a F-center trap. The CR plate is then read by the use of a stimulating red-wavelength laser that stimulates releases the trapped electrons resulting in a blue-g reen wavelength light emission. The output signal travels through a fi beroptic guide to a photomultiplier tube whose signal is converted to digital using a analog-digital converter. The stored signal
22 contains the corresponding gray scale value al ong with a spatial posit ion generated by the CR reader. Digital radiography (DR) has two major categor ies of detectors indi rect and direct. Indirect DR uses an intens ifying screen on top of the photodetector to convert the x-ray energy into light. Cesium Iodide (C sI) is the popular choi ce for the intensifying screen because it improves spatial resolution with its columnar crystals. The typical photodetector used for this application is a fl at panel discretized de tector. Each discrete element contains the electronics and a light -sensitive area. The phot oconductor converts the light signal emitted from the intensifying screen into electrons. The photoconductor is made up of transistor that is read in a series fashion retaining positional orientation. Spatial resolution is limited by the disc retized detectors. Direct DR uses a photoconductor on top of the thin-film transist or which converts x rays directly to electrons without the use of an intensifying screen. The phot oconductor is typically made of amorphous selenium. Unlike CR and indirect DR, direct DR does not trade off spatial resolution for sensitivity. Breast Tissue Density Breast tissue density (glandul ar tissue content) and co mpressed breast thickness are the main tissue parameters of interest for producing the mammography image of the fibroglandular tissues. The associated parenchy mal patterns in the image have in recent years been associated with the risk of breas t cancer by studies that have correlated cancer incidence with breast tissue density.32-45 A confounding factor to th e associated risk is the difficulty of the radiologist in detec ting a neoplasm amongst the dense breast parenchymal patterns from a radiograph.32 Generally speaking, diagnostic information
23 (i.e., calcifications and soft tissue masses) is more hidden from the radiologist as the breast tissue density increases. Density Measurement Tissue density, composition or glandular ity is determined by quantifying the glandular content of the breast. The earliest atte mpt to classify tissues and their associated risks from mammograms was made by John Wo lfe, who introduced a qualitative method of evaluating breast tiss ue density in 1976 by four qualitative categories.36 The Wolfe categories are: N1-lowest risk, P1-low risk, P2-high risk, and DY-highest risk.36 WolfeÂ’s category system, having been developed from a qualitative perspective, has little to offer the medical physicist in the ev aluation of patient dosimetry. At present, the ACR Breast Imaging Reporting and Data Sy stem (BI-RADS) categories ar e the accepted standard for classification. BI-RADS partitions breast densities into four categories11: Â• Almost entirely fat (<25% fibroglandular) Â• Scattered fibroglandular dens ities (25-50% fibroglandular) Â• Heterogeneously dense (51-75% fibroglandular) Â• Extremely dense (>75% fibroglandular) The BI-RADS categories are annotated as part of the patientÂ’s medical record; however, they do not constitute a quantitative measur ement of tissue density by the interpreting radiologist. Wolfe and colleagues in 1976 used a com puter-assisted planimetry method to measure breast density. They used an acetate overlay with a trace of the breast outline and parenchymal regions to identify the dense areas.36 Recently, digi tized mammograms have been segmented and tissue compositions similarly quantified.46 These twodimensional methods ignore the three-di mensional nature of the breast.
24 In 2003, Pawluczyk and colleagues devel oped a digitized method in which the imaging system was calibrated by using a tissue-equivalent step-wedge simulating differing breast densities.47 The authors created a three-di mensional plane function that correlated breast thickness, per cent glandularity and the logari thm of the exposure. Based on the three-dimensional plane, the digitized image was evaluated pixel by pixel, and a resultant aggregated breast density was determined. Tissue Compression Tissue compression is desirable in ma mmography because it reduces overlapping anatomy and decreases tissue thickness.31 The decrease in tissue thickness results in reduced geometric blurring, reduces the dyna mic exposure range, and lowers patient dose. However, compression of the breast is highly dependent on the compression pressure, compression paddle size, patient pos itioning and the patien t. Current Federal regulations mandate that the compression th ickness indicator of the mammographic unit must be accurate to 5 mm under moderate compression and reproducible to 2 mm for compressed breast thickness between 1 to 8 cm.12 It is the intention of tissue compression to cause anatomical distor tion of the breast tissue. The distortion from compression divides th e breast into two major regions: the tissue that is under full compression, making the tissue uniform in thickness, and the breast edge region. Figure 2-7 illustrates th e two regions. The edge of the compressed breast has tissue that is not uniform in th ickness. The breast-edge effect has been previously discussed by Highnam et al. and ha s been explored after collection of the patient data in this study.48
25 Tissue Characteristics Johns and Yaffe characterized the radiol ogical attenuating properties of normal and neoplastic breast tissues. The tissue sample s were harvested from breast reductions (normal), and cancer mortality autopsies (neoplastic).49 They reported that neoplastic (DCIS) tissue had a four-percent increase in the tissue linear attenuation coefficient, as compared to normal glandular tissue. Current tissue references do not appear to have taken this discrepancy in attenuation into account for determining the atomic composition of breast tissues. Further studies are needed to determine the atomic constituents of normal and neoplastic tissues. Breast Dosimetry Reviews Various attempts have been made to deri ve a dose measurement that appropriately quantifies the energy absorbed by the radiosensitive tissue in the breast. It is now known that this radiosensitive tissu e is the active glandular tissue.8,9 Some of the dosimetric measures that have been used in the past are skin-entrance dose, mid-breast dose and total energy imparted to the breast.28,50 All of these measurements have not adequately quantified the dose absorbed by the tissue at risk. Today, average glandular dose (AGD) is used as a measure of th e dose to the tissue at risk.50 However, an in-vivo measurement of the mean glandular dose is not feasible . Historically, methodologies were devised to relate the AGD to the incident exposure, or now the air ke rma using conversion factors. These conversion factors started as direct meas urements in tissue equivalent material and have since progressed to Mont e Carlo evaluations based on th e original, simplified model of the compressed breast. This review is in tended to provide the pertinent historical perspective to the currently-util ized methods of calculating AGD.
26 Hammerstein et al., 1979 Prior to the work of Hammerstein et al ., mammography dose and its associated risk were based on the surface skin exposure measurement. Hammerstein et al. proposed several dose measurements to be used to document patient dose in mammography. The authors investigated the mid-breast dose, to tal energy absorbed by the breast and energy absorbed in tissue at high risk (gland ular). Two systems (xeromammography and screen/film imaging systems) were examined, because they were the standard of practice at the time. At the time this research was being conducted direct film mammography fell out of favor due to its high doses. The focus then became on deciding which system could produce the lowest patient dose, sc reen-film mammography or xeromammography. The xeromammography-imaging system had a t ungsten target with an aluminum filter, with a half-value layer in the range of 0.36-1.21 mm Al and a kVp range of 40Â–50 kVp. The screen/film imaging system had a molybde num target and filter with a HVL of 0.31 mm Al at 28 kVp. Hammerstei n et al. did indeed find that AGD dose was higher with xeromammography-imaging system when compar ed with screen/film combination. Determining the atomic composition of breast tissue was also in its infancy, and the authors evaluated skin, adipose tissue, and glandular tissue samples from mastectomy procedures. They used Archimedes princi ple to determine the tissue density. They utilized dehydration and chemical extraction to determine the atomic composition of the samples. The authors indicated that the adip ose and glandular tissue had large variations in carbon and oxygen due to the contamination from other tissues. The authors however, did not indicate whether the contaminating ti ssue was free of neoplasms. In view of Johns and YaffeÂ’s attenuation characteristics of neoplastic and normal tissues, the effective
27 atomic composition of Hammerstei n et al. may have had a systematic error resulting from neoplastic tissue contamination.49 In order to evaluate the mid-breast dose , total energy absorbed by the breast and dose absorbed in tissue at high risk (glandular), three tissue substitutes were utilized that were matched for the radiation interaction prope rties: they were water, adipose tissue and 50% glandular tissue(50% water/50% adipose).51 The tissue substitutes were manufactured into homogeneous Â“DÂ” shaped 1-cm sections with 93 cm2 area. The tissue doses were measured in the tissue substitute using thermoluminescent dosimeters placed in machined cavities. The doses were evalua ted for breast thickness ranging from 1 to 7 cm. Glandular tissue dose was based on embeddi ng a small mass of glandular tissue at a depth of 3 cm in a tissue subs titute slab (50% glandular). Hammerstein et al. did not indicate the mass size of th e glandular tissue used in the embedding process. Hammerstein et al. introduced the concept of Â“average breastÂ” dose, which was based on a simplified model of the breast as a 6 cm thick homogeneous glandular volume (50% adipose/50% water equi valent) surrounded by a 0.5 cm of adipose tissue, figure 28. Hammerstein et al. readily admitted in th e article that the "average breast" concept was assumed due to the lack of population data . Hammerstein et al . suggested that the average breast dose would be an appropriate measure in comparing different systems but should not be used for quantifying individual risks until such time that population data would be available for breast tissue composition. Hammerstein et al. also pr oposed the procedure for determining dose at any depth in an appropriate tissue substitute, equation (2 -5). The normalized average glandular dose (DgN) was based on a measure of exposure in the tissue substitute (Xg), the appropriate
28 exposure-to-dose conversion factor (fg), the mass of the target tissue (mt) and the exposure free-in-air (FIA). 1 1) ( FIA t g g gNX m f X D (rad R-1) (2-5) The authors stated that the energy abso rbed in glandular tissue was the most pertinent quantity for the ri sk evaluations in mammography.28 Hammerstein et al. concluded the article by stati ng that the "mean dose to the gland" (AGD) for an "average breast" (50% glandular) could be used for comparing radiographic techniques. Stanton et al., 1984 Stanton and colleagues evaluated a similar imaging system as Hammerstein et al.28 The authors expanded the Hammerstein idea of Â“average breastÂ” phantom, figure 2-8(a). The authors developed a phantom that was se micircular (15 cm diameter) and composed of BR 12 tissue substitute, figure 2-8(b-d). They examined and established that the area of a compressed breast had little effect on breast dose (~10%) when examining breasts with a compressed area of 35-270 cm2. Stanton and colleagues refined the DgN calculation proposed by Hammerstein et al.28 and developed working curves to determine glandular dose based on equation (2-6). 1 FIA g gNX D D (rad R-1) (2-6) The authors concluded that the average glandular dose was independent of breast tissue composition. This conclusion was not supported by their measurements, which suggested a change of 10% for xero mammography and 20% for screen-film mammography for changes in glandularity of 20-60%. The FDA Handbook of Glandular Tissue Doses in Mammography and the Nati onal Council on Radiation Protection and Measurements Report 85 adopted the concept of average glandular dose because it
29 represented the best estimate of th e glandular tissue absorbed dose.6,52 The methodology proposed by Hammerstein et al. and refined by Stanton wa s used extensively in the reports.28 The FDA handbook did however, modify the adipose tissue layer to 0.4 cm, based on Monte Carlo work by Rosenstein et al..52,53 Dance, 1990 Dance developed a Monte Carlo model that expanded the simplified model proposed by Hammerstein et al. and Stant on et al. by adding the compression paddle and the film receptor.28,54,55 The Stanton et al. glandular dos e calculation was modified to include a phantom conversion factor (p ) shown in equation (2-7), where DgN is the dose conversion factor, K is the air kerma and Dg is the mean glandular dose.54 1) ( p K D Dg gN (mGy mGy-1) (2-7) The conversion factor of phantom and sta ndard breast enabled the author to use material that was not necessar ily tissue-equivalent but rather convenient and inexpensive. The material used was Perspex, one of the trade names for PMMA (polymethyl methacrylate or polymethyl-2-methylpropanoate ), which is commercially-available as inexpensive acrylic. The simulation was made without the patient being modeled and it ignored backscatter from the patient back in to the imaging field. The authors used 90 different x-ray spectra with a variety of beam qualities to compute p and DgN, table 2-3. Wu et al., 1991, 1994 Wu et al. used Monte Carlo techniques to determine the standard-breast-conversion factor (DgN), equation (2-8). 1 FIA g gNX D D (rad R-1) (2-8)
30 Their model for the Monte Carlo calculati ons was adopted from the FDA model, a semi-elliptical phantom with a 0.4-cm adipos e layer covering a homogeneous mixture as described previously, figure 2-8 (a,d). Howeve r, the Wu model did not take into account the image receptor and cassette. The cal culations took into account all photon interactions previous discussed, for example, photoelectric effect, in coherent scattering and coherent scattering. DgN values were based on one million particle histories. Extensive tables for DgN were developed that correl ated kVp (23-35), HVL (0.24-0.43 mm Al equivalent) and breast thickness (3-8 cm) for Mo/Mo, Mo/Rh and Rh/Rh targetfilter combinations. The tables were created for 100% glandular tissue, 50% glandular/50% adipose tissue and 100% adi pose tissue. The authors indicated that glandularity could be adjusted by interpolation between the provided tables. The authors recommended that XFIA be measured 4 cm from the chest wall and centered transversely in the field slightly above the grid assembly. They further recommended that HVL be measured with the compression paddle in a narrow, collimated field. Sobol and Wu, 1997 Sobol and Wu developed a parametrization algorithm in Visual Basic to enable computerized calculation of the average gla ndular dose calculation based on breast tissue composition, kVp, HVL and breast thickness, using the data collected previously.56,57 The authors explicitly warned that the comput er codes presented should not be used to extrapolate beyond the ranges of the original work. The code was developed for use with ranges of kVp(23-35), HVL(0.240.43 mm Al equivalent) and breast thickness (3-8 cm) for Mo/Mo, Mo/Rh, and Rh/Rh ta rget-filter combinations.
31 Dance et al., 1999 Dance et al. discussed the estimati on of breast dose during a diagnostic mammogram. They suggested that dose measurements be quantified by using thermoluminescent dosimeters (TLD) placed on the breast or by correlating patient dose with machine exposure parameters. They de scribed that a major pitfall could be the position of the ionization chamber, because the position could cause a variation in the estimate of mean glandular dose by 12%. They suggested that machine output be measured during the annual medical physics surv ey to give an indica tion of the stability of the mammography unit. In discussing TLD use, they suggested that placement would have to be such that it would not interfere with the clinical image. The other drawback to using a TLD on a patient is the obvious pro cessing time and quality control needed to provide an accurate measurement. The aut hors also discussed the strong dependence of the TLD with the dosimeter thickness at lo w energies. At the e nd of the article, the authors reviewed the national protocols in Europe and the United States and concluded that, because of the variation in protocols, it was hard to compare them. They reported that the European protocols called for 1.0-1.5 background optical density (OD) in mammograms, whereas the United States and United Kingdom called for 1.4-1.8 OD. Dance et al. suggested that th ere was a need for an international consensus protocol. Simulating Breast Tissue Population Demographics A clear understanding of the population an thropometric measurements for the tissue of interest is needed before tissue equivalent materials can be developed as substitutes. Unless otherwise i ndicated, the accuracy of the va lues presented is plus or minus one standard deviation.
32 Geise and Palchevsky, 1996 The research by Geise and Palchevsky in Minnesota was performed to identify the best composition of tissue equivalent materi al to test automatic exposure controlled units.29 They underscored the lack of substantiation for the ACR's description of the Â“average breastÂ” as being representative of the typical breast. The ACR 50%adipose and 50%-glandular tissue ph antom had in fact no basis in population studies; instead, as described above, it was suggested by Hammer stein et al. in their paper in 1979 to evaluate units against a single model. The Geise and Palchevsky evaluated a manufacturerÂ’s phantom that was composed of phantoms of simulated 0, 30, 50, 70 and 100% glandular tissue to rela te mathematically the mammography unit parameters to patient breast glandularity. After calcula ting an empirical formula that correlated glandularity to machine parameters, equa tion (2-9), they evaluated 1578 mammograms from 415 patients retrospectively. Equation (2-9 ) was developed to relate glandularity (g) to compressed breast thickness (t), mAs, and kVp. ) / 1 . 7812 / 3 . 1935 9 . 1481 ( ) (ln( ) / 24 . 237 / 37 . 109 409 . 39 (2 8 2t t kVp mAs t t g (2-9) Using equation (2-9), Geise and Palchevs ky determined that the average breast thickness for their population was 5.2 cm a nd the average effective glandularity was 34%. They concluded that a 30% glandular ti ssue composition was the best glandularity content to simulate AEC response of an aver age breast in their su rvey population. They also suggested that 50% glandul ar material could be utilized, but the thickness of the phantom needed adjustment to simula te the same phototimer response.
33 Dance et al., 1999 Dance et al. evaluated the use of the dose conversion factors to determine mean glandular dose in the UK and Europe.50 They evaluated two popul ations of women: 40-49 years of age and 50-64 years of age. In determining the AGD conversions, Dance et al. evaluated these two groups for breast size a nd glandularity, but they did not determine the mean breast thickness or mean breast glandu larity of the two popul ations. Instead, the authors stated that the 50-64 year-old group had breasts that were on average compressed to 4.5-5.0 cm thick with an associated glandularity of 41% to 33%. Figure 2-9 demonstrates the demographics of the tw o groups in comparison to other authors discussed. Maskarinec and Meng, 2000 Maskarinec and Meng conducted a case-control study that evaluated a breast cancer population in Hawaii to determine correlation be tween breast density and risk of cancer. 33 The authors stated that Hawaii has a unique population that has e ssentially equal parts of Chinese, Japanese, Filipino, Native Hawa iian, and Caucasian ancestry. The authors evaluated glandularity by a planimetry met hod of digitized area (p ercent dense area). The control population of 647 subjects was select ed to match the age, race, glandularity, breast thickness and various other parameters of the breast cancer cases (647 patients) used in the study. The authors determined th at their average compressed breast thickness was 3.13 Â± 1.45 cm with an average glandularity of 26.8 Â± 16.5%. Rosenberg et al., 2001 Rosenberg it al. evaluated a method to cap ture machine parame ters from routine mammograms in New Mexico.58 They evaluated 2738 patie nts with 13,621 radiographs, in which the glandularity was based on the equation determined by Dance et al. 55 They
34 determined that their average compressed breast thickness was 4.9 1.3 cm, with an associated breast glandularity of 42.9 21%. Kruger and Schuler, 2001 Kruger and Schuler surveyed the mean glandular dose for women undergoing mammography in their institution in Minnesota.59 They evaluated 6,006 patients with 24,471 radiographs; the age distribution was not specified. They concluded that their average breast was 5.1 1.3 cm thick, with an associated 28% glandularity. Other researchers Various other researchers have made AGD me asurements to the standard breast as it currently exists. The data from their research that is pertinent to this discussion is the mean compressed breast thickness measurem ents, which averaged to be 5.03 0.69 cm.59-62 Volumetric Breast Measurements: Fife, 1990 The innovative concept of this study by Fife is that the author m easured the breast, not only in compressed thickness, but also in compressed width and breadth, giving insight into a possible range for deve loping a true population phantom series.63 The authors evaluated the demogr aphics of their patient population (216 patients) and determined that the average breast co mpressed thickness was 5.2 1.1 cm, with associated compressed breadth of 8.1 2.1 cm and a width of 18 2.4 cm. The authors defined the width of the breast to be the di stance measured from the radiograph along the axilla to sternum and breadth is the distan ce from the chest wall to the mammilla, figure 2-7. The authors did not indi cate the average compression pressure used for their compressed thickness measurements.
35 Protocol Tools The relevant historical background and mammographic unit details have been presented preceding sections, but in order to develop a viable clinical protocol several tools will be needed. The tools that will be utilized in this study are tissue-equivalent breast phantoms, dosimetric devices and Monte Carlo modeling technique. The proceeding sections will provide the re levant background tools be utilized. Tissue Equivalent Material In Vienna in 1906, R. KeinbÃ¶ch was the first to publish the use of tissue-equivalent (TE) materials by using water as a substitute fo r muscle tissue in his pursuit to develop a dosimetry method for Roentgen rays (x rays).64 After KeinbÃ¶ch efforts, many attempts were made to develop TE materials that had characteristics for being reused and molding. After the introduction of wax based TE mate rials by Baumeister in 1922, numerous radiological characteristic improvements were made to wax that spanned from 1937 to 1956.51 Then in 1956, Shonka et al. introduced a conducting plasti c an alternative to wax. Soon thereafter in 1961-1962, natural rubber a nd isocyanate rubber substitutes were introduced by Stacey et al. and Alderson et al. r espectively.51 The non-wax based materials developed had retained the molding characteristics but had superior mechanical characteristics. Concurrently to the deve lopment of TE materials, plastics, first introduced in 1862, had become perv asive in society and industry.65 The molding and mechanical characteristics of plastics made th en a viable candidate for TE materials. In 1977, White used a comprehensive appr oach to develop TE materials taking commercially-available plastics and adding chemical compone nts to mimic radiological interactions of tissues.51 White chose materials to mimic the reference tissue in both mass density and radiological interact ion characteristics. Several authors have evaluated breast
36 and adipose tissues and their compositions are listed in table 2-4 along with Whiteâ€™s group glandular TE material (BR 12) and Adipose TE material (AP6). The most commonly used TE breast material is BR 12. Dosimetry Devices A major challenge for mammography dosim etry is choosing the appropriate dosimeter to quantify the air kerma needed to calculate the AGD. The dosimeter must be sensitive to the low-energy photons that ar e used in mammography. In this study, the metal oxide semiconductor field effect tran sistor (MOSFET) and fiber-optic coupled dosimeters will be evaluated for use in quantifying air kerma dose. Metal oxide semiconductor field effect transistor In 1974 a metal oxide semiconductor fiel d effect transistor (MOSFET) was reported to measure ionizing radiation by Holmes-Siedle .66 The technology in fact quantifies radiation damage to a semiconductor. The damage is irreversible, and therefore the dosimeters have a limited lifetime that result in a recurring cost. The MOSFET has a threshold voltage on the me tal gate that allows current to flow between the source and the drain, figure 2-10. During radiation intera ctions, electron-hole pairs are created in the silicon oxide layer. A positive bias (3 to 10 volts) applied on the gate during the exposure separa tes the electron -hole pairs. The electrons subsequently move to the gate, whereas the holes move toward the silicon oxi de, creating a positive charge. The positive charge produces a negative shift in the voltage required to allow current to pass through the MOSFET.67 The shift in voltage (a s little as 1 mV) is proportional to the absorbed-dose deposited.68 The sensitivity can be increased by increasing the applied positive bias or increasing the thickness of the silicon oxide.67
37 The MOSFET dosimeter has been evaluate d for use in diagnostic radiology, radiotherapy, and radioimmune therapy.67-73 The feasibility of using MOSFETs in diagnostic or screening mammogr aphy was first evaluated by Dong et al., and a follow up study was conducted by Benevides and Hintenlang.74-76 The system used in this study has an oxide layer that is one m thick, whereas MOSFETs used in radiotherapy are only half as thick.67 The increased thickness of the oxide increases the number of available electron-hole pairs leadi ng to improved statistics a nd reproducibility. Fiber optic-coupled dosimeter A variety of materials produce luminescen ce immediately (radioluminescence, RL) or when heated (thermoluminescence), or stimulated with a visible light spectrum (optically-stimulated luminescence, OSL) after having been exposed to ionizing radiation. The basic theory is that scinti llating photons are libera ted when the electronhole pairs formed from the exposure to ionizing radiation recombine at the activator site. Luminescent materials such as copper-doped silicon dioxide (SiO2:Cu) have several unique characteristics. Their prompt RL is large in magnitude. They have an OSL capability and they can be fused to glass, such as fiber optic fibers. A direct application of this material has been made in the fi ber-optic-coupled dosimeter (FOCD). The FOCD dosimeter principle theory of operatio n is the detection of RL from Cu+1-doped fused quartz dosimeter. After being e xposed to ionizing radiation, Cu+1 ions are raised to an excited state, which de-excite by RL. 77 Given that the material can be fused to glass, it is fully compatible with the fiber optic fibers wh ich are used to transmit the resulting signal from an exposure.
38 Monte Carlo Method The Monte Carlo (MC) method provides an accurate solution to a variety of radiation interaction problems by perfor ming statistical sampling experiments. The method was named after the capital of Monaco, Mont e Carlo, because of the association with the game of Roulette, which represen ts a simple random number generator. The history of the MC method be gan in early 1871, when a Civil War soldier recovering from wounds wrote a paper on an experimental method of determining .78 In 1908, W.S. Gosset, a student of a British statis tical school used experimental sampling to determine the distribu tion of the correlation coefficien t for two populations. In 1931, A. N. Kolmogorov (1903-1987) showed the re lationship between Markov stochastic processes and certain integr o-differential equations. Nicholas Metropolis and Stanislaw Ulam did the majority of the work toward developing the formal MC method during the Manhattan Project (c irca 1945). Their work simulated probabilistic problems such as random neutron diffusion in fissile material. J. Von Neumann and Stanislaw Ulam also developed two routines used today called Russian Roulette and splitting. Th e method was published in a paper of the Journal of the American St atistical Association in 1949 titled "The Monte Carlo Method". In recent years, MC methods have b een applied in diagnostic radiology to determine patient doses.55,57,79-82 The major drawbacks to using the MC technique are the computing time to accumulate a statistically -significant number of histories and the detailed knowledge of the geometry of the in cident radiation beam. Monte Carlo Neutral Particle Code version 5 (MCNP-5) devel oped by the US Department of Energy was utilized in the study. MCNP-5 was written in ANSI-Standard Fortran 90 global data
39 format that is accessed by individual modules.10 MCNP-5 accounts for coherent scattering, incoherent scattering, photoelec tric effect absorp tion with fluorescence emission and pair-production.10 The preceding sections have provided the hist orical perspective and tools needed to develop a dosimetry protocol that can be uti lized in the clinical environment to estimate an individual's average glandular dose resulting from mammography imaging with greater accuracy than is currently possible.
40 (Upper Left Quadrant) (Lower Left Quadrant) (Upper Right Quadrant) (Lower Right Quadrant) TDLU Duct Lumen Deep fascia Pectoralis major Subcutaneous adipose Retromammary adipose Rib Sinus lactifer Areola Mammilla Ductus lactifer Pars infundibularis Extralobular terminal duct Porus excretoris Cancer genesis area Myoepithellial cells Epithellial cells Lobes Granular acini Figure 2-1. Anatomical features of the human breast with microsc opic detail of the TDLU and its ducts. Adapted from Kopans and Tortora and Anagnostakos.13,83
41 (A) (B) (C)R egion Materials 1. 1.56 mm nylon fiber9. 0.32 mm specks 2. 1.12 mm nylon fiber10. 0.24 mm specks 3. 0.89 mm nylon fiber11. 0.16 mm specks 4. 0.75 mm nylon fiber12. 2.00 mm tumor-like mass 5. 0.54 mm nylon fiber13. 1.00 mm tumor-like mass 6. 0.40 mm nylon fiber14. 0.75 mm tumor-like mass 7. 0.54 mm specks15. 0.50 mm tumor-like mass 8. 0.40 mm specks16. 0.25 mm tumor-like mass 00001234 5 6 78 9 10 11 12 13 141516 Figure 2-2. (A) American College of Ra diology accreditation phantom. (B) A ACR accreditation phantom image. (C) Schema tic showing image quality objects.
42 Photon Energy (keV) 01020304050Interaction Coefficient (cm2/gm) 0.001 0.01 0.1 1 10 100 1000 10000 Coherent Scattering, R / , Glandular tissue Incoherenr Scattering, / , Glandular Tissue Photoelectric Effect, / , Glandular Tissue R / , Adipose Tissue / , Adipose Tissue / , Adipose Tissue Photon Energy (keV) 1020304050Total Linear Attenuation Coefficient (cm-1) 0.1 1 10 100% Adipose Tissue (ICRU-46) 100% Glandular Tissue (ICRU-46) (A) (B) Figure 2-3. (A) Mass attenuation comparison between glandular and adipose breast tissue. Data derived from National In stitutes of Standards and Technology, XCOM databases.84 (B) The graph shows a comparison between the interaction coefficients of ICRU-44 gl andular and adipose breast tissue.
Figure 2-4 (A) Raw x-ray tube target spectra at 28 kVp, spectra filtered with 0.03 mm of Mo, 0.025 mm of Rh, and linear attenuation coefficient for Mo in the same energy range. (B) Raw x-ray tube target spectra at 29 kVp, spectra filtered with 0.025 mm of Rh, and linear attenuati on coefficient for Rh in the same energy range. The figures illustrate why the characteristic x rays are not significantly attenuated in the 18-20 kVp range and 19-23 kVp as a result of the significant drop in value of the linear attenuation coefficient near the Mo and Rh K-edges respectively. Graph data were derived from National Institutes of Standards and Technol ogy, XCOM databases and the FDA, Center for Devices and Radiol ogical Health, X-ray spectra.84.
44 (A)Energy (kVp) 051015202530Photons 104 cm-2 mAs-1 0 5 10 15 20 25 cm-1) 0 200 400 600 800 1000 Raw Mo 28 kVp spectra Filtered Mo 28 kVp with Mo 0.030 mm Mo Linear Attenuation Coefficient Rh Linear Attenuation Coefficient Filtered Mo 28 kVp with Rh 0.025 mm Energy (kVp) 051015202530Photons 104 cm-2 mAs-1 0 2 4 6 8 10 cm-1) 0 200 400 600 800 1000 Raw Rh 29 kVp spectra Filtered Rh 29 kVp with Rh 0.025 mm Rh Linear Attenuation Coefficient (B)
45 6 degrees 24 degrees X-ray tube Tube port Filtration Collimation Compression Paddle Grid Image receptor with cassette Automatic Exposure Control Sensor Cathode Anode 0Â° Anode tilt 16Â° Anode tiltTypical Mammography Unit Tube Geometries 6Â° Tube tilt 24Â° Tube tilt(B) (A) Figure 2-5. Illustration of t ypical mammography unit and typi cal tube geometries. (A) Common components found on a typical mammography unit. (B) Two types of tube geometries available in curre nt commercial units. Adapted from Bushberg et al.31
46 Heel Effect 024Cathode 0 degree Anode tilt SID=65cm12 degreesReference Axis Central Axis 29 cmFocal Spot Anode 24 degrees ChestMamilla12==(B) (A) lengthwidth}} 24 degrees Figure 2-6. Illustration of focal spot and heel e ffect. (A) Heel effect of a unit with a zero degree ( ) anode tilt and a 24 degree ( ) tube tilt.31 The figure demonstrates that photons traveling along the 1 path are filtered less by the anode than those that travel along 2 path. The difference in phot on filtration results in an decreasing-intensity continuum from the chest wall to the mammilla. (B) Focal spot geometry of a 24 degree tube head tilt.31 The figure demonstrates the projected focal spot (PFS) va riation along the cat hode-anode axis. Adapted from Bushberg et al.31
47 Breast Edge Region Mammogram CC view Skin edge Width Breadth (Posterior Nipple Line) Uniform Region Dense region(A) (B) Figure 2-7. (A) A cross-sectional view of the compressed-breast edge and uniform region in a mammographic unit. (B) A craniocaudal mammogram view of skin edge, dense region, width, posterior nipple line, compressed-breast edge and uniform region.
48 Adipose/Gland Uniform Mixture Adipose tissue, 0.5 cm, taUniform Phantom Uniform PhantomAdipose Tissue, 0.4 cm Adipose/Gland Uniform Mixture sagittalAnterior-posterior cranial-caudal A. Hammerstein modelet al. B. S tanton modelet al. C. NCRP modeltbt-tba Figure 2-8. (A) Hammerstein et al. simplified model with an 0.5 cm layer of adipose tissue. The thickness of adi pose tissue is shown as ta and homogeneous glandular tissue is shown as tb. (B) Stanton et al. model with an 0.4 cm adipose layer tissue. (C)Uniform com position phantoms used in dosimetry measurements.6,28,54
49 Compressed Thickness (cm) 024681012Glandularity (%) 0 20 40 60 80 100 Geise et al.,1996 (+/SD) Dance, 2000, 40-49 yo Series Dance, 2000, 50-59 yo Series Mean Compressed Thickness (cm) 024681012Mean Glandularity (%) 0 20 40 60 80 100 Geise et al.,1996 Rosenberg et al. , 2001 (+/SD) Kruger and Schuler, 2001 Maskarinec and Meng, 2000, (+/-SD) Jamal et al., 2004 (+/SD) (A) (B) Figure 2-9. (A) The graph show s the relationship between br east tissue glandularity and compressed thickness. (B)The mean va lues tend to aggregate about 30-45% glandularity and 4.0-5.0 cm in thickness.29,33,55,58-60
50 p c h a n n e l M O S F E T X-ray bea m K a p t o n A u N i P o l y s i l i c o n G a t e S i O2S i O2g a t e o x i d e ( s e n s i t i v e r e g i o n )A l A lS o u r c eS i O2 P+P+S i S u b s t r a t e ( n t y p e )D r a i n Epoxy Figure 2-10. A MOSFET dosimeter functi onal schematic.Adapted from Thompson Nielsen.85
51 Table 2-1. Anatomical distribution of breast cance r in the right or left breast as described in Kopans13 Breast Quadrant Breast Cancer Distribution (%) Upper outer 52 Upper inner 15 Areola 14 Lower outer 11 Lower inner 8
52 Table 2-2. Radiographic features suggestive of malignancy as describe in Kopans. 12 Neodensity or new calcifications New masses or architectural distortion New clustered calcifications Findings that suggest a high probability of malignancy Spiculated lesions Fine, linear, br anching calcifications Findings that should arouse suspension A lesion with ill-defined margins A lesion with microlobulated margin Architectural distortion A distorted parenchymal edge Density increasing over time Clustered microcalcifications Hanging calcifications Findings that are probably benign Solitary circumscribed mass Solitary asymmetric duct Round, regular clustered calcifications Findings that support the po ssibility of malignacy Asymmetric breast tissue Asymmetric ducts Asymemteric veins Skin and trabecular thickening Nipple retraction, deviation, or inversion Enlarged axillary lymph nodes
53 Table 2-3. X-ray Spectra examined by Dance in developing DgN tables.26 HVL range (mm Al equiv) Target/Filter Combination Filter Thickness (Âµm) kV Range 0.25-0.45 Mo/Mo 30 25-35 0.45-0.70 W/Mo 60 23.35 0.50-0.80 W/Rh 50 24-35 0.55-0.90 W/Pb 50 25-35 0.50-2.00 W/Al na 23-50
54Table 2-4. The elemental constituents of adipose, breast glandu lar tissue and two composite plas tic tissue equivalent materials (BR 12 and AP 6)as described by Hammestein et al., White et al. and ICRU-44.22,23 Elemental Composition (% by mass) Source Tissue H C N O Others Density (gm cm-3) Adipose ICRU44 11.4 59.8 0.7 27.8 0.1 Na,0.1 S, 0.1 Cl 0.9500 White77 12.0 64.0 0.8 23.3 0.9200 Hamstn79 11.2 61.9 1.7 25.1 0.1 Ash (S,P,K,Ca) 0.930 Glandular ICRU44 10.6 33.2 3.0 52.7 0.1 Na, 0.1 P, 0.2 S, 0.1 Cl 1.020 White77 11.70 38.04 50.26 0.960 Hamstn79 10.2 18.4 3.2 67.7 0.5 Ash 1.040 Tissue Equivalent White77 BR12Glandular 8.68 69.95 2.37 17.91 0.14 Cl, 0.95 Ca 0.970 White77 AP6-Adipose 8.36 69.14 2.36 16.93 3.07 F, 0.14 Cl 0.920
55 CHAPTER 3 CHARACTERIZATION OF MOSFET DO SIMETERS FOR APPLICATION IN CLINICAL MAMMOGRAPHY Introduction Glandular tissue in the breast has been shown to be radiosensitive, yet current clinical practices do not record the individual patient doses for mammography procedures. The Mammography Quality Standa rds Act (MQSA) of 1992 instead requires that each facility performing mammogra phy have its dedicated mammography units deliver less than to 3.0 mGy (300 mrad) dos e for the standardized acrylic phantom specified by the American College of Radiology.12 Although the risk associated with ionizing radiation is low, and the benefit of mammography is without question, in recent times, ionizing radiation doses received dur ing diagnostic and interventional x-ray procedures have become a widening con cern as exemplified by pediatric CT and interventional cardiac fluoros copic procedures. However, i ndividual patient doses are not required to be documented even thou gh mammography constit utes a reoccurring exposure of breast tissues to ionizing radiation. In addition to current imaging technology, the continued clini cal demand for detailed imaging of patients has promoted new imaging modalities such as computed tomography of the breast, general digital mammography, and breast tomosynthesis that will require detailed studies to determine the risk associated with thes e procedures. These technolog ies promise to reduce patient dose and provide a better platform for visu alizing disease states without invasive procedures. Dosimetry can play a vital role in providing the data necessary to document
56 patient dose for future risk-benefit analysis. On of the methods to evaluate patient risk from a radiological procedure is to document patient dose while not interfering with the diagnostic information in the image. Evaluating the risk associated with a di agnostic radiological procedure has been acknowledged by the 2005 draf t recommendations from the International Commission on Radiological Protection (ICRP 2005). The ICRP 2005 states " That exposure is not limited by any regulatory process, but is controlle d by the physician, who therefore should be aware of the risks and benefits of the procedures involved" . 86 The direct outcome of the ICRP recommendations is that a physician mu st then balance the risk versus benefit equation by maximizing the diagnostic inform ation while maintaining the exposure to ionizing radiation as low as diagnostically achievable (ALADA). The physician and physicist that ignore ALADA have the potential of increasing patient dose unnecessarily. Until relatively recently, the primary method used by physicists to evaluate patient doses in general radiology was th e Bureau of Radiological Health1) handbooks that were published in the 1980s and present dose as a f unction of measured entrance skin exposure (Xese) .52 The current method used in ma mmography is described in ACR Quality Control Manual where a "free-in-air" (Xese(FIA)) entrance-skin-exposure measurement is coupled to the dose conversion factor resulting in an "average gl andular dose" to the breast.12 In either method, the physicist measures the HVL and Xese or XeseFIA with an ionization chamber and then selects the appropriate tabu lated dose-conversion factor for the target tissue or organ at the specified tube potent ial. The procedure is not designed for recording patient dose and is usually performed post-exposure, which is time-consuming. 1 Food and Drug Administration, Center for De vices and Radiological Health, Rockville, MD
57 In addition, the ionization chambe rs typically used are too large to be used in real-time because they would interfer e with the diagnostic image for individual patients. One possible method of acquiring XESE or XFIA is to use a thermoluminescent dosimeter (TLD) during or after the procedur e. However, TLDs require significant postexposure processing; they are visible in the diagnostic image; and are seldom used in a diagnostic setting. A recent paper by Forwar d and Dugan explored the possibility of using a new thin TLD that reso lves the issue of being visi ble, but the authors did not address the lengthy post-exposure processing.87 The primary drawback with TLDs is their inability to provide the patient dose im mediately post-exposure, that is, real-time. The ideal diagnostic radiology dosimeter would be one that provides a immediate response to the radiation field (real-time), is accurate, tissue-equivalent, inexpensive, easy to operate, exhibits no angul ar dependence and does not in terfere with the diagnostic information in the image. An ionization chamber meets the real-time requirement and accuracy but is typically too large to be used in the c linical image. The ionization chamber also does not satisfy the tissue-equiva lence and it has an angular dependence. An alternate possibility is a MOSF ET dosimeter for real time mammography exposure measurement. MOSFETs have been used to measure ionizing radiation since 1974.66 The medical applications of MOSF ET dosimeters have ranged from diagnostic radiology, to radiotherapy and radioimmune therapy.67-73 Previous diagnostic radiology evaluations used x-ray tube potential s throughout the energy range of 40-120 kVp. 67,70,71 The use of MOSFET dosimeters in mammography was introduced by Dong et al. as a means to estimate the mean glandular dose. Dong et al. evaluated MOSFET2 dosimeters 2 Thompson Nielsen Electronics Ltd.,25B, Northside Road, Ottawa, Canada K2H8S1
58 for tube potentials of 25 to30 kVp and reporte d an energy dependence of less than three percent. Their finding is c ontrary to the present study.74 In this paper, several models of MOSFET dosimeters have been characterized for use in clinical mammography. This research characterized calibra tion factors, linearity, angular dependence, and reproducibility of the five MOSFET models, TN-502RDS, TN1002RDI, TN-1002RDS, TN-1002RDM a nd TN-1002RD MOSFET, utilizing techniques initially described in Ch apter 2 and in Bower and Hintenlang.67 Material and Methods MOSFET and Patient Dose Verification System The current study characterized the 5 high-sensitivity MOSFET2) dosimeter models TN-502RDS (Micro), TN-1002RDI, TN-1002R DS (Micro), TN-1002RDM (Micro) and TN-1002RD for use in clinical mammography. Figure 3-1 shows the schematic structure of a MOSFET dosimeter. The dosimeters were utilized in conjunction with the Patient Dose Verification System2, model number TN-RD-15 (v ersion 3.6) consisting of a dosimeter reader and up to four dual-bias power supplies, (model TN-RD-22). In addition to the TN-RD-22 dual-bias power supply, a very-high sensitivity dual-bias power supply was characterized. The bias supply can accommodate up to five dosimeters. The TN-RD-22 dual-bias power supp ly has two sensitivity settings: standard and high. The very-high sensitivity dual-bias power supply has one sensitivity setting and requires no adjustment for use with high-sens itivity dosimeters. The standard, high and very-high sensitivity settings of the bias s upplies change the dosimeter bias voltage, correspondingly changing the dosimeter respon se. Typically, the dosimeter sensitivity requirements are matched with the bias supply sensitivity setting. The dosimeter reader displays the dosimeter response (mV) and th e accumulated dosimete r response (mV) for
59 each dosimeter. Computer-dosimeter interface software (Autosense version 2.1, Thomson and Nielsen Electronics, Ltd.) was used to facilitate data co llection and storage. The Autosense software enables the user to store and apply the calibration factor for each MOSFET, providing an automated conversi on of exposure in mV to roentgen or C kg-1. The reader display has a resolution of 1 mV which allows a minimum exposure of 8.0 X 10-6 C kg-1 (30 mR) to be recorded when utilizing very-high bias power supply. The high-sensitivity MOSFET dosimeter is composed of two MOSFET devices mounted together under a 1-mm bubble of black epoxy on a thin, flexible 20-cm-long cable, which is connected to the bias pow er supply by a 1.4 m long rugged cable. The size of the epoxy bulb that covers the MOSF ET dosimeter has lateral dimensions of 2.5 mm x 8 mm with a maximum thickness of 1.3 mm for a standard dosimeter, and 1.0 mm x 3.5 mm with a maximum thickne ss of 1.0 mm for a microdosimeter.88 Diagnostic Radiography Units General Electric Senographe DMR A General Electric Senographe DMR3) (GE-DMR) mammography x-ray unit provided the mammography x-ray field used in measurements of se nsitivity, linearity, angular response and reproducibility. The GE -DMR uses a high frequency generator and an x-ray tube, with a dual bi-metal track ta rget of molybdenum (Mo) and rhodium (Rh). The x-ray spectra were filtered using Rh or Mo filters. Tube potentials used for the characterization were incremented in two kV p increments over a ra nge appropriate to each target-filter combination: Mo/Mo (2434 kVp), Mo/Rh (30-40 kVp) and Rh/Rh (3648 kVp). The tube current-time stations used correspond to stations used in clinical 3 General Electric Medical Systems, Waukesha, WI
60 practice (50-320 mAs). The unit had a fixe d source-to-image receptor distance of 660 mm. Picker condenser discharge mobile radiography unit A Picker4) condenser discharge mobile ra diography unit, model UG-5M-02M, provided the diagnostic x-ray field used in th e angular response evaluation. The unit uses and operates in the diagnostic range of 50120 kVp with a tungsten target. The tube potential used for the angular characterization was 80 kVp. Exposure Measurements The ionization chamber used for referen ce exposures and dosimeter calibration was a RADCAL Corporation5) model MDH 1015 radiation monitor with a mammography probe (model 10x5-6M). The ionization ch amber has a dynamic exposure range of 5.16 x 10-9 C kg-1 (0.02 mR) to 0.026 C kg-1 (99.9 R) and exposure duration range of 0.1 msec to 99.99 sec. The mammography ionization chamber has a metal-coated polyester window with a polyacetal exterior and an active volume of six cm3. Dosimeter Angular Response The dosimeter angular response was evaluate d by rotating the dos imeter within the stationary x-ray field. The dosimeters were ro tated about their cable axis free-in-air with the bubble side facing and perpendi cular to the x-ray beam at 0 position. The dosimeters were exposed until they accumulated an indi cated exposure of approximately 200 mV. The dosimeters were evaluate d at 27 kVp and 80 kVp in 10 increments for a full 360 rotation. The 80 kVp evaluation was performe d to provide a standard for previous 4 Picker Dunlee Corporation, Bellwood, IL 5 RADCAL Corporation, 426 West Duarte Road, Monroma, CA 91016
61 measures of angular response. The dosim eters were evaluated in the mammography energy range at 27 kVp with a target-filter co mbination of Mo/Mo at 0 degrees and in 10 increments from 90 to 230 to evaluate the angular response resulting from the dosimeter construction. The repor ted value is an average of five exposures was taken at each 10-degree increment at both the di agnostic and mammographic energy ranges. The angular response was evaluated to dete rmine the working angular range of the dosimeters for clinical applications. In our methodology the MOSFET dosimeter are exposed in a fixed orientation so angular res ponse does not dramatically affect the results. Other investigators may wish to use dosimeters in other configurations. Dosimeter Sensitivity The MOSFET dosimeter sensitivity was measur ed over the x-ray tube potentials of 24 to 48 kVp to evaluate the energy response across in the mammography energy range. The dosimeters were fixed to the unde rside of the GE-DMR compression paddle approximately one centimeter from the face of the ACR phantom, as illustrated in figure 3-2. The target-filter combinations eval uated were Mo/Mo (24-34 kVp), Mo/Rh (30-40 kVp), and Rh/Rh (36-48 kVp). Five (5.16x10-4C kg-1 (2 R)) exposures were made at each energy range. The total exposure was kept c onstant for different kVps by adjusting the tube current-time product (mAs) station. The primary use of the dosimeter is to de termine the free-in-air exposure during the clinical imaging. This measurement can subs equently be used in the calculation of average glandular dose. The dosimeter position was selected to ensure that the dosimeter would not interfere with the phantom be ing imaged. The position also reduces backscatter to the dosimeters from the image receptor. The very-high bias power supply
62 was evaluated with the Mo/Mo target-filter combin ation at tube potentials of 24, 28, and 34 kVp using dosimeter models TN-1002 RDM and TN-1002 RD. Dosimeter Linearity Dosimeter linearity with exposure and e xposure rate were evaluated at tube potentials of 24 to 48 kVp at tube currenttime product stations of 63-320 mAs to cover the range used in clinical mammography. Th e dosimeters were exposed with the epoxy side to the x-ray beam as prev iously described and as illust rated in figure 3-2. The targetfilter combinations evaluated were Mo/M o (24, 28, 34 kVp), Mo/Rh (30, 34, 40 kVp), and Rh/Rh (36, 40, 42, 48 kVp). Five exposures were made at each tube potential (kVp) and tube current-time (mAs) station. Dosimeter Reproducibility Reproducibility was evaluated by exposi ng two dosimeters representing each model type to 30 kVp, 100 mAs (3.72 x 10-4 C kg-1 (1.44 R)), Mo/Mo for 20 total exposures. The dosimeters were ex posed as shown in figure 3-2. Results and Discussion Dosimeter Angular Response Angular response data are shown in figur e 3-3 through figure 3-5 normalized to the mean angular response of the dosimeters. The 0 orientation indicates that the epoxy bubble side of the dosimeter is facing and perpendicular to the x-ray field. The angular response for all dosimeter models show a ma rked decrease in sensitivity between the angles of 120 and 150 and again between 190 and 220 . A similar decrease in response was previously reported in the diagnostic energy range (70 kVp) using dosimeter model TN-502 RDI by Roshau et al . and Gladstone et al . 72 The asymmetric construction and aluminum components of the MOSFET dosimeter contribute to this
63 observed variation in the a ngular dependence, although Roshau and Hintenlang reported that the angular dependence was eliminated when the dosimeters were embedded in tissue-equivalent material.72 In the proposed clini cal application, mammography dosimeters will monitor a free-in-air exposure normal to the x-ray beam, which will be used to calculate the mean glandular dose. Since this is performed at 0o, the decreased response for certain orientations does not impact the clinical application in mammography. Dosimeter Sensitivity The sensitivity across the evaluated mammography energy range extended from 3.83 x 104 mV per C kg-1 for model TN-502RDS to 11.66 x 104 mV per C kg-1 for model TN-1002RDS. The sensitivity as a function of tube potential (kVp) for each MOSFET model is shown in Tables 3-1 through 3-5 and illustrated in figures 3-6 through 3-10. The same vertical scale is used in these figur es to facilitate inter-comparison of dosimeter performance. All dosimeter models showed an energy dependence for the tube potentials evaluated. The two models that showed the least energy dependence were TN-502RDS and TN-1002RDI, which were originally mark eted for radiotherapy applications. The highest sensitivity dosimeter is the TN-1002R DS, and the least sensitive is TN-502RDS. The TN-1002RDS was designed to be very sens itive to radiation fo r its application in skin dosimetry, and very low-dose environmen ts. The use of the very-high bias power supply increases in the dosimeter sensitivity by 1.45 Â± 0.04 for TN-1002RDM and TN1002RD dosimeters. The very-high bias power supply increases the dos imeter sensitivity without having to change dosimeter models . The observed energy dependence of the sensitivity response differs from th at previously reported by Dong et al . , who showed a flat energy response.74 Dong et al . varied tube potentials from 25 to 30 kVp at 100 mAs,
64 resulting in a variable expos ure, while this study deliver ed a constant exposure by varying the mAs for each tube potential.74 This methodology en sures the delivery of a constant exposure in the sensitivity evaluation.67,71 Dosimeter Linearity All dosimeter models showed a linear res ponse to exposure for th e tube potentials and tube current-time stati ons evaluated as shown in figure 3-11 through figure 3-14. This finding is consistent with previo us investigations in the diagnostic and mammography energy ranges. 67,74 Dosimeter Reproducibility The reproducibility of the dosimeters varied with model, as shown in Table 3-6. Reproducibility of a single dosimeter measurement ranged from 15.5% to 31.8%. In addition to relative low reproducibilit y, MOSFET dosimeters would randomly not provide a reading contrary to adjacent dosimeters. Improve d accuracy therefore requires the utilization of either multiple measurements or multiple dosimeters used for a single exposure measurement in order to improve accuracy. Conclusions The overall performance of the MOSFET m odels evaluated in this study indicates that they can be clinically applicable in the mammography energy range provided that the conditions described below ar e adequately addressed. The decreased angular response of the dosimeters when irradiated between 120 and 150 and again between 190 and 220 in reference to the average angular response would not significantly influence clinical mammography applications, which would most likely radiate the dosimeter in a normal orientation.
65 The energy dependence must be taken into ac count, as it will affect the measured exposure. Energy dependent correction factors are required to accommodate the varying clinical tube potentials. The reproducibility is inadequate for a single dosimeter; however, it can be improved by utilizing multiple dosimeters, which, facilitates identifying aberrant dosimeter readings and statistical errors. Fu rther improvement in reproducibility can be made by reading the dosimeters at a constant time following exposure preferably within 20 seconds post-exposure. The combination of the Micro MOSFET model TN-1002 RDS used in conjunction with the high-sensitivity m ode of the TN-RD-22 dual bias power supply, or MOSFET model TN-1002RDM used in conjunction with the very-high dual bi as power supply are the pairs best-suited for clinical mammogra phy due to their inhere ntly-high sensitivity responses. Thomson Nielsen MOSFET dosimeters may be used in the evaluation of mammography dose, provided appropriat e controls are exercised. A possible measurement protocol would utilize five dosimeters integrated through a single bias supply, placed as illustrated in figure 3-1, fixed to the underside of the compression paddle and approximately one centimeter anterior to the anterior portion of the breast. The dosimeters should be calibrated to the free-in-air exposure at the chest wall to facilitate the calculation of average gla ndular dose. Under th ese conditions, MOSFET dosimeters can provide a viable option as a real-time dosimeter in the mammography energy range (22-50 kVp).
66 E p o x y A u N i P o l y s i l i c o n G a t e S i O2S i O2g a t e o x i d e ( s e n s i t i v e r e g i o n )A l A lS o u r c eS i O2 p c h a n n e l M O S F E T P+P+S i S u b s t r a t e ( n t y p e )D r a i n X-ray beam (A) (B) Figure 3-1. (a) A schematic of th e typical clinical orientation of the dosimeter to the x-ray beam. (b) The schematic illustra tes a P-channel MOSFET dosimeter indicating internal compositi on and regions of interest.
67 ACR phantom ionization chambe r MOSFETS Image recepto r Compression Paddle Figure 3-2. The schematic illustrate s the experimental setup used fo r the MOSFET dosimeter evaluations.
68 Angle(degrees) 0100200300400Angular Reponse -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 TN-502RDS A ngular Response (Diagnostic) Angle_Mammographic vs TN-502RD Figure 3-3. The graph shows the angular re sponse of the dosimeter model TN-502RDS in the diagnostic and mammographic en ergy range. The angular-response ratio was normalized, (mVi mV-1 0 Ã»), to the zero-degree angular response of the dosimeter. The dosimeter response at 0Â° corresponds to the dosimeter being exposed epoxy bubble side normal to the x-ray beam.
69 Angle(degrees) 0100200300400Angular Reponse -2 -1 0 1 2 3 TN-1002 RD A ngular Response (Diagnostic) Angle_Mammographic vs TN-1002RD Figure 3-4. The graph shows the angular re sponse of the dosimeter model TN-1002RD in the diagnostic and mammographic en ergy range. The angular-response ratio was normalized, (mVi mV-1 0 Ã»), to the zero-degree angular response of the dosimeter. The dosimeter response at 0Â° corresponds to the dosimeter being exposed epoxy bubble side normal to the x-ray beam.
70 Angle(degrees) 0100200300400Angular Reponse -2 -1 0 1 2 3 TN-1002 RDM Angular Response (Diagnostic) Angle_Mammographic vs TN-1002RDM Figure 3-5. The graph shows the angular re sponse of the dosimeter model TN-1002RDM in the diagnostic and mammographic en ergy range. The angular-response ratio was normalized, (mVi mV-1 0 Ã»), to the zero-degree angular response of the dosimeter. The dosimeter response at 0Â° corresponds to the dosimeter being exposed epoxy bubble side normal to the x-ray beam.
71 Tube Potential (kVp) 222426283032343638404244464850Sensitivity (mV kg C-1) 3.0e+4 4.0e+4 5.0e+4 6.0e+4 7.0e+4 8.0e+4 9.0e+4 1.0e+5 1.1e+5 1.2e+5 1.3e+5 1.4e+5 Mo/Mo Mo/Rh Rh/Rh Figure 3-6. The graph shows the sensitiv ity response in the mammographic energy region of MOSFET dosimeter model TN1002RDI used in conjunction with a high sensitivity dualbias power supply.
72 Tube Potential (kVp) 222426283032343638404244464850Sensitivity (mV kg C-1) 3.0e+4 4.0e+4 5.0e+4 6.0e+4 7.0e+4 8.0e+4 9.0e+4 1.0e+5 1.1e+5 1.2e+5 1.3e+5 1.4e+5 Mo/Mo Mo/Rh Rh/Rh Figure 3-7. The graph shows the sensitiv ity response in the mammographic energy region of MOSFET dosimeter model TN502RDS used in conjunction with a high sensitivity dualbias power supply.
73 Tube Potential (kVp) 2224262830323436384042Sensitivity (mV kg C-1) 3.0e+4 4.0e+4 5.0e+4 6.0e+4 7.0e+4 8.0e+4 9.0e+4 1.0e+5 1.1e+5 1.2e+5 1.3e+5 1.4e+5 Mo/Mo Mo/Rh Figure 3-8. The graph shows the sensitivity response in the mammographic energy region of MOSFET dosimeter model TN-1002RDS used in conjunction with a high sensitivity dual-bias power supply used in conjunction with a high sensitivity dual-bias power supply..
74 Tube Potential (kVp) 222426283032343638404244464850Sensitivity (mV kg C-1) 3.0e+4 4.0e+4 5.0e+4 6.0e+4 7.0e+4 8.0e+4 9.0e+4 1.0e+5 1.1e+5 1.2e+5 1.3e+5 1.4e+5 Mo/Mo Mo/Rh Rh/Rh Mo/Mo Very-high dual Bias Power Supply Figure 3-9. The graph shows the sensitiv ity response in the mammographic energy region of MOSFET dosimeter model TN -1002RDM used in conjunction with a high and very-high sensitivity dual-bias power supply.
75 Tube Potential (kVp) 222426283032343638404244464850Sensitivity (mV kg C-1) 3.0e+4 4.0e+4 5.0e+4 6.0e+4 7.0e+4 8.0e+4 9.0e+4 1.0e+5 1.1e+5 1.2e+5 1.3e+5 1.4e+5 Mo/Rh Mo/Mo Rh/Rh Mo/Mo, Very-high Dual Bias Power Supply Figure 3-10. The graph shows the sensitiv ity response in the mammographic energy region of MOSFET dosimeter model TN -1002RD used in conjunction with a high and very-high sensitivity dual-bias power supply.
76 Tube Current-time Product (mAs) 50100150200250300350Dosimeter Response (mV) 0 20 40 60 80 100 24 kVp (Mo/Mo) 28 kVp (Mo/Mo) 34 kVp (Mo/Mo) 30 kVp (Mo/Rh) 34 kVp (Mo/Rh) 40 kVp (Mo/Rh) 36 kVp (Rh/Rh) 36 kVp (Rh/Rh) 48 kVp (Rh/Rh) Figure 3-11. The graph shows the linearit y of the dosimeter response in the mammographic energy region of MO SFET dosimeter model TN-1002RDI, (R2 > 0.98).
77 Tube Current-time Product (mAs) 50100150200250300350Dosimeter Response (mV) 0 20 40 60 80 100 24 kVp (Mo/Mo) 28 kVp (Mo/Mo) 34 kVp (Mo/Mo) 30 kVp (Mo/Rh) 34 kVp (Mo/Rh) 40 kVp (Mo/Rh) 36 kVp (Rh/Rh) 42 kVp (Rh/Rh) 48 kVp (Rh/Rh) Figure 3-12. The graph shows the linearit y of the dosimeter response in the mammographic energy region of MOSF ET dosimeter model TN-502RDS, (R2 > 0.98).
78 Tube Current-time Product (mAs) 50100150200250300350Dosimeter Response (mV) 0 20 40 60 80 100 120 24 kVp (Mo/Mo) 28 kVp (Mo/Mo) 34 kVp (Mo/Mo) 30 kVp (Mo/Rh) 34 kVp (Mo/Rh) 40 kVp (Mo/Rh) 36 kVp (Rh/Rh) 40 kVp (Rh/Rh) 42 kVp (Rh/Rh) Figure 3-13. The graph shows the linearit y of the dosimeter response in the mammographic energy region of MOSF ET dosimeter model TN-1002RD, (R2 > 0.98).
79 Tube Current-time Product (mAs) 050100150200250300350Dosimeter Response (mV) 0 20 40 60 80 100 120 140 160 180 200 24 kVp (Mo/Mo) 28 kVp (Mo/Mo) 34 kVp (Mo/Mo) 30 kVp (Mo/Rh) 34 kVp (Mo/Rh) 40 kVp (Mo/Rh) 36 kVp (Rh/Rh) 40 kVp (Rh/Rh) 42 kVp (Rh/Rh) Figure 3-14. The graph shows the linearit y of the dosimeter response in the mammographic energy region of MO SFET dosimeter model TN-1002RDM, (R2 > 0.99).
80 Table 3-1. MOSFET dosimeters, model TN-1002 RDI, sensitivity conversion factors at mammography energy range. TN-1002 RDI Sensitivity Tube Potential (kVp) Target/Filter (mV kg C-1) x 104 (mV kg C-1) ( ) x 104 (mV R-1) (mV R-1) ( ) 24 Mo/Mo 4.17 0.38 10.75 0.97 26 Mo/Mo 4.52 0.27 11.65 0.70 28 Mo/Mo 4.99 0.44 12.88 1.13 30 Mo/Mo 4.96 0.31 12.80 0.79 32 Mo/Mo 5.21 0.48 13.45 1.23 34 Mo/Mo 5.41 0.31 13.95 0.79 30 Mo/Rh 5.63 0.37 14.52 0.97 32 Mo/Rh 5.41 0.24 13.96 0.63 34 Mo/Rh 6.07 0.30 15.66 0.77 36 Mo/Rh 5.90 0.31 15.22 0.80 38 Mo/Rh 5.92 0.29 15.27 0.74 40 Mo/Rh 5.91 0.34 15.26 0.87 36 Rh/Rh 6.88 0.31 17.75 0.79 38 Rh/Rh 7.03 0.37 18.14 0.95 40 Rh/Rh 7.44 0.30 19.20 0.78 42 Rh/Rh 7.14 0.35 18.42 0.91 44 Rh/Rh 7.56 0.25 19.51 0.65 46 Rh/Rh 7.38 0.25 19.04 0.64 48 Rh/Rh 7.57 0.36 19.53 0.94 Average 6.06 0.08 15.63 0.20
81 Table 3-2. MOSFET dosimeters, model TN-502 RDS, sensitivity conversion factors at mammography energy range. TN-502 RDS Sensitivity Tube Potential (kVp) Target/Filter (mV kg C-1) x 104 (mV kg C-1 ) ( ) x 104 (mV R-1) (mV R-1) ( ) 24 Mo/Mo 3.83 0.34 9.89 0.88 26 Mo/Mo 4.12 0.38 10.64 0.97 28 Mo/Mo 4.27 0.18 11.03 0.46 30 Mo/Mo 4.36 0.44 11.24 1.14 32 Mo/Mo 4.55 0.18 11.74 0.47 34 Mo/Mo 4.16 0.35 10.74 0.90 30 Mo/Rh 4.15 0.26 10.70 0.68 32 Mo/Rh 4.39 0.22 11.33 0.57 34 Mo/Rh 4.50 0.33 11.61 0.84 36 Mo/Rh 4.43 0.33 11.44 0.85 38 Mo/Rh 4.42 0.17 11.41 0.44 40 Mo/Rh 4.34 0.15 11.20 0.38 36 Rh/Rh 4.94 0.15 12.75 0.39 38 Rh/Rh 4.70 0.32 12.13 0.84 40 Rh/Rh 4.53 0.27 11.70 0.69 42 Rh/Rh 4.56 0.21 11.75 0.54 44 Rh/Rh 4.67 0.21 12.05 0.54 46 Rh/Rh 4.77 0.23 12.30 0.59 48 Rh/Rh 4.78 0.24 12.33 0.61 Average 4.45 0.63 11.47 0.16
82 Table 3-3. MOSFET dosimeters, model TN -1002 RDS, sensitivity conversion factors at mammography energy range. TN-1002 RDS Sensitivity Tube Potential (kVp) Target/Filter (mV kg C-1) x 104 (mV kg C-1) ( ) x 104 (mV R-1) (mV R-1) ( ) 24 Mo/Mo 10.68 1.17 27.55 3.01 26 Mo/Mo 10.23 1.22 26.38 3.16 28 Mo/Mo 11.62 1.50 29.99 3.87 30 Mo/Mo 11.35 1.27 29.29 3.28 32 Mo/Mo 10.38 0.49 26.77 1.29 34 Mo/Mo 11.26 1.39 29.04 3.59 30 Mo/Rh 11.58 1.48 29.88 3.82 32 Mo/Rh 11.66 1.08 30.09 2.78 34 Mo/Rh 11.49 1.41 29.64 3.63 36 Mo/Rh 11.66 1.12 30.08 2.89 38 Mo/Rh 10.56 1.34 27.23 3.46 40 Mo/Rh 11.60 1.51 29.93 3.90 Average 11.20 0.37 28.82 0.95
83Table 3-4. MOSFET dosimeters, model TN1002 RDM, sensitivity conversion factors at mammography energy range. TN-1002 RDM Sensitivity Tube Potential (kVp) Target/Filter (mV kg C-1) x 104 (mV kg C-1 ) ( ) x 104 (mV R-1) (mV R-1) ( ) 24 Mo/Mo 8.16 0.58 21.06 1.51 26 Mo/Mo 8.38 0.95 21.63 2.44 28 Mo/Mo 8.60 0.26 22.19 0.66 30 Mo/Mo 7.70 1.80 19.85 4.65 32 Mo/Mo 9.30 0.95 23.99 2.46 34 Mo/Mo 8.55 0.50 22.07 1.30 30 Mo/Rh 8.81 0.59 22.72 1.52 32 Mo/Rh 8.98 1.04 23.17 2.68 34 Mo/Rh 9.40 0.59 24.26 1.52 36 Mo/Rh 9.76 0.38 25.19 0.98 38 Mo/Rh 8.99 0.34 23.18 0.88 40 Mo/Rh 9.31 0.89 24.02 2.30 36 Rh/Rh 10.50 0.44 26.98 1.14 38 Rh/Rh 10.40 1.17 26.83 3.02 40 Rh/Rh 10.60 1.00 27.35 2.59 42 Rh/Rh 10.70 0.43 27.56 1.10 44 Rh/Rh 10.00 0.49 25.91 1.26 46 Rh/Rh 11.60 0.61 30.02 1.57 48 Rh/Rh 10.60 0.59 27.32 1.51 Average 9.49 0.18 24.49 0.48
84 Table 3-5. MOSFET dosimeters, model TN1002 RD, sensitivity conversion factors at mammography energy range. TN-1002 RD Sensitivity Tube Potential (kVp) Target/Filter (mV kg C-1) x 104 (mV kg C-1) ( ) x 104 (mV R-1) (mV R-1) ( ) 24 Mo/Mo 4.09 0.96 10.56 2.48 28 Mo/Mo 4.77 1.05 12.32 2.72 34 Mo/Mo 5.12 0.80 13.22 2.08 30 Mo/Rh 5.80 0.98 14.95 2.54 34 Mo/Rh 6.20 0.20 16.00 0.51 38 Mo/Rh 6.27 0.24 16.19 0.61 36 Rh/Rh 6.61 0.65 17.07 1.67 42 Rh/Rh 7.10 0.32 18.33 0.82 48 Rh/Rh 7.42 0.34 19.15 0.87 Average 5.93 0.23 15.31 0.60
85 Table 3-6. Reproducibility of MOSFET dosimeters. Dosimeter Model Reproducibility TN-502 RDS 15.5% TN-1002 RD 27.7% TN-1002 RDM 28.8% TN-1002 RDI 31.8%
86 CHAPTER 4 CHARACTERIZATION OF FIBEROPTIC-COUPLED DETECTOR FOR DOSIMETRY IN CLINICAL MAMMOGRAPHY Introduction In vivo dosimetry has been the most dire ct way to monitor, document, improve quality control and provide safeguards ag ainst excessive exposures for patients undergoing an external beam radiation therapy.89 In vivo dosimetry in diagnostic examination can also bring the same benefits provided to radiation therapy. Documented doses can provide vital information for risk analysis by the physician, as recommended by the 2005 recommendations by the Inte rnational Commission on Radiological Protection (ICRP).86 Ideally, the risk analysis should us e a risk model that is derived from the cancer incidence of a larg e population that has been exposed to similar diagnostic xray procedures. In 2004, an article by Gonzalez and Darby came under scrutiny because the authors based their risk of cancer from diagnostic x-rays imaging on a linear-no threshold risk model that was based on the Japanese atomic bomb survivors.90-93 At the core of the discussions was the lack of su fficient data at the low-dose region to predict cancer risks accurately. Diagnostic x-ray examination ove rwhelmingly save lives, but the direct evidence for their potential detriment is lacki ng. Current data shows th at an individual is likely to receive nearly one x-ray image pe r year (962/1000 people) in the United States. 90 Worldwide, the number of annual diagnos tic x-ray images per 1000 people ranged from 565 to 1477 for the fourteen count ries reviewed by Gonzalez and Darby.90 The
87 recent concerns with patient doses in in terventional radiology, whole body CT and pediatric CT indicate that there is need to at least document dose received by patients during these procedures.94-96 The documentation of patient-sp ecific doses is the first step in providing a comprehensive database that would provide the direct evidence needed to evaluate the risk associated w ith diagnostic x-rays imaging. A likely population candidate for the st udy of patient doses is the screening mammography population. According to the US Census Bureau, there are approximately 69 million women over the age of 40 of which 62.6% or approximately 43 million women, have had a recent mammogram5 according to the 2000 Behavioral Risk Factor Surveillance System. Screening mammogr aphy is a large population of healthy individuals that have recurring exposures to ionizing radiation, representing an excellent opportunity to evaluate the risk of diagnosti c x-ray imaging. Using Land's estimates for case-control study size, a population of at l east 10 million would be needed to provide sufficient data to access the risk associated with low dose exposure( 10 mGy).97 Approximately one fourth of the curren t screening mammography population would be sufficient, using Land's estimation, to provide sufficient data to put to rest the risk associated with low dose exposures. It is not that the risk-benefit analysis for using mammography is in question because screen ing mammography has been responsible for reducing breast cancer fatalities. Small studi es have been conducted to document patient doses, but there is very little direct evidence in large populat ion studies as the risk of radiation-induced cancers.98,99 A contributing factor for not documenting patient doses in the United States is that unit output is re gulated, as opposed to patient doses being monitored. The Food and Drug Administra tion regulates design parameters and
88 equipment manufactures. However in mammography, the Mammography Quality Standards Act (MQSA) requires an annual check to verify that each facility performing mammography have its dedicated mammographi c units deliver less than to 3.0 mGy (300 mrad) dose for a standardized acrylic phantom with imbedded test objects, specified by the American College of Radiology.12,100 Another compelling reason to document patient doses is that new imaging modalities, such as breast computed tomography, digital mammography, and breast tomosynthesis, wh ich have all been designed to deliver reduced doses. However, patient-specific dos es from these technol ogies in a clinical setting are still needed to verify their design premise. Technology until recently has not been ab le to provide the tools to measure individual patient doses without interfering with th e diagnostic examination procedure. Today, there are several dosimeters that have b een or are about to be marketed that may provide an opportunity for in viv o individual dosimetry without a large impact on the use of diagnostic modalities. One such system is the metal oxide semiconductor field-effect transistors (MOSFET) dosimeter that based on the quantification of radiation damage to transistors.67,71,74 Another approach that is wellsuited for in vivo point dosimetry involves the use of optical fiber coupled sens ors. There are several different types of optical fiber dosimeters that differ primarily in the radiation sensitivity element that is used. These include, for example, scintilla ting plastic fiber, carbon-doped aluminium oxide, ruby spheres, and copper-doped qua rtz. The carbon-doped aluminum oxide (Al2O3:C) dosimeter is based on measuring the prompt radioluminescence (RL) and optical stimulated luminescence (OSL) resul ting from an ionizing radiation exposure to quantify the dose imparted.101,102 The sensitivity of the Al2O3:C device, 0.48 mm in
89 diameter and 2 mm length, is dependent on th e optical coupling and manufacturing of the Al2O3:C crystal.101 Aznar et al. showed in mammography that the Al2O3:C device had a 18% energy dependence between 23 and 35 kV p for a molybdenum (Mo) target and a Mo filtered mammography unit. In this paper, we will characterize a pr ototype copper-doped fused quartz, fiberoptic-coupled dosimeters (FOCD) is evaluate d for use in clinical mammography. We characterize the calibration factors, linearity, angular dependence, and reproducibility of the FOC dosimeters and readout module were developed and manufactured by Naval Research Laboratory, utilizing techni ques described by Bower and Hintenlang.67 The proprietary distribution of this technol ogy is available through Global Dosimetry Solutions Incorporated6). The theory of operation of the FOC dosimeter has been previously discussed in several papers and will only be summarized in this paper.77,103 A variety of materials produce luminescence immediately or when stimulated with heat (thermoluminescence), or stimulated w ith visible light (OSL ) following exposure to ionizing radiation. The fiber-optic-coupled (FOC) detector is a copper doped fused quartz, SiO2:Cu, dosimeter which has a RL signal th at is large enough to permit use as a dosimeter. After being expos ed to ionizing radiation, Cu+1 ions are raised to an excited state, which de-excite by a 150 Âµs ec RL. A desirable characteristic of this material is that it is fully compatible with fiber optic transmi ssion of the RL signal. The detector can be fused directly to the optic fiber w ithout a visible boundary in radiographs. 6 Global Solutions Incorporation, Irvine, CA 92614
90 Materials and Methods Fiber-Optic-Coupled Dosimeters (FOCD) The current study characterized fiberoptic-coupled dosimeters spanning four copper-doped quartz detector sizes, Table 4-1. The dosimeters are described by the detector size to facilitate discussion. Each fiber optical dos imeter was enclosed in an opaque protective coating that prevents light penetration. The dosimeters were used in conjunction with a photomultiplier module (Hamamatsu model HC135-11) 7), and power supply module (Acopian model 5EB100)8) as shown in figure 4-1. Acquisition data were transferred via a RS-232 computer cable to a laptop computer. Th e dosimeter acquisition data were evaluated using a LabVIEW) program. The computer program module is illustrated in figure 4-2, which provided a di splay of high voltage status, communication port selection, integrat ion time (10 msec 1 sec), total counts, total duration, background counts, and average count rate. Total counts we re an integration of the peak counts after subtracting the integrated background count s. FOC dosimeters background count rate has contributions from several sources, incl uding PMT dark counts that represent mostly thermal emission of electrons from the photoc athode, room light that penetrates through small openings in the fiberoptic connector s and residual traps released from the Cu+1doped quartz due to previous radiation expos ures. The background count levels are very constant over periods that are long compar ed to the diagnostic procedure and are accounted for by the software by recording a baseline prior to accumulating exposure data. 7 Hamamatsu Corporation, Bridgewater, NJ 08807 8 Acopian, Easton, PA 18044 9 National Instruments Corporation, Austin, TX 78759-3504
91 X-Ray Field A General Electric Senographe-DMR10) mammography x-ray unit provided the mammography x-ray field used throughout the experiments. The Senographe-DMR uses a high-frequency generator and an x-ray tube with a dual bi-metal track target of molybdenum (Mo) and rhodium (Rh). The xray output was filtered using Rh or Mo filters. Tube potentials used for the char acterization were incremented in two kVp increments over 25-31 kVp in each target-f ilter combination: Mo/Mo, Mo/Rh and Rh/Rh. The tube current-time product stations used were within the typical clinical practice. The unit had a fixed source -to-image receptor distance of 660 mm. Exposure Measurement The ionization chamber used for referen ce exposures and dosimeter calibration was a Victoreen NEROÂ®11) mAx model 8000 radiation m onitor with a model 6000-529 mammography ion chamber. The mammography ionization chamber has a metal-coated polycarbonate window with a thickness of 9.5 mg cm-2 and an active volume of 3.3 cm3. Dosimeter Reproducibility Reproducibility was evaluated by exposi ng the 1.9, 1.6 and 1.1 mm dosimeters to 30 kVp, 100 mAs (3.72 x 10-4 C kg-1 (1.44 R)), Mo/Mo for 10 total exposures. The 4.0 mm dosimeter was exposed to 28 kVp, 100 mA s (1.21R) for 10 total exposures. The dosimeters were exposed as shown in figure 4-4. 10 General Electric Medical Systems, 3135 Easton Turnpike, Fairfield, CT 06431 11 Inovision Radiation Measurements, 6045 Cochran Road, Cleveland, OH 44139-3303
92 Dosimeter Angular Response The dosimeter axial-angular evaluation was performed by rotating the dosimeter about its cable axis free-in-air, perpendi cular to the x-ray beam, figure 4-3a. The dosimeter normal-to-axis-angular evaluati on was performed by rotating the dosimeter parallel to its cable axis, fi gure 4-3b. The axial and normalto-axis evaluations of the 1.1, 1.6, 1.9 and 4.0 mm dosimeters were perfor med using a Mo/Mo, 28 kVp, 71 mAs to accumulate an exposure of approximately 1R. The 4.0 mm dosimeter was exposed to 26 kVp, 28 mAs, to accumulate an exposure of approximately 340 mR, to avoid overwhelming the PMT circuitry. The dosimeters were rotated 360 axial and normal-toaxis as indicated in figure s 4-5 through figure 4-16. The angular response was evaluated to determine the working angular range of the dosimeters for clinical applications. In the present clinical application, the dosimeters are exposed in a fixed axial orientation. However, investigators might wish to use dos imeters in other axia l or normal-to-axis configurations. The reported values were an average of three e xposures taken at the indicated angular increment. Dosimeter Sensitivity The dosimeter sensitivities were evaluate d over the x-ray energies of 25 to 31 kVp to evaluate the energy response across th e mammography energy range. The dosimeters were fixed to the underside of the compre ssion paddle approximately one centimeter from the face of the ACR phantom as illust rated in figure 4-4. The target-filter combinations evaluated were Mo/Mo, Mo /Rh, and Rh/Rh. Three (700mR) exposures were made at each tube-potential energy set ting. The total exposure was kept constant for different tube potentials by adjusting the tube current-time product (mAs). The reported values were an average of three exposures.
93 The primary use of the dosimeter during clinical imaging is to determine the freein-air exposure to determine the patient glandul ar dose for an individual. Therefore, the dosimeter position was selected to ensure that the dosimeter would not interfere with the anatomical features being imaged. Dosimeter Linearity Dosimeter linearity was evaluated at t ube potentials of 25, 27, 29 and 31 kVp at tube current-time product stations of 40, 90 a nd 140 mAs within the clinical range. The target-filter combinations evaluated were Mo/Mo, Mo/Rh, and Rh/Rh. The dosimeters were exposed as previously described and as illustrated in figu re 4-4. The reported values were the average of three exposures taken at each tube potential (kVp) and tube current-time (mAs) station. Results and Discussion Dosimeter Angular Response The axial-angular responses are shown in figure 4-5 through figure 4-8 and were normalized to the dosimeter response at zero degrees. The axial-angu lar responses for all dosimeter models show nearly uniform re sponse without any marked decrease in sensitivity. However, the normal-axis-angular response showed a marked decrease in sensitivity about 0Â° and 180Â°, as shown in fi gure 4-9 through figure 4-16. The decrease in angular response appears to be a result of the photon attenuation in the dosimeter and cross-sectional volume to the low energy b eam, the cylindrical construction of the dosimeter and fiber optic cable fusion to the dosimeter. In the proposed clinical application, mammography dosimeters are utili zed free-in-air and axial to the x-ray beam, thereby eliminating the variations shown in the normal-to-axis response.
94 Dosimeter Sensitivity The sensitivities of the dosimeters are shown in Table 4-2 through Table 4-5, and figure 4-17 through figure 4-20 showed energy de pendence. The sensitivity as a function of tube potential increased in Mo/Mo and Rh /Rh exposed dosimeters by 4 to 7% and by 2.1 to 2.2 % for dosimeter exposed to Mo/Rh. Overall, the order of sensitivity from highest to lowest was 4.0, 1.9, 1.6, and 1.1 mm dosimeters. The dosimeter model average sensitivity as a function of detect or volume demonstrated a linear response (8 810 70 . 4 10 45 . 1 Volume y Sensitivit , and R2 = 95 %) shown in figure 4-21. Dosimeter Linearity All dosimeter models showed a linear response (R2 0.997) to exposure for all target-filter combinations, tube potentials and tube cu rrent-time product stations evaluated, as shown in figure 4-22 through figur e 4-25. The linear response is consistent with previous investigations in radiotherapy.77,103 Dosimeter Reproducibility The dosimeter sensitivity va ried according to the activ e length of the sensor material, but all were within 2%, as s hown in Table 4-6. The reproducibility was consistent with previous investigat ions in the radiot herapy energy range.104 Conclusions The overall performance of the FOC dosim eters evaluated in this study indicated that they provide a reasonable response at exposures and energies characteristic of clinical mammography. The FOC dosimeter had a very uniform axial-angular response. The normal-to-axis angular response showed a marked decrease near 0Â° and 180Â°, corresponding to the fiber-optic coupling region and the reduced cross-sectional interaction area. The mammography applica tion, however, would utilize the dosimeters
95 in an axial position; therefore the normal-to -axis response should not be of concern for this, or for most dosimetry applications in planar imaging. The FOC dosimeter demonstrated reproducibility w ithin 2%, and was linear for al l target-filter combinations. The dosimeter sensitivity as a function of tube potential ha d an average increase of 4.72 Â± 2.04% for dosimeter models and target-fil ter combinations tested. The energy dependence at each target-filter combination affects the final dosimetry measurement and must be taken into account for accuracy. The 4.0 mm dosimeter was found to be the best suited for mammography applications, because it provides the largest volume of interaction sites (elect ron and hole traps).
96 Cudoped quartz+1 AC+ Power supply PMT X-Ray Source Figure 4-1 The schematic illustrate s the fiber-optic-coupled dosimeter system for dosimeter evaluations.
97 . Figure 4-2 The fiber-optic-coupled dosimet er computer interface software display.
98 45Â° 0Â° 180Â° 270Â° 135Â° 90Â° 270Â° 180Â° 0Â° 45Â° 135Â° 90 Â° (B) Normal-to-axis (A) Axial Figure 4-3. (a) A schematic illustrates the ax ial-angular evaluation geometry. (b) A sc hematic illustrates the normal-to-axis e valuation geometry.
99 ACR phantom ionization chamber FOC dosimeter Image receptor Compression Paddle RCC FOC dosimeterA B Figure 4-4. (A) The schematic illustrates the experimental setup used for FOC dosimeter evaluations. (B) The schematic illustra tes the clinical setup of the FOCD. Th e FOCD would be fixed to the underside of the compression padd le without interfering with the diagnostic image.
100 Axial Rotation(degrees) 0 . 0 2 2 . 5 4 5 . 0 6 7 . 5 9 0 . 0 1 1 2 . 5 1 3 5 . 0 1 5 7 . 5 1 8 0 . 0 2 0 2 . 5 2 2 5 . 0 2 4 7 . 5 2 7 0 . 0 2 9 2 . 5 3 1 5 . 0 3 3 7 . 5 3 6 0 . 0Dosimeter Response Ratio 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00 1.01 1.02 1.03 1.04 1.05 Figure 4-5. The graph shows the axial-angu lar response of the 4.0 mm dosimeter model using a 26 kVp, Mo/Mo spectrum at 28 mAs. The angular-response ratio was normalized to zero-degree a ngular response (counts counts-1 0 Ã») of the dosimeter.
101 Axial Rotation (degrees) 0 . 0 2 2 . 5 4 5 . 0 6 7 . 5 9 0 . 0 1 1 2 . 5 1 3 5 . 0 1 5 7 . 5 1 8 0 . 0 2 0 2 . 5 2 2 5 . 0 2 4 7 . 5 2 7 0 . 0 2 9 2 . 5 3 1 5 . 0 3 3 7 . 5 3 6 0 . 0Dosimeter Response Ratio 0.95 0.96 0.97 0.98 0.99 1.00 1.01 1.02 1.03 1.04 1.05 Figure 4-6. The graph shows the axial-angu lar response of the 1.9 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angular-response ratio was normalized to zero-degree a ngular response (counts counts-1 0 Ã») of the dosimeter.
102 Axial Rotation(degrees) 0 . 0 2 2 . 5 4 5 . 0 6 7 . 5 9 0 . 0 1 1 2 . 5 1 3 5 . 0 1 5 7 . 5 1 8 0 . 0 2 0 2 . 5 2 2 5 . 0 2 4 7 . 5 2 7 0 . 0 2 9 2 . 5 3 1 5 . 0 3 3 7 . 5 3 6 0 . 0Dosimeter Response Ratio 0.95 0.96 0.97 0.98 0.99 1.00 1.01 1.02 1.03 1.04 1.05 Figure 4-7. The graph shows the axial-angu lar response of the 1.6 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angular-response ratio was normalized to zero-degree a ngular response (counts counts-1 0 Ã») of the dosimeter.
103 Axial Rotation(degrees) 0 . 0 2 2 . 5 4 5 . 0 6 7 . 5 9 0 . 0 1 1 2 . 5 1 3 5 . 0 1 5 7 . 5 1 8 0 . 0 2 0 2 . 5 2 2 5 . 0 2 4 7 . 5 2 7 0 . 0 2 9 2 . 5 3 1 5 . 0 3 3 7 . 5 3 6 0 . 0Dosimeter Response Ratio 0.95 0.96 0.97 0.98 0.99 1.00 1.01 1.02 1.03 1.04 1.05 Figure 4-8. The graph shows the axial-angu lar response of the 1.1 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angular-response ratio was normalized to zero-degree a ngular response (counts counts-1 0 Ã») of the dosimeter.
104 0.00.20.40.60.81.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0 30 60 90 120 150 180 210 240 270 300 330 Figure 4-9. The polar-plot graph shows the normal-to-axis angular response of the 4.0 mm dosimeter model using a 26 kVp, Mo/Mo spectrum at 28 mAs. The angular-response ratio was normalized to zero-degree angular response (counts counts-1 0 Ã») of the dosimeter.
105 0.00.20.40.60.81.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0 30 60 90 120 150 180 210 240 270 300 330 Figure 4-10. The polar-plot graph shows th e normal-to-axis angular response of the 1.9 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angular-response ratio was normalized to zero-degree angular response (counts counts-1 0 Ã») of the dosimeter.
106 0.00.20.40.60.81.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0 30 60 90 120 150 180 210 240 270 300 330 Figure 4-11. The polar-plot graph shows th e normal-to-axis angular response of the 1.6 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angular-response ratio was normalized to zero-degree angular response (counts counts-1 0 Ã») of the dosimeter.
107 0.00.20.40.60.81.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0 30 60 90 120 150 180 210 240 270 300 330 Figure 4-12. The polar-plot graph shows th e normal-to-axis angular response of the 1.1 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angular-response ratio was normalized to zero-degree angular response (counts counts-1 0 Ã») of the dosimeter.
108 Normal-Axial Rotation (degrees) 0 . 0 2 2 . 5 4 5 . 0 6 7 . 5 9 0 . 0 1 1 2 . 5 1 3 5 . 0 1 5 7 . 5 1 8 0 . 0 2 0 2 . 5 2 2 5 . 0 2 4 7 . 5 2 7 0 . 0 2 9 2 . 5 3 1 5 . 0 3 3 7 . 5 3 6 0 . 0Dosimeter Response Ratio 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Figure 4-13. The graph shows the normal -to-axis angular response of the 4.0 mm dosimeter model using a 26 kVp, Mo/Mo spectrum at 28 mAs. The angularresponse ratio was normalized to zero -degree angular response (counts counts1 0 Ã») of the dosimeter.
109 Normal-Axial Rotation (degrees) 0 . 0 2 2 . 5 4 5 . 0 6 7 . 5 9 0 . 0 1 1 2 . 5 1 3 5 . 0 1 5 7 . 5 1 8 0 . 0 2 0 2 . 5 2 2 5 . 0 2 4 7 . 5 2 7 0 . 0 2 9 2 . 5 3 1 5 . 0 3 3 7 . 5 3 6 0 . 0Dosimeter Response Ratio 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 Figure 4-14. The graph shows the normal -to-axis angular response of the 1.9 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angularresponse ratio was normalized to zero -degree angular response (counts counts1 0 Ã») of the dosimeter.
110 Dosimeter Rotation (degrees) 0 . 0 2 2 . 5 4 5 . 0 6 7 . 5 9 0 . 0 1 1 2 . 5 1 3 5 . 0 1 5 7 . 5 1 8 0 . 0 2 0 2 . 5 2 2 5 . 0 2 4 7 . 5 2 7 0 . 0 2 9 2 . 5 3 1 5 . 0 3 3 7 . 5 3 6 0 . 0Dosimeter Response Ratio 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Figure 4-15. The graph shows the normal -to-axis angular response of the 1.6 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angularresponse ratio was normalized to zero -degree angular response (counts counts1 0 Ã») of the dosimeter.
111 Normal-Axial Rotation(degrees) 0 . 0 2 2 . 5 4 5 . 0 6 7 . 5 9 0 . 0 1 1 2 . 5 1 3 5 . 0 1 5 7 . 5 1 8 0 . 0 2 0 2 . 5 2 2 5 . 0 2 4 7 . 5 2 7 0 . 0 2 9 2 . 5 3 1 5 . 0 3 3 7 . 5 3 6 0 . 0Dosimeter Response Ratio 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 Figure 4-16. The graph shows the normal -to-axis angular response of the 1.1 mm dosimeter model using a 28 kVp, Mo/Mo spectrum at 71 mAs. The angularresponse ratio was normalized to zero -degree angular response (counts counts1 0 Ã») of the dosimeter.
112 Tube Potential (kVp) 242526272829303132Sensitivity(108 counts kg C-1) 2 4 6 8 10 12 14 Mo/Mo Mo/Rh Rh/Rh Figure 4-17. The graph shows the sensitivity response of the 4.0 mm dosimeter model.
113 Tube Potential(kVp) 242526272829303132Sensitivity (counts kg C-1) 2 4 6 8 10 12 14 Mo/Mo Mo/Rh Rh/Rh Figure 4-18 The graph shows the sensitivity response of the 1.9 mm dosimeter model.
114 Tube Potential(kVp) 242526272829303132Sennsitivity (counts kg C-1) 2 4 6 8 10 12 14 Mo/Mo Mo/Rh Rh/Rh Figure 4-19. The graph shows the sensitivity response of the 1.6 mm dosimeter model.
115 Tube Potential (kVp) 242526272829303132Sensitivity (108counts kg C-1) 2 4 6 8 10 12 14 Mo/Mo Mo/Rh Rh/Rh Figure 4-20. The graph shows the sensitivity response of the 1.1 mm dosimeter model .
116 Dosimeter Volume (mm3) 123456Sensitivity (counts kg C-1) 5.0e+8 6.0e+8 7.0e+8 8.0e+8 9.0e+8 1.0e+9 1.1e+9 1.2e+9 1.3e+9 Mean Sensitivity Figure 4-21. The graph shows the mean sens itivity response plotte d as a function of dosimeter volume.
117 Tube Current Time Product (mAs) 20406080100120140160Dosimeter Response (counts) 0 1e+5 2e+5 3e+5 4e+5 5e+5 6e+5 7e+5 Mo/Mo, 25 kVp Mo/Mo, 27 kVp Mo/Mo, 29 kVp Mo/Rh, 25 kVp Mo/Rh, 27 kVp Mo/Rh, 29 kVp Mo/Rh, 31 kVp Rh/Rh, 25 kVp Rh/Rh, 27 kVp Rh/Rh, 29 kVp Rh/Rh, 31 kVp Figure 4-22. The graph shows the linearit y of the dosimeter response of the 4.0 mm dosimeter model.
118 Tube Current Time Product (mAs) 20406080100120140160Dosimeter Response (counts) 0 1e+5 2e+5 3e+5 4e+5 5e+5 6e+5 Mo/Mo, 25 kVp Mo/Mo, 27 kVp Mo/Mo, 29 kVp Mo/Rh, 25 kVp Mo/Rh, 27 kVp Mo/Rh, 29 kVp Rh/Rh, 27 kVp Rh/Rh, 29 kVp Rh/Rh, 31 kVp Figure 4-23. The graph shows the linearit y of the dosimeter response of the 1.9 mm dosimeter model .
119 Tube Current Time Product (mAs) 20406080100120140160Dosimeter Response (Counts) 0 1e+5 2e+5 3e+5 4e+5 5e+5 Mo/Mo, 25 kVp Mo/Mo, 27 kVp Mo/Mo, 29 kVp Mo/Rh, 25 kVp Mo/Rh, 27 kVp Mo/Rh, 29 kVp Mo/Rh, 31 kVp Rh/Rh, 25 kVp Rh/Rh, 27 kVp Rh/Rh, 29 kVp Rh/Rh, 31 kVp Figure 4-24. The graph shows the linearit y of the dosimeter response of the 1.6 mm dosimeter model.
120 Tube Current Time Product (mAs) 20406080100120140160Dosimeter Response (counts) 0.0 5.0e+4 1.0e+5 1.5e+5 2.0e+5 2.5e+5 3.0e+5 3.5e+5 Mo/Mo, 25 kVp Mo/Mo, 27 kVp Mo/Mo, 29 kVp Mo/Rh, 25 kVp Mo/Rh, 27 kVp Mo/Rh, 29 kVp Mo/Rh, 31 kVp Rh/Rh, 25 kVp Rh/Rh, 27 kVp Rh/Rh, 29 kVp Rh/Rh, 31 kVp Figure 4-25. The graph shows the linearit y of the dosimeter response of the 1.1 mm dosimeter model.
121 Table 4-1. Fiber-optic-coupled dosimeter sizes. Dosimeter Length (mm) Â± 0.2 Dosimeter Diameter (Âµm) Dosimeterfiber optic fiber length (m) 4.0 400 1.0 1.9 400 1.0 1.6 400 1.0 1.1 400 1.0
122Table 4-2. FOC dosimeter, 1.1 mm model, sensi tivity conversion factors over the mammography energy range. Average values were used to co mpare sensitivity response of dosimeter models. Sensitivity Tube potential (kVp) Target/Filter (counts kg C-1) x108 (counts kg C-1) ( ) x108 (counts mR-1) (counts mR-1) ( ) 25 Mo/Mo 5.41 2.67 x 10-3 139.62 0.69 27 Mo/Mo 5.42 2.18 x 10-3 139.82 5.62 29 Mo/Mo 5.51 2.40 x 10-3 142.25 6.20 31 Mo/Mo 5.77 5.74 x 10-3 148.78 1.48 25 Mo/Rh 6.07 3.66 x105 156.65 0.09 27 Mo/Rh 6.10 3.24 x 10-3 157.28 0.84 29 Mo/Rh 6.16 6.09 x 10-3 158.87 1.57 31 Mo/Rh 6.20 7.01 x 10-3 159.92 1.81 25 Rh/Rh 6.33 4.15 x 10-3 163.26 1.07 27 Rh/Rh 6.50 7.38 x 10-3 167.70 1.90 29 Rh/Rh 6.65 1.85 x 10-3 171.49 0.48 31 Rh/Rh 6.77 8.87 x 10-3 174.69 2.29 Average 6.07 7.50 x 10-3 156.69 0.79
123Table 4-3. FOC dosimeter, 4.0 mm model, sens itivity conversion factors at mammography energy range. The PMT failed to provide a response at 31-kVp Mo/Mo spectrum. Average values were used to compare sensit ivity response of dosimeter models. Sensitivity Tube potential (kVp) Target/Filter (counts kg C-1) x108 (counts kg C-1) ( ) x108 (counts mR-1) (counts mR-1) ( ) 25 Mo/Mo 10.41 0.06 268.66 1.60 27 Mo/Mo 10.68 0.05 275.53 1.37 29 Mo/Mo 10.80 0.001 278.62 3.58 31 Mo/Mo NM 25 Mo/Rh 11.58 0.02 298.94 0.60 27 Mo/Rh 11.77 0.001 303.76 3.97 29 Mo/Rh 11.96 0.02 308.48 4.77 31 Mo/Rh 11.85 0.05 305.65 1.33 25 Rh/Rh 12.27 0.08 316.68 2.13 27 Rh/Rh 12.55 0.08 323.80 2.01 29 Rh/Rh 12.87 0.08 332.00 2.06 31 Rh/Rh 12.97 0.09 334.50 2.45 Average 11.79 0.03 304.24 0.80
124 Table 4-4. FOC 1.9 mm dosimeter model sensitivity conversion factors at mammography energy range. Average values were used to compare sensitivity response of dosimeter models. "Red" Sensitivity Tube potential (kVp) Target/Filter (counts kg C-1) x108 (counts kg C-1) ( ) x108 (counts mR-1) (counts mR-1) ( ) 25 Mo/Mo 7.75 0.04 199.88 0.92 27 Mo/Mo 8.00 0.01 206.27 0.29 29 Mo/Mo 8.06 0.08 207.96 1.98 31 Mo/Mo 8.15 0.04 210.37 1.09 25 Mo/Rh 8.68 0.01 224.06 0.35 27 Mo/Rh 8.77 0.09 226.38 0.22 29 Mo/Rh 8.95 0.05 230.96 1.24 31 Mo/Rh 8.90 0.001 229.72 2.73 25 Rh/Rh 9.18 0.10 236.79 2.56 27 Rh/Rh 9.37 0.04 241.73 1.03 29 Rh/Rh 9.52 0.04 245.60 1.15 31 Rh/Rh 9.72 0.09 250.68 2.25 Average 8.75 0.03 225.87 0.45
125 Table 4-5. FOC dosimeter, 1.6 mm model, sensitivity c onversion factors at mammogr aphy energy range. Average values were used to compare sensit ivity response of dosimeter models. Sensitivity Tube potential (kVp) Target/Filter (counts kg C-1) x108 (counts kg C-1) ( ) x108 (counts mR-1) (counts mR-1) ( ) 25 Mo/Mo 6.94 0.05 178.95 1.30 27 Mo/Mo 7.08 0.10 182.56 0.25 29 Mo/Mo 7.31 0.03 188.72 0.86 31 Mo/Mo 7.45 0.03 192.31 0.93 25 Mo/Rh 7.87 0.05 203.07 1.19 27 Mo/Rh 7.94 0.07 204.85 1.81 29 Mo/Rh 8.06 0.10 207.98 2.46 31 Mo/Rh 8.03 0.05 207.05 1.20 25 Rh/Rh 8.22 0.03 212.04 0.68 27 Rh/Rh 8.39 0.03 216.43 0.83 29 Rh/Rh 8.63 0.07 222.76 1.69 31 Rh/Rh 8.73 0.05 225.32 1.26 Average 7.89 0.02 203.50 0.38
126 Table 4-6. Reproducibility of FOC dosimeters. Dosimeter ModelReproducibility 1.9 mm 0.99% 1.1 mm 0.98% 1.6 mm 0.34% 4.0 mm 1.98%
127 CHAPTER 5 A BREAST TISSUE EQUIVALENT PHANTOM SERIES FOR USE IN CLINICAL MAMMOGRAPHY Introduction Since their introduction in 1906 by KienbÃ¶ch, pha ntoms have played a vital role in diagnostic radiology, radiation dosimetr y, calibration and image assessment.64 In mammography, phantoms have been an integral part of the quality assurance, image quality and radiation dosimetry evalua tions required by the Mammography Quality Standards Act of 1992 (MQSA). The most commonly used Food and Drug Administration (FDA) approved phantom is the phantom described by the American College of Radiology (ACR).12 The phantom design is a modification of the Wisconsin Mammographic Phantom (Random Phantom) developed at the University of Wisconsin.105 Today's ACR phantom is used to evaluate quality assurance, image quality and to provide a standardized measure of the average glandular dose (AGD). The AGD measured using the ACR phantom is a benchm ark dose to monitor th e x-ray output of a mammographic unit for inter-comp arison among mammography units and mammography facilities to verify regula tory compliance with the dose limit. The AGD measured using the ACR phanto m was developed to represent the average glandular dose received by an averag e compressed breast ( 4.5 cm) composed of 50% fibroglandular (glandular) and 50% adipose tissue when it is imaged with the automatic exposure control system of a mammography unit. The ACR phantom is composed of a 3.4 cm base of acrylic, a 0.7-cm thick paraffin insert (surrounded by
128 acrylic on the sides) containing image quality test objects and a cover of acrylic (0.4 cm), as shown in figure 5-1. The ACR phantom is not representative of the screening mammography population as s hown by several studies.29,33,50,58,59,81 Several worldwide studies have shown that the average br east ranges from 4.5 to 5.2 cm compressed thickness with a glandularity ra nge of 26-41% glandular tissue.29,33,50,58,59 The lateral dimensions of the average breast have not b een extensively studie d, but a small study by Fife showed that the average breast was a pproximately 8.1 Â± 2.1 cm in the dimension of the posterior-nipple line (PNL), 18 Â± 2.4 cm in width and 5.2 Â± 1.1 cm in thickness in a craniocaudal (CC) view mammogram.63 Breast PNL is the perpendicular distance from the chest wall to the nipple and the breast width is the distance from the axilla medially to the sternum in a CC view mammogram. Given the complex nature of breast tissu e one must also consider the elemental composition and the effects of aging on breast tissues. The effect of age on glandular content was demonstrated by Prechtel when he showed that the adipose content increases and the glandular content decreases as an individual woman ages.106 The elemental composition of breast tissue has also been stud ied by several authors, the results of which were compiled into the International Commission on Radiol ogical Units and Measurements (ICRU) Publica tion 46, shown in table 2-4.28,107,108 Population studies of tissue compositions ha ve resulted in the formulation of tissueequivalent phantoms that approximate the elemental composition, mass density, and radiological properties of the target tissue to provide esti mates of patient dose. In mammography, the radiological properties of breast tissue can best be approximated by the percent weight content of the glandular and adipose compone nt tissues of the breast.
129 In 1977, White developed a widely-used epoxy resin-based tissue substitute that mimicked 50% glandular breast tissue ( BR 12) in mass density and radiological interactions of the 0.01 to 100 MeV energy range.51 White stated that he decided to use a 50% glandular tissue content because "Due to the lack of reliable data on the chemical composition of this tissue, an 'average breas t tissue' of 50% fat and 50% water by weight was accepted".51 BR 12 is often used to determine patient doses; however, it does not reflect the distribution charac teristics of the mammography sc reening patient population. Another popular tissue-equiva lent phantom is the CIRS12 tissue-equivalent phantom (model 011A) for mammography, for which the manufacturing method and phantom design was based on the work of Fatouros and Skubic.109-111 The CIRS phantom is realistically shaped (12.5 cm length and 18.5 cm width), composed of Hammerstein's composition of 50% glandular tissue, with a variety of image quality objects and a varying glandularity step wedge.28,112 Here too, the CIRS phantom is lacking the target population char acteristics in size and glandula rity to be used effectively for patient dose calculations. On the othe r hand, the breast tissue-equivalent series (BRTES) developed at the University of Fl orida by Argo et al. was developed to mimic glandular tissue content from 20 to 70% and was intended to be fabricated into breast sizes that mimic a realistic size di stribution of the target population.113 In this study, a BRTES phantom series was developed to represen t, not only breast tissue elemental composition, mass density, and ra diological interactio ns coefficients, but also the scattering characteristics related to the compressed breast thickness, PNL, and width. In addition, we modified the epoxy-resin components and manufacturing 12 CIRS, Norfolk, VA
130 techniques of Argo et al. wa s modified in order to improve the radiographic homogeneity of the phantoms. Materials and Methods Phantom Manufacturing As in the BRTES phantom, breast-tissue equivalent modified series (BRTESMOD) homogeneous phantoms were developed using an epoxy-resin matrix, constructed to represent a range of gl andularities and compressed br east thicknesses described by Geise and Palchevsky, figure 5-2. The com ponents added to the epoxy resin mixture were selected to modify the properties of the phantom to ma tch those of breast tissues. The epoxy-resin hardener matrix will unde rgo a polymerization reaction during the curing process resulting in a hardened plastic phantom. The BRTES-MOD curedphantom elemental composition was assumed to be equivalent to the epoxy-resin matrix components. The components are nonvolatile, and form a highly visc ous liquid which has the same mass before and after the curing process.113 Independent elem ental analysis has not been identified because of the physical properties and the hi gh hydrogen content of the epoxy-resin components. Selection crit eria for the epoxy-resin matrix components were based upon their elemental compositi on, commercial availability, long-term stability and ease of use. The final epoxy-resin matrix mixture cl osely matched the elemental composition, the mass attenuation (Âµ/ ), mass-energy absorption (Âµen/ ) coefficients, and the mass density ( ) of breast tissue over the mammography energy range (10 -50 keV). XCOM Version 3.1, a software package available fr om the National Institute of Standards and Technology (NIST), was used to match the to tal mass attenuation coefficients between BRTES-MOD and breast tissues with com positions described in ICRU Report 44
131 (ICRU-44) and mean values in ICRU Repor t 46 (ICRU-46) for glandular and adipose tissues.84,107,114 The percent weights of the component materials were adjusted to minimize the differences (within 3%) in elemental composition, Âµ/ and between BRTES-MOD and the ICRU-46 using a referen ce energy of 25 keV. The selection of 25 keV was based on the primary photon interacti ons of photoelectric effect, incoherent (Compton) scattering and cohere nt (Raleigh) scattering. The dominance of an interaction is directly dependent on the photon energy, as demonstrated in figure 5-3. The dominant interaction between 0-25 keV is the phot oelectric effect while Compton scattering dominates above 25 keV. Raleigh scattering is never dominant and contributes to less than 12% of interactions for the entire energy range examined. Selecting the 25 keV reference energy ensured that the coefficien ts would not be biased toward a dominant interaction without having to evalua te the entire energy spectrum. The Bragg rule or "mixture rule"was used to proportion the mass-fractions ( w ) of the epoxy-resin matrix elemental compone nts determine the resulting total massattenuation coefficient (Âµ -1) , equation 5-1, or total massenergy attenuation coefficient (Âµen -1), equation 5-2.115,116 i i B B A A mixw w w .... (5-1) i en B B en A A en enw w w .... (5-2) The percent glandularity of referenced tissues were determined by the weighted fractions of 100% adipose and 100% glandular elemental compositions listed in ICRU46. The BRTES-MOD phantom series tissues we re the mass-weighted fractions of the elemental composition of the epoxy-resin matr ix components. The target mass-density
132 was estimated from the densities listed for 100% adipose and 100% glandular elemental compositions in ICRU Report 46. The commercial sources for phantom com ponents are listed in table 5-1.The component weight fraction for each glandularity is listed in table 5-2. The BRTES components used by Argo et al . were modified by replaci ng the original microspheres used with thermoplastic microspheres.113 This change in microspheres was made to improve the radiographic homogeneity of th e mammographic image of the phantom. The epoxy-resin matrix component recipe was deve loped to minimize the differences between the mass attenuation coefficient of the BRTES-MOD and tissue compositions as delineated in ICRU-46. Once the percent-we ight component mixture was determined, table 5-2, a series of five te st samples (100 ml) was made with variations in the percentweight of the microspheres to evaluate th e cured epoxy-resin matrix mass density. The variation in microspheres conten t did not adversely affect the radiological characteristics of the final cured material but did affect the final mass density. The percent weight of the microspheres corresponding to th e desired percentage was se lected, and the final epoxyresin matrix mixture was ready for manufacturing. The epoxy-resin matrix components for each selected glandularity of BRTESMOD composition were accurately weighed (Â± 0.01 gm) and combined in the following order: Araldite (epoxy-resin), polyethelene powder, microspheres, MgO, and Jeffamine (hardener). The components were thoroughly hand-mixed until uniformly distributed and smooth (approximately 5-10 min) to prevent excessive introducti on of air into the mixture. Once the components were hand-mixe d, they were transferred to an evacuated container and mixed continually for 30 min. The vacuum in the chamber caused the
133 epoxy-resin matrix mixture to rise as air es caped the matrix; however, the rotating mixing blade assisted the escape of trapped gas and kept the mixture contained. The BRTES' manufacturing method developed by Argo et al. did not utilize a vacuum chamber as part of their protocol.113 The vacuum mixing produced a pr oduct that had fewer air bubbles in the cured phantom, which had also been reported by White.51 After vacuum mixing, the epoxy-resin matrix was poured into Teflon molds that either D-shaped for the compressed breast phantoms or rectangular mo lds for the step phantom, to a depth of approximately 2.5 cm and then allowed to cure at room temperature for 48 hours. Once cured, five small samples (approximately 12 cm3) were taken from the BRTES-MOD phantom to evaluate the cured mass densit y. Mass density was determined using the Archimedes Principle with isopropyl alcohol ( = 0.8011 Â± 0.0053 g cm-3, 23.43Â° Â± 1.78 Â°C ) as the liquid medium.116 After the density was verified to be within Â±3% of the target density, the phantom sections were ready to be machined into their final dimensions. The BRTES-MOD phantoms were machined into semielliptical cylinders with the minor axis half-length (Posterior-Nipple Line (PNL)) of 8 cm and the major axis length (width) of 18 cm (width). Phantom dimensi ons were selected to correlate to a nominal medium-size breast as delineated in the FDA Handbook of Glandular Tissue Doses in Mammography, which also correlates well with the average size breast as determined by Fife, et al.52,63 The range in glandularity was ba sed on polynomial fit to demographic data reported by Geise and Palchevsky s hown in figure 5-2 by breast thickness.29 The final BRTES-MOD phantom thicknesses ranged from 1 to 8 cm in approximately 1 cm sections for all glandularities, shown in fi gure 5-4. In addition to the D-shaped phantoms, a step wedge was made that contai ned 10 individual sec tions, each of which
134 had steps ranging from 1-8 cm with glandular ities ranging from 100% glandular to 100% adipose, as shown in figure 5-5. The D-shaped phantoms were used to meas ure the patient average glandular dose, whereas the step wedge phantom was used as a calibration tool to evaluate individual patient glandularity. Phantom homogeneity was evaluated by mammographic examination visual inspection. A General Electric Senographe-DMR13) mammography unit was used to make the mammography image. The Senogra phe-DMR uses a high-frequency generator and an x-ray tube, with a dual bi-metal track target of molybdenum (Mo) and rhodium (Rh). The x-ray spectrum used was filtered using Rh or Mo filters. Radiograph images were made only after processor quality cont rol was completed and was found to be within control limits as delineated in MQ SA regulations and ACR guidelines.12,100 The Kodak14) Min-R 2000 screen-film combination was used for imaging in the homogeneity evaluation; these images were processed in a Kodak XOMAT 5000 RA film processor. BRTES-MOD phantom sections that contained imperfections exceeding 1 cm in diameter were rejected and subsequently remanufact ured. The 1 cm diameter criterion was selected to ensure that phantom imperfec tions would not adversely interfere with subsequent image quality analysis. The attenuation of the BRTES-MOD pha ntom was examined by evaluating the tube-current time product (mAs) response by the automatic exposure control (AEC) system of the mammography unit. The AEC syst em uses a phototimer that terminates the 13 General Electric Medical Systems, 3135 Easton Turnpike, Fairfield, CT 06431 14 Eastman Kodak, Rochester, NY
135 exposure once the radiation level has reach ed a predetermined calibration level corresponding to a desired film optical densit y. Four phantoms, (i.e. total thickness of 4 cm) of 67.8, 42.6, 25.4 and 16.2 % glandularity phantoms were evaluated for their AEC response using a Mo target and Mo filte r at a tube potential of 26 kVp. In order to evaluate the cured phantom tissu e equivalency to the reference tissues of ICRU Report 46, a full spectrum comparison was conducted (10-50 keV) of the Âµ -1 and Âµen -1 of BRTES-MOD, acrylic and BR 12 phantom materials. The acrylic and BR 12 were used for comparison purposes. A ratio was calculated for Âµ -1 or Âµen -1of the cured-phantom elemental composition to th e elemental composition of the reference tissues derived from ICRU-46. Results and Discussion The final array of BRTES-MOD compositi ons for the range of glandularities designed in this study is delineated in table 5-2. As noted in table 5-2, compositions of ingredients were adjusted to match ICRU46 tissue compositions. The criterion used to determine the acceptability of a batch of epoxy -resin matrix was the compatibility (Â± 3%) of the Âµ -1 and to the reference tissue. The 3% criterion was based on 95% confidence interval of the mass-density in the isopropyl alcohol mass-de nsity reference, and various steps in the manufacturing process, such as material weighting (Â± 0.01 gm), mixing and vacuum pressure. The Âµen -1 was calculated using equation 5-2 for each epoxy-resin batch was compared to ICRU-46 reference tissu e after the initial matrix components were finalized. The elemental compositions, effective atomic number ( Zeff), effective photoelectric effect atomic number (PE effZ ) and mean mass density are listed in table 5-3
136 for ICRU-46 referenced tissue and table 54 for BRTES-MOD phantom tissue equivalent compositions. As Attix and Johns and Cunningham have described, Zeff and PE effZ to be good predictors of photoelectri c absorption. Therefore, Zeff and PE effZ have been used to indicate the compatibility of the BRTES-MOD phantom Âµ -1 and Âµen -1 to ICRU-46 reference tissue.115,117 The difference in between Zeff and PE effZ BRTES-MOD and ICRU46 reference tissue were less than 2.5% for th e entire range of glandularities examined, figure 5-6. The average mass-density closel y matched (Â± 2%) the ICRU-46 reference tissue. The BRTES-MOD mean mass density was ba sed on the matrix batches that meet all manufacturing criteria, being within 3% of the referenced ti ssue in mass-density, Âµ -1 and Âµen -1. The 3% criterion was selected to acc ount for variations in the mass density of the BRTES-MOD phantom, turned out to be th e most difficult to control with low massdensities, as noted in figure 5-7 BRTES-M OD density variability about ICRU-46 target density. The factors that were best in controlling the mass density were the vacuum mixing and the microsphere content in the matrix. One method of evaluating the attenuation of a material is to maintain an optical density that that will accentuate tissue diffe rences by utilizing the AEC system in the mammographic unit. The AEC system desi gn increases mAs as percent glandularity increases. The required tube-current tim e product (mAs) at a constant kVp is an indication of the overall attenua tion characteristics of the pha ntom material as a function of glandularity. The attenua tion characteristic of the BRTES-MOD phantom was linear as shown in figure 5-8, with a 0.999 R2. The results were consistent with clinical mammography images (higher glandular cont ent corresponds to higher attenuation as
137 shown in a mammography image) and Âµ -1-glandular response seen in figure 5-9. These results indicate that BRTES-MOD phantom series is in good agreement with mammography clinical imagi ng. The differences between BRTES-MOD and ICRU-46 target Âµ -1 at 25 keV are shown in figure 5-8. The largest differences in Âµ -1 was noted with 100% adipose at 2.97 % and 100% gla ndular at 1.03 % tissue which was directly attributed to the 3% criteri on selected to manufacturing th e phantom. A correction was made to the criterion for the remaining glandularities of less than 0.5 % difference between the Âµ -1 and the ICRU-46 derived tissue at 25 keV. A further analysis was performed by comparing the Âµ -1 and Âµen -1 over the entire mammography energy range (10 -50 ke V) for BRTES-MOD, methyl-methacrylate (acrylic) and BR 12 to ICRU-46 reference ti ssue illustrated in fi gure 5-10 and figure 511. The ratio of Âµ -1 between acrylic and BR 12 were comp ared to 50% glandular tissue composition derived from ICRU-46 referen ce tissues compositions. The ratio of Âµ -1 between BRTES-MOD phantom (67.8, 50, 42.6, 25.4 and 16.2 % glandular content) to ICRU-46 reference tissue perf ormed better overall than both acrylic and BR 12 as illustrated in figure 5-10 and figure 5-11. The ratio of Âµ -1 between BRTES-MOD phantom 100% glandular Âµ -1 ratio to ICRU-46 referen ce tissue ranged from 0.96 to 1.02 whereas 100% adipose ranged from 0.99 to 1.07. However, the BRTES-MOD phantom (67.8, 50, 42.6, 25.4 and 16.2 % glandular content) Âµ -1 ratio to ICRU-46 reference tissue ranged from 1.01 to 0.99 and Âµen -1 ratio ranged from 1.01 to 1.03. Acrylic Âµ -1 ratio to ICRU-46 reference tissue ranged from 0.89 to 0.96 and Âµen 1 ratio ranged from 0.87 to 0.97. Overall ac rylic tended to undere stimate the ICRU-46 reference values throughout the energy range. The BR 12 Âµ -1 ratio to ICRU-46
138 reference tissue ranged from 0.95 to 1.02 (7%) and Âµen -1 ratio ranged from 1.01 to 1.09 (8%); the BR 12 Âµen -1 ratio overestimated the ICRU-46 reference values. BR 12 and the BRTES-MOD phantom series both overestima te the ICRU-46 reference values for low energy photons, which may be a result of the hi gher concentration of chlorine, as noted by White et al..51 The BRTES-MOD phantom series has a grea ter flexibility than either BR 12 or acrylic primarily because it provides compatibil ity with a larger range of glandularities and therefore is more representative of clinical mammography populations. In addition, the BRTES-MOD phantom series provides a cl oser match in mass density (Â±3%), Âµ -1 and Âµen -1 (Â±2%) as compared with ICRU-46 reference tissues. Conclusions The BRTES-MOD phantom is presented as a breast tissue equivalent series of phantoms that closely mimics the radiological characteristics of ICRU-46 adipose and fibroglandular tissues for ma mmography applications. The modification performed to BRTES has improved the overa ll homogeneity by changing the type of microspheres used and reducing the incorporated ai r by mixing under vacuum. The BRTES-MOD series also has the added benefit of being the geometrical dimensions of the dosimetric model used by Wu and adopted by the ACR, w ith a range of percent glandularities. Patient doses based on this series of phantoms will be more representative of patient population. The BRTES-MOD phantom se ries out performed either acrylic or BR 12 when evaluated by their Âµ -1 and Âµen -1 in comparison with their reference tissues as described in ICRU Report 44. Further refine ment is needed for the 100% glandular and 100% adipose BRTES-MOD to further reduce the differences from their respective ICRU-46 reference tissues. The BRTES-MOD pha ntom series provides the platform to
139 perform patient glandularity re ferenced measurements and i ndividual patient doses in the clinical environment. A potential for the BRTES-MOD phantom is to supplement the ACR phantom to evaluate AGD for a variety of glandularities in orde r to provide a better assessment of the AGD for an individual patient.
140 (A) (B) (C) Parafin insert image quality object descriptions 1. 1.56 mm nylon fiber9. 0.32 mm specks 2. 1.12 mm nylon fiber10. 0.24 mm specks 3. 0.89 mm nylon fiber11. 0.16 mm specks 4. 0.75 mm nylon fiber12. 2.00 mm tumor-like mass 5. 0.54 mm nylon fiber13. 1.00 mm tumor-like mass 6. 0.40 mm nylon fiber14. 0.75 mm tumor-like mass 7. 0.54 mm specks15. 0.50 mm tumor-like mass 8. 0.40 mm specks16. 0.25 mm tumor-like mass 00001234 5 6 78 9 10 11 12 13 141516 Figure 5-1. (A) ACR phantom. (B) ACR phantom radiograph. (C) Image quality objects located within the paraffin insert with their respective descriptions.
141 Compressed Breast Thickness (cm) 0246810Glandularity (%) 10 20 30 40 50 60 70 80 90 Figure 5-2. Glandularity as a function of compressed breast thickness based on Geise and Palchevsky work.29
142 XY PlaneXZ Plane YZ Plane 3D Plane 18 cm 1-8 cm 8 cm X Y Z Figure 5-3. The BRTES-MOD phantoms have a PNL of 8cm and a width of 18 cm in 1 cm thick sections. Each representative BRTES-MOD percent glandularity has 8 sections.
143 100.0% 83.4% 67.8% 54.2% 42.6% 33.0% 25.4% 19.8% 16.2% 0.0% 0.0% 0.0%0.0% 8 cm 8 c m 1 cm 1 cm 8 cm 1 2 c m 9 cm 1 c m(A) (B) (C)100.0% Figure 5-4. (A) The BRTES-MOD step phantom dimensions (B) Individual step dimensions (C) BRTES-MOD step phantom percent glandularit.
144 Energy (keV) 1020304050Total Attenuation Interaction Coefficient Fraction 0.0 0.2 0.4 0.6 0.8 1.0 100% Glandular Tissue (ICRU-46), R -1 100% Glandular Tissue 100% Glandular Tissue 100% Adipose Tissue (ICRU-46), R 100% Adipose Tissue 100% Adipose Tissue Figure 5-5. The fractional contribution to the total attenuation coefficient from photoelectric effect ( -1), Compton effect ( -1), and Raleigh scattering ( R -1)as a function of photon energy for 100 % adipose and 100% glandular tissues.
145 Glandualrity (%) 020406080100Effective Atomic Number 5.6 5.8 6.0 6.2 6.4 6.6 6.8 7.0 7.2 7.4 7.6 BRTES-MOD Zeff ICRU-46 Zeff ICRU-46 Zeff PE Upper and Lower 2.5% BRTES-MOD Zeff PE Figure 5-6. The graph shows th e effective atomic number (Zeff) and the photoelectric effective atomic number(Zeff PE) for ICRU-46 reference tissues and BRTESMOD phantoms as a function of percen t glandularity. The upper and lower 2.5% lines are in refere nce to the corresponding Zeff and Zeff PE.
146 Glandularity (%) 020406080100Density (gm cm-3) 0.90 0.92 0.94 0.96 0.98 1.00 1.02 1.04 1.06 ICRU-44 Target Density Upper and Lower 3% Limit BRTES-MOD Figure 5-7. Mass density of BRTES-MOD phantom series and the corresponding ICRU46 reference tissues as a func tion of percent glandularity.
147 Glandularity (%) 1020304050607080Tube-Current Time Product (mAs) 65 70 75 80 85 90 95 100 105 Figure 5-8. Tube-current time product (mAs ) as a function of phantom composition.
148 Glandularity (%) 020406080100Mass Attenuation Coefficient ( -1(25 keV)) 0.36 0.38 0.40 0.42 0.44 0.46 0.48 ICRU-44 Target -1 BRTES-MOD -1 Upper and Lower 3% Limit Figure 5-9. Mass attenuation co efficient (25 keV) of BRTES -MOD phantom series, their corresponding ICRU-46 target tissues, and upper and lower 3% limit as a function of glandularity.
149 Photon Energy (keV) 1020304050 Mass Attenuation Coefficient ( -1) Ratio 0.85 0.90 0.95 1.00 1.05 1.10 BRTES-MOD 0% Glandular BRTES-MOD 100% Glandular BRTES-MOD Midrange Methyl-methacrylate BR 12 Figure 5-10. Ratio of Âµ -1 for BRTES-MOD phantom series, acrylic, and BR 12 tissue substitutes to their corresponding ICRU46 reference values as a function of photon energy. The BRTES-MOD phantom series were plotted as three separate entities of 100 % adipose, 100 % glandular, and a midrange from 16.2-67.8 % glandularities. The midrange Âµ -1 ratio is the average Âµ -1 ratio of 67.8, 50.0, 42.6, 25.4 and 16.2%. The e rror bars of the BRTES-MOD midrange glandularities repres ent the variabil ity of the Âµ -1 ratio to ICRU-46 for the midrange percent glandularities.
150 Photon Energy (keV) 1020304050Mass Energy Absorption Coefficient ( en) Ratio 0.7 0.8 0.9 1.0 1.1 1.2 BRTES-MOD 100% Glandularity BRTES-MOD 0% Glandularity BRTES-MOD Midrange Glandularity Methyl-methacrylate BR 12 Figure 5-11. Ratio of Âµen -1 for BRTES-MOD phantom series, acrylic, and BR 12 tissue substitutes to their corresponding ICRU46 reference values as a function of photon energy. The BRTES-MOD phantom series were plotted as three separate entities of 100% adipose, 10 0% glandular, and a range from 16.267.8% glandularities. The midrange Âµen -1 ratio is the average Âµen -1 ratio of 67.8, 50.0, 42.6, 25.4 and 16.2%. The error bars of the BRTES-MOD midrange glandularities repres ent the variabil ity of the Âµen -1 ratio to ICRU46 for the midrange percent glandularities.
151 Table 5-1. Commercial suppliers used in th is study for epoxy-resin matrix components. Epoxy-matrix component Chemical Family Source Araldite (GY-60-10), liquid epoxy resin Epoxide Vantico Inc., Brewster, NY Jeffamine (T-403), epoxy resin hardener Alkyl ether amine Huntsman, Houston, TX West System 410 Microlight microspheres Thermoplas tic polymer Gougeon Brothe rs, Inc., Bay City, MI Polyethelene Powder (medium density) [C2H4]n Aldrich Chemical Co., Milwaukee,WI Magnesium Oxide MgO Fisher Scientific, Fairlawn, NJ
152Table 5-2. Breast tissue equivalent modified series (BRTES-MOD) ma terial composition by percent. Fibroglandularity (%) Epoxy Resin Matrix Components (% by mass) 100.00 83.40 67.80 54.20 50.00 42.60 33.00 25.40 19.80 16.20 0.00 Araldite GY 60-10 (Epoxy) 49.41 47.08 47.39 47.96 45.40 51.56 51.37 50.90 53.54 47.11 48.59 Jeffamine T-403 (Hardener) 19.77 18.83 18.96 19.18 18.16 20.62 20.55 20.36 21.42 18.84 19.44 Polyethylene Powder (Medium Density) 18.50 23.50 23.95 24.60 28.04 20.58 21.57 22.72 19.58 28.60 26.36 Magnesium Oxide 11.42 9.90 9.00 7.50 7.48 6.29 5.65 5.16 4.67 4.80 4.70 West System 410 Microlight (Thermo Plastic microspheres) 0.90 0.69 0.70 0.76 0.92 0.95 0.86 0.86 0.79 0.66 0.91
153Table 5-3. Elemental composition, effective atomic number and the mass density of the BRTES-MOD epoxy-resin matrix. Effective atomic number was calculated by taking the sum of the percent by we ight atomic number of element.15 Elemental composition (% by weight) of fibroglandular tissue content (%) Elements 100 83.4 67.8 54.2 50 42.6 33 25.4 19.8 16.2 0 C 0.6330 0.6531 0.66000.67110.67630.67140.67800.6833 0.68160.69700.6925 O 0.1916 0.1779 0.17530.17120.16450.17760.17410.1708 0.17620.15760.1626 H 0.0815 0.0860 0.08700.08860.09070.08680.08800.0892 0.08760.09340.0918 Mg 0.0689 0.0597 0.05430.04530.04520.03800.03410.0312 0.02820.02890.0284 N 0.0207 0.0197 0.01980.02010.01900.02160.02150.0213 0.02240.01970.0203 Si 0.0020 0.0015 0.00150.00170.00200.00210.00190.0019 0.00170.00140.0020 Cl 0.0012 0.0011 0.00110.00120.00110.00120.00120.0012 0.00130.00110.0012 Na 0.0007 0.0006 0.00060.00060.00070.00080.00070.0007 0.00060.00050.0007 Al 0.0004 0.0003 0.00030.00030.00040.00040.00040.0004 0.00030.00030.0004 Ca 0.0001 0.0001 0.00010.00010.00010.00010.00010.0001 0.00010.00010.0001 effZ 6.4425 6.3312 6.28876.22026.19726.20466.16656.1363 6.13736.06846.0892PE effZ 7.2639 7.1158 7.05106.94456.93126.89796.83686.7915 6.77056.70426.7335 Density 1.0095 1.0187 0.99961.00320.96620.97590.99170.9669 0.98130.95990.9542 Density 0.0013 0.0027 0.00080.00170.00050.00050.00120.0005 0.00270.00110.0014 N 15 5 30 5 20 20 5 30 5 45 10 15i i i effZ w Z and 5 . 3 5 . 3 i i PE effZ a Z where i i i i i i i iA Z w A Z w a/, wi , Zi, Ai are the mass fraction, atomic number, and mass number, respectively of element i.
154Table 5-4. Elemental composition, effective at omic number, effective photoelectric eff ect atomic number and the mass density of the ICRU-46 reference tissue. Only 0 a nd 100% fibroglandular tissue composition was taken form ICRU-46. All other glandularities were percent weight com positions of these two ICRU-46 tissues. 16 Elemental composition (% by weight) of fibroglandular tissue content (%) Elements 100 83.4 67.8 54.2 50 42.6 33 25.4 19.8 16.2 0 H 0.1060 0.1073 0.1086 0.1097 0.1100 0.1106 0.1114 0.1120 0.1124 0.1127 0.1140 C 0.3320 0.3762 0.4177 0.4538 0.4650 0.4847 0.5102 0.5304 0.5453 0.5549 0.5980 N 0.0300 0.0262 0.0226 0.0195 0.0185 0.0168 0.0146 0.0128 0.0116 0.0107 0.0070 O 0.5270 0.4857 0.4468 0.4130 0.4025 0.3841 0.3602 0.3412 0.3273 0.3183 0.2780 Ca 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Cl 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 Na 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 0.0010 P 0.0010 0.0008 0.0007 0.0005 0.0005 0.0004 0.0003 0.0003 0.0002 0.0002 0.0000 S 0.0020 0.0018 0.0017 0.0015 0.0015 0.0014 0.0013 0.0013 0.0012 0.0012 0.0010 effZ 6.5690 6.4765 6.3896 6.3139 6.2905 6.2493 6.1958 6.1535 6.1223 6.1022 6.0120 PE effZ 7.2163 7.1198 7.0258 6.9412 6.9146 6.8670 6.8039 6.7530 6.7147 6.6899 6.5751 Density gm cm-3 1.0200 1.0077 0.9964 0.9867 0.9838 0.9786 0.9720 0.9669 0.9631 0.9607 0.9500 16 i i i effZ w Z and 5 . 3 5 . 3i i PE effZ a Z where i i i i i i i iA Z w A Z w a/, wi , Zi, Ai are the mass fraction, atomic number, and mass number, respectively of element i.
155 CHAPTER 6 ANTHROPOMETRIC VARIATIONS IN MAMMOGRAPHY Introduction The adoption of standardized phantoms to evaluate image quality and radiation dose of mammography units has dramatica lly improved the practice of screening mammography.5,15,17-22 Unfortunately, the phantoms in wide spread use do not accurately reflect the variations of br east glandularity or geometry of the mammography population. A phantom requires tissue equivalence and popula tion correct anatomical geometry to be used effectively to determine individua l doses in a clinical environment. Tissue equivalence ensures that the radi ological characteristics of the phantom tissue-substitutes match those of the tissues of interest in patients. The anatomic geometry plays an important role in the sca ttering characteristics of the exposed tissue, which directly contributes to the deposited dose.118 Tissue composition has been addressed with the publication of Intern ational Commission on Radiation Units and Measurements Reports 44 and 46 which compile an aggregate of published data for all major tissues in the human body.107,114 Unfortunately, no equiva lent publication exists for the radiographic anatomical geometries for mammography. Numerous studies in mammography have restricted their population measurements to determination of the av erage compressed breast thickness (4.9 5.7 cm ) and percentage breast glandularity (34 47%).29,50,58-62,81,119,120 An expanded human population study provides the demogra phic data of anatomical vari ations that is needed to design and manufacture realis tically-shaped dosimetry pha ntoms for mammography. In
156 this study, we retrospectively studied 253 mammography patients to determine the anthropometric variation of the breast under screening mammography conditions. Materials and Methods A three-month retrospective demographic study was conducted that compiled mammography data acquired from 253 patients seen at a U.S. Navy health care facility in Jacksonville, Florida. One thousand and fo rty mammograms taken in the period from January through March 2005 were reviewed and digitized fo r this study. The clinical mammography study population consisted of wome n who were seen either for screening mammography or as a result of a referral from their primary care physician for a diagnostic mammography exam. The typical mammography examination cons ists of four images taken in two views for both right and left breasts (craniocaudal (CC) and mediolateral oblique (MLO)). The mammographic units used to im age patients utilize a proprietary automatic exposure control modality that optimizes unit parameters, by selecting the tube potential (kVp), tube current-time product (mAs), ta rget and filter combination after short, preliminary x-ray exposure. The patients that were selected fo r the study had a American College of Radiology (ACR) breast imaging repor ting and data system BI-RADSÂ® assessment categories of 1 (Negative) or 2 (Benign Fi ndings) assigned to them by the interpreting radiologist.11 A negative BI-RADS assessment categor y indicates that patient had nothing for comment, and the breasts were symmetric with no masses, architectural distortion or suspicious calcifications. A benign BI-R ADS assessment category indicates a normal assessment, but the radiologist annotated a benign finding such as involution, calcified fibroadenoma, multiple secretory calcifications, fat-containing lesion or mixed-density
157 hamartoma. The radiologist also may choose to report a patient as category 2 if there was evidence of a prior surgery, breasts implant or architectural distortion. Overall, BIRADS category 1 and 2 assessments are indica tive of normal breasts with no evidence of malignancies. A General Electric Senographe DMR (G E-DMR) or a GE Senographe-800T (GE800T) provided the mammography x-ray imaging used throughout the study. The Senographe DMR used a high-fre quency generator and an x-ray tube with a dual bi-metal track target (T) of molybdenum (Mo) and rhodi um (Rh). The x-ray sp ectrum was filtered using a Mo or Rh filter (F). The unit had a fi xed source-to-image receptor distance of 660 mm and a nominal focal spot size of 0.3 mm. The Senographe-800T used a highfrequency generator and an x-ray tube with a single metal track target of Mo. The x-ray spectra were filtered using Mo or Rh filters. The unit had a fixed source-to-image receptor distance of 660 mm and a nominal focal spot size of 0.3 mm. Both the GEDMR and GE-800T were operated in automatic optimization parameter (AOP) mode whenever possible. The AOP mode gave pr iority to dose reducti on (AOP-D), subject contrast (AOP-C) or a compromise between dose reduction and subjec t contrast (AOP-S). The AOP mode selected the target (T), filter (F), tube potential (kVp), and tube currenttime product (mAs), through the use of a pre-ex posure which evaluate d the radiographic beam penetration. The system used the prelimin ary x ray exposure of the breast to select the proper value in the calibration table that designates the T/F, kVp and mAs for the individual patient breast imag e. The radiation emerging fr om the breast passes through the patient support table, the antiscatter grid, the cassette front, film base before being absorbed by the intensifying screen and film emulsion.
158 As a U.S. Food and Drug Administrati on certified and American College of Radiology (ACR) accredited facility, the studie d facility maintains its operations in accordance with Mammography Quality Standa rds Act (MQSA) regulations and the ACR Mammography Accreditation Program (ACR-MAP) guidelines as specified.12,100 Throughout the study, the facility performed th e required daily quality control activities and ensured that the prescribed control limits were met prior to processing any patients.12 In addition, and as part of the MQSA a nd ACR-MAP, the mammography x-ray unit units were surveyed on an annual basis to verify optimum performance.12 The Kodak17) Min-R 2000 screen-f ilm combination was used throughout the study. The film was developed using a K odak X-OMAT 5000 RA film processor. Processor quality control was performed prio r to patientsÂ’ films being developed, in accordance with MQSA and ACR-MAP. The data collected for each mammogram are shown in Table 6-1. The data were compiled from each mammogram using paper fo rms that were subsequently transferred to a database program designed for this study18). The statistical analysis performed on the population data was done using SigmaStat19) (version 3.119). The two main statistical tests utilized for the study were the MannWhitney (MW) rank sum test and the KruskalWallis (KW) one-way analysis of variance on ranks Non-parametric statistical tests were utilized to evaluate differences between two or more groups. 17 Eastman Kodak Company, Rochester,NY 18 Microsoft Office Access 2003 , Microsoft, Redmond, WA 19 SigmaStat 3.1, Systat Software, Inc., Point Richmond, Ca
159 Digitization of mammograms and daily pr ocessor control film s was done using a Kodak LS-75 film digitizer with a limiting resolution of 5 lines pair per mm (0.1mm).121 All mammograms were digitized at a resoluti on width of 2048 pixels with the number of lines determined by the film size. The matrix sizes for the 18 x 24 cm2 and 24 x 30 cm2 film sizes were 2048 x 2812 and 2048 x 2628 pixe ls, respectively. The pixel size was approximately 0.1 mm x 0.1 mm, corresponding to the limiting resolution of the scanner.121 All digitized mammogr ams were archived in a Digital Imaging and Communications in Medicine (D ICOM) format for processing. Manual segmentation was performed using ImageJ20 (version 1.34i), image analysis software developed by the National Institutes of Health.122 Each CC view and MLO view of each mammogram was manually se gmented into the regions and segments illustrated in figure 6-1. The criteria for manual segmentation are described below for each radiographic position. The CC view incl uded two distance measurements and four area measurements. The CC width of the breas t was measured as the distance from the medial plane to the axilla. The posterior ni pple line (PNL) was defined in the CC view as the distance from the chest wa ll anteriorly to the nipple. The MLO PNL was defined as the perpendicular distance from the pectoral muscle anteriorly to the nipple. The whole (whole) breast area was defi ned as that area encompassed by the skin edge in both CC and MLO views. The breast edge (bedge) in the CC and MLO views was defined to be the region anterior to the glandular region of the breast that is primarily comprised of adipose tissue and the associated skin layers. To define th e breast edge region in ImageJ, a rectangular 20 ImageJ, National Institutes of Health, Bethesda, MD
160 region of interest was made that encompassed the entire breast, excl uding labels such as the patient identification or view markers, a nd then pixel intensities were plotted as a function of distance from the chest wall, figur e 6-2. Two values are established from the plot to identify the minimum pixe l value of the plateau region (X2) and minimum pixel value of the valley (X1) shown in figure 6-2. Using equa tion (6-1), a maximum threshold value (TV) was established based on X1 and X2, figure 62. 21 2 1X X X TV (6-1) The breast edge area was ultimately defined by thresholding the image for pixel intensities less than or equal to the calculated TV. The uniform region of the breast was defi ned to be the whole breast area minus the breast edge in all images. The de nse fibroglandular region was subjectively established by visual inspection of each image. This dense region was defined independent of other views taken of the sa me subject for both CC and MLO views. The subjective criterion used was that the mini mum pixel value corres ponding to glandular tissue was 64% of the maximum pixel intensity of 4095. Contrary to clinical practice, the approach ensures that dense region of each mammogram is defined without the influence of the other views. The primary reason for this method was to determine whether the dense regions varied between the left and right views. The MLO view had four distance measurements and five area measurements. In addition to the aforementioned measurements th e MLO view also had the pectoral muscle (PecMaj) area measured. The MLO Pectoral length (PecMajD) was defined as the mammogram's cranial caudal distance of the pectoral muscle on the image. The MLO axilla distance was defined as the distance from chest wall anteriorly to the skin edge in
161 the MLO view. The MLO breast length (CCD) was defined as the cranial caudal distance from the axilla to the inferi or skin edge at the chest wall. All measurements from the MLO and CC views were treated as separate populations because the MLO view includes a large portion of the pectoral musc le, while the CC view does not. The anthropometric measurements collected from the CC and MLO views were used to determine an analytical expressi on for the CC and MLO whole breast areas, to estimate the thickness of the subcutaneous adi pose and skin layers and to determine the optimal location for the AEC ionization ch amber with respect to the PNL. The CC and MLO area analytical expressi ons were based on simple geometry models using the measurements collected illu strated in figure 6-3. The area equation for the breast geometry in the CC view is based on a semielliptical geometry, where the major axis is of the CC width and the minor axis is the PNL measurement illustrated in figure 6-3, equation (6-2). The area equa tion for MLO view was derived from the rotation of the CC area to accommodate a triang ular pectoral muscle , where the base of the triangle is the axilla measurement, a nd the height of the triangle is the CCD measurement illustrated in figure 6-3, equati on (6-3). The area equations were fitted using Sigmaplot21) (version 9) which uses a Marqua rdt-Levenberg nonlinear least-squares algorithm, where aCC and aMLO are the corresponding fitting parameters.123 width PNL a CCCC area 4 (6-2) CCD axilla width PNL a MLOMLO area ) 2 1 ( 4 (6-3) The estimated thickness of the subcutan eous adipose and skin was based on theare of the breast edge and perimeter measur ements. The breast edge was modeled by 21 Sigmaplot 9.0, Systat Software, Inc., Point Richmond, CA
162 equating the breast edge of th e CC or MLO view to an equi valent rectangular region as, illustrated in figure 6-4. The rectangular regi on was defined at the horizontal mid-plane of the compressed breast illustrated in figure 6-4. The measurements of the breast edge area and perimeter were used to determine the breast edge width based on a rectangular area by using equations 6-4 and 6-5, where th e breast edge measured perimeter is PBedge, the measured area is ABedge, the source-to-image receptor distance is SID and the sourceto-object distance is SOD. Adjustment was made to the breast edge width for image magnification, as shown in equation (6-5). 4 162 Bedge Bedge Bedge LengthA P P Bedge (6-4) ) ( SOD SID Bedge A BedgeLength Bedge Width (6-5) The current clinical method of determ ining the optimal location for the AEC ionization sensor is evaluati ng previous mammograms. Unfort unately, this is not possible for new patients. In this case, typically the technologists place the ionization chamber where they believe the densest part of the breast to be. The results of this study will provide technologists an estimate of the mo st likely location for the AEC sensor. The PNL ratio was derived by dividing the PNL di stance by the dense tissue centroid distance from the chest wall. Results and Discussion The study population of 253 subjects ha d a mean age of 54.32 Â± 12.11 years. The range of ages in the study group was 30-89 y ears old, with the median age being 52 years old. The distribution of subj ects was consistent with the recommendations of the American Cancer Society for screening mammography population age (>40 years), figure 6-5.5 Subjects under the typical age (<40) for mammography imaging were likely to
163 have been referred by a primary care physic ian for a diagnostic study to evaluate potential abnormalities. Out of the 253 subj ects reviewed for this study, 143 were assigned a BI-RADS assessment category 1 w ith the remainder being assigned BI-RADS assessment category 2. The total number of mammograms (1040) reviewed included subjects that had multiple images (>4) and th e repeated images were not included in the population statistics. The BI-RADS density category is subject ively assigned to the patient by the interpreting radiologist, base d on professional experience, training and guidelines provided in the ACR BI-RADS. It is important to note that a radiol ogist assigns the BIRADS density category to the entire mammogr aphy examination for a patient and not to an individual image. In order to evalua te the fibroglandular content of the study population, each BI-RADS density category wa s assigned a mean fibroglandular value related to the mid-point of th e BI-RADS density categories, as shown in Table 6-2. The BI-RAD density category assignment was eval uated with its consistency to population trends in hormone replacemen t therapy (HRT), compressed breast thickness, and subject age. Current population trends indicate that women receiving HRT have a higher glandular tissue content and therefore the population trend for HRT patients is a higher BI-RADS density categories.44,124 Population studies indicate that glandularity decreases as compressed breast thickness increases and therefore the populat ion trend would be reflected in a decreasing BI-RADS density categories.29,82 The last factor utilized to evaluate the BI-RADS density category assignment was age, as previous studies have shown a decrease of glandularity with increasing age.106,120 The general distribution of BI-RADS density categories was similar to pr evious studies as il lustrated in figure 6-
164 6.119,125,126 The general distribution from the studies shown in figure 6-5 and this study indicate that the distributi on of BI-RADS density categories are approximately ~10%, ~40%, ~40%, ~10% for the respective BI-R ADS density categories of <25%, 25-50%, 51-75% and >75%. However, when glandul arity distribution is analyzed for each contributing radiologist, Tabl e 6-3, the distributions were not consistent with the aggregate data or previous studies. BI-RADS density category <25% ranged from 0-22% among the 9 radiologists, 25-50% ranged from 23-70%, 51-75% ranged from 9-63% and >75% ranged from 0-19%. This type of vari ation amongst density category distributions (for individual radiologist) ha s been previously reported as a result of differences in their education and professional experience.37,127 The variation in BI-RADS density category distributions for individual radi ologists could be adjusted if a facility were to establish a unified protocol for interpretation of BI-RADS density categories. Only 51 out of 253 subjects in this st udy were receiving HRT. When they are compared to women not receiving HRT of th e same age (39-80 years old, 183 subjects), Table 6-4, no statistically si gnificant difference in the medi an BI-RAD density categories (p= 0.365, MW) was found. Contrary to previous studies, HRT was not found to increase glandularity in this study.44,124 The percent glandularity as a function of compressed breast thickness showed a statistically significant tre nd (p=0.001, KW) of decreasing percent glandularity and increasing with compressed breast thickness, as indicated in Table 6-5. The trend was qualitatively similar with prev ious published data, that is percent glandularity decreases with increasing compressed breast thickness with a slope of -3.7 (R2=0.89).29,120
165 Glandularity as a function of age group, shown in Table 6-6 showed a statistically significant trend (p 0.001, KW) indicating an increase in adipose tissue with age group. The trend related to age group was qualitative similar to previous published data, that is glandularity decreases with age with a slope of -2.34 (R2=0.81).106,120 Compression pressure, unlike BI-RADS glandularity, is applied by the technologist to each individual breast in each view. Therefore the data compressions were evaluated by view (CC or MLO). The co mpression pressures a pplied to CC and MLO views, figure 6-7, were statistically different (p 0.001, MW) as expected, it is possible that the MLO view requires additional pressu re to fix its position as a result of the pectoral muscle. The mean compression pre ssure applied in the CC view was 9.91 Â± 3.28 daN. The mean compression pressure applied in the MLO view was 12.38 Â± 3.06 daN. The compression pressure applied during a mammogram is highly dependent on the training and experience of the mammogra phy technologist. The differences among technologists were evaluated. The result was that the application of compression pressure differed among technologists, Table 6-7, (p 0.001, KW). However, all technologists were within Â± 68% of the mean pressure (11.33 Â± 1.77 daN) applied except for technologist 004. Technologist 004 or 005 only had one patient each and therefore were not included in the mean value calcula tion. In evaluating compression as a function of breast thickness a statistically signif icant trend was observed, Table 6-8, (p 0.001, KW) that indicated that compression pressu re was increasing with compressed breast thickness. This trend with breast size was expected because a greater amount of tissue requires greater force to reduce overlappi ng anatomy, decrease tissue thickness and reduce motion blur. No statistical correlation was noted in this study between subject age
166 and compression pressure (p=0.170, KW). Previ ous studies have treated the compression pressure in both CC and MLO as statistically equal and therefore aggregated the data; however, the data observe d in this study do not support this approach.29,50,58-60,62,81,119,120 Compressed breast thickness was measur ed and annotated on the mammogram by the mammography unit. The mammography units utilized in this study complied with the compression thickness scale accuracy and reproducibility limits of MQSA and ACR.12,100 The regulations require that under moderate compression (6.7 to 8.9 daN), the thickness measurement scale on the mammographic unit must be reproducible within 2 mm and accurate to 5 mm for compressed breast thickness of 1 to 8 cm.12 The compressed breast thickness distribution by CC and MLO views is shown in figure 6-8. The difference in compressed breast thickness between CC and MLO views was statis tically significant (p 0.001, MW), indicative of two distinct populations. The mean compression thicknesses for the CC view and MLO view s were 4.46 Â± 1.17 cm and 5.22 Â± 1.39 cm respectively. The difference in thickness be tween CC and MLO can be directly attributed to pectoral muscle being included in the MLO view and excluded in the CC view. The study data do not support the approach of aggregating the compre ssed breast thickness into a single population, in cont rast to previous studies. 29,50,58-60,62,81,119,120 The distribution of tube potential (kVp) as a function of CC, MLO, and target/filter combina tion are shown in figure 6-9. The tube current-time product (mAs) distribution is shown in figure 6-10. Table 6-9 shows the unit parameters as a function of view (CC or MLO) and target/filter combina tion. The data provide insights into the parameterization algorithm (AOP) of the ma nufacturer. The data show that the target/filter combination tracks the compresse d breast thickness, whic h results in higher
167 tube potentials and higher t ube-current time products. Th e tube potential graph shows that the MLO view requires a wider range of tube potentials when compared to the CC view. In addition, that data show that 29 kVp was chosen by the unit less frequently than 28 or 30 kVp with Mo/Rh. This effect was not noted with Rh/Rh. Anthropometric breast measurements as a function of compressed breast thickness and CC or MLO views are shown in Table 6-10 and Table 6-11, respectively. The CC and MLO view dimensions and areas increase as a function of compressed breast thickness, and glandularity decreases as prev iously discussed. The CC mean values of width and PNL are larger than those used by Wu dosimetric model by approximately 3 cm and 2 cm respectively. The increase in PN L and width results in a 46% increase in CC breast volume. When compared to the average breast phantom used by ACR, the study population average CC view results indicate th at the average breast has a approximately 28% increase in PNL distance, 17% increase in width, resulting in a 46% increase in volume with no appreciable in crease in compressed breast th ickness or glandularity. The data indicate that the Â“average breastÂ” curr ently used by Wu and ACR is similar to the study population average breast except for the vo lume increase which is likely to have a small impact (<1%) in glandular dose. MLO da ta results show that the pectoral muscle takes up 21% of the whole breast area in the MLO view. The current Wu dosimetry model does not adjust for the MLO view doing so would require new dose conversion factors based on a breast containi ng the pectoral muscle. An analysis was also conducted to dete rmine an analytical expression predicting the whole breast area as function of PNL, widt h, CCD and axilla. Equations (6-3) and (64) were good predictors of the measured areas corresponding to CC and MLO views as
168 shown in figures 11 and 12. The equations provi de the data needed to develop realistic anthropometric dosimetry phantoms that are population-based. The population distribution of the breast edge widths (Bedgewidth) calculated are shown in figure 6-13. The mean Bedgewidth for CC view was 0.20 Â± 0.03 cm and MLO view was 0.22 Â± 0.03 cm. The calculated Bedgewidth indicates that the current dosimetry models (0.4 cm) are overestimating the sk in thickness. Given that current dose conversion factors decrease with increasing gl andularity, the decrease in adipose region would result in an higher gla ndularity that corresponds to a decrease in glandular dose. The positioning of the automatic expos ure control (AEC) ionization chamber under the densest tissue ensures proper image quality for disease de tectability. There was no statistical difference between the PNL ratio as a function of co mpressed breast tissue thickness for CC and MLO views (CC, p = 0.509, MLO p =0.382), Table 6-12. However, the difference in median PNL ratio s for the CC and MLO views is statistically significant, (p<0.001, MW). The PNL ratio s for CC and MLO views are 0.45 Â± 0.13, and 0.51 Â± 0.14, respectively, Table 6-12. Instead of using two values, a recommended value of 0.50 Â± 0.08 for both CC and MLO views woul d ensure that at least 54% of the population would be within the value, taking into account the width of the AEC sensor. The clinical application of th e PNL ratio is that the techno logist will use the product of the PNL measurement and the PNL ratio to se lect the appropriate position of the AEC sensor that is most likely to be located under the dense portion of the breast. The direct clinical impact of the PNL ratio is the possi ble reduction in repeat films for new patients that have resulted from improper placement of ionization chamber by the technologist. The clinical application of the PNL ratio is that the technologist will use the PNL
169 measurement then select the appropriate positi on of the AEC sensor that is likely to be located under the dense portion of the breast. Conclusions The main thrust of the retrospective study was to define the average breast as imaged in CC and MLO views by mammography for a clinical populat ion, the results of which are shown in table 6-10 and table 6-11. The data provide the anthropometric data needed to develop realistic phantoms. In addition, the data sugg est that the current MQSA/ACR dosimetry model lin ear dimensions are smaller than this study population. This study did not take into account several f actors that may influence the anthropometric measurements of a population, such as race, menopausal status and body index measurements. The anthropometric measurements of this study population clearly showed that adipose tissue increases with age and compre ssed breast thickness. The study presented evidence that CC and MLO are two distinct populations. This makes it important to distinguish the two views in population studies, unlike previous studies that combined the views as a single aggregate group.29,30,60,81,119,128 In this study, the CC and MLO views constituted two statis tically separate groups in compre ssion pressure, compressed breast thickness, tube potential and tube current-time product. The underlying reason for this is the anatomical inclusion or exclusion of the pectoral mu scle in CC and MLO views. Equations 6-3 and 6-4 were derived in or der to estimate the whole breast areas in CC and MLO views based on rudimentary m easurements of PNL, width, CCD, and axilla. These equations facilitate the deve lopment of anatomica lly correct mammography phantoms.
170 The estimated measurement of the skin la yer indicates that current models that assume a 0.4 cm skin layer need to be re vised. The smaller skin layer would result in lower glandular doses with current dose conversion factors. In addition, we were also able to de termine the optimal position (0.45 of PNL) for the AEC detector in respect to the dense tissue for those pati ents which previous films do not exist or could not be accessed. The use of the PNL ratio would directly contribute to reducing retake rates for new patients. The study clearly demonstrated the pote ntial benefit of characterizing patient populations to define anthropometric paramete rs for the manufactur ing of anatomicallycorrect phantoms.
171 Mammogram CC view Breast Edge Uniform Dense PNLW i d t h Mammogram MLO view Pectoral Major (PecMaj) AxillaP e c M a j DWholeP N L C C D Figure 6-1 CC and MLO segmentation regi ons and lengths used in this study.
172 Breast Edge X=1740.72 X=888.71 RCC P atient Information Unit Parameters 888.7 PNL Distance (mm) 0.0 123.8 1933.9 P i x e l I n t e n s i t y(A) (B)C h e s t w a l l N i p p l eT=1314.7v(C) Figure 6-2. (A) Breast edge region illustra tion in the geometry of a clinical mammography unit. (B) Image illustration in a CC view of the breast edge region and the rectangular region used to evaluate threshold values. (C) CC f average pixel intensity graph of the do tted rectangular regi on shown in (B) as a function of distance from the chest wall. The threshold value was then derived using X1 and X2 values in equation (6-1).
173 Mammogram CC view PNLW i d t h Mammogram MLO view AxillaC C D Breast Geometry Model Figure 6-3. Breast geometry model to determine CC and MLO whole area.
174 Mammogram CC viewRCC Patient Information Unit Parameters Mammogram MLO viewRMLO Patient Information Unit Parameters X =Breast Edge width Y Breast Edge Width Model Horizontal mid-plane Figure 6-4. Breast edge width model to es timate the adipose region thickness .
175 Age Group (years) 3031 4041 5051 6061 7071 80> 80Frequency (%) 0 10 20 30 40 Figure 6-5. Distribution of age groups determined from the study population.
176 BI-RADS Density Categories <25%25-50%51-75%>75%Frequency (%) 0 10 20 30 40 50 60 Current Study Lewin et al., 2001 Wang et al., 2003 Venta et al., 2001 Figure 6-6. BI-RADS density cat egory distribution for our st udy compared to Lewin et al., Wang et al., and Venta et al.,.119,125,126
177 Compression Pressure (daN) 2468101214161820Frequency (%) 0 5 10 15 20 25 30 CC MLO Figure 6-7 Distribution of CC and MLO compression pressu re used on patients during imaging.
178 Compressed Breast Thickness (mm) 0102030405060708090100110120Frequency 0 20 40 60 80 100 120 140 160 180 CC MLO Figure 6-8. Distribution of patient compressed breast th ickness for CC and MLO views under compression.
179 Tube Potential (kVp) 252627282930313233Frequency (%) 0 10 20 30 40 50 CC Mo/Rh MLO Mo/Rh MLO Mo/Mo CC Mo/Mo Figure 6-9. Study population tube potential distribution for CC and MLO views using Mo/Mo and Mo/Rh target-filter combinations.
180 Tube Current-Time Product (mAs) 50 75 100 125 150 200 225 250 300 325 350 400 425 450 475 500 Frequency (%) 0 5 10 15 20 25 30 35 CC MLO Figure 6-10. Distribution of tube currenttime product for CC and MLO views from 1040 mammography images.
181 Predicted CCarea (cm2) 0100200300400500600Measured CC whole area (cm2) 0 100 200 300 400 500 600 Correlation of predicted area to measured area One-to-one correlation Figure 6-11. The graph correlates the measured CC view whole area to the predicted area defined by width PNL CCArea 4 9494 . 0 , R2=0.97.
182 Predicted MLOarea (cm2) 0100200300400500600Measured MLO whole area (cm2) 0 100 200 300 400 500 600 Correlation of predicted area to measured area One-to-one correlation Figure 6.12. The graph correlates the measured MLO view whole area to the predicted area defined by CCD axilla PNL width MLOarea 2 1 4 6558 . 0 , R2=0.90.
183 Bedge Edge Region Width (cm) 0.100.150.200.250.300.350.40Frequency (%) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 MLO CC Figure 6-13. Distribution of the calculate d breast-edge region width for MLO and CC views using equation 6-5.
184 Table 6-1. Data collected in the retrospective study. Collected Data Descriptor Characteristics BI-RADSÂ® Assessment Category Cat 1 or Cat 2 (only) Note: Assigned by interpreting radiologist BI-RADSÂ® Density Category <25%, 25-50%, 51-75%, >75% Note: Assigned by interpreting radiologist Mammography Technologist Code Anonymity maintained in study Interpreting Radiologist Code Anonymity maintained in study Individual Age If 89 years old of greater assigned " 89" Hormone replacement status Yes or No Radiographic Positioning Craniocaudal (CC) and mediolateral oblique (MLO) views, le ft (L), right (R) Codes: LCC, LMLO, RCC, RMLO Radiographic Unit GE-800T or GE-DMR Compressed Breast Thickness Distance in millimeters (mm) Compression Pressure Pressure in decaNewtons (daN) Radiographic Unit Operating Mode AOP-S, AOP-C, AOP-D, AEC Tube Potential kVp Tube Current-Time Product mAs Focalspot size 0.1 mm or 0.3 mm Target Mo or Rh Filter Mo or Rh Positioning Angle Degrees
185Table 6-2. BI-RADS density categories used for population analysis. BI-RADS Category Description Glandularity Range Mean Glandularity Assigned (%) 1 Almost entirely fat <25% fibroglandular 12.5 2 Scattered fibroglandular dens ities 25-50% fibroglandular 37.5 3 Heterogeneously dense 51-75% fibroglandular 63.0 4 Extremely Dense > 75% fibroglandular 87.5
186Table 6-3. Interpreting ra diologist assigned BI-RADSÂ® density categories as a function of individual interpreting radiologists. BI-RADS Density Categories Interpreting Radiologist <25% 25-50% 51-75% >75% Subjects N 001 0.00% 50.00% 50.00% 0.00% 27 002 5.26% 57.89% 34.21% 2.63% 38 004 21.74% 69.57% 8.70% 0.00% 23 005 8.77% 68.42% 17.54% 5.26% 57 006 8.00% 36.00% 48.00% 8.00% 25 007 12.20% 51.22% 24.39% 12.20% 41 008 9.09% 54.55% 18.18% 18.18% 11 009 3.33% 23.33% 63.33% 10.00% 30 Mean 8.37% 52.99% 32.27% 6.37% 251
187Table 6-4. Hormone replacement therapy BI-RADS density category comparison to comparable aged subjects. Mean Age Age Range Glandularity (%) n HRT status 59.02 Â± 10.41 39 80 45.00 Â± 19.70 51 Yes 57.32 Â± 11.43 39 80 47.50 Â± 17.30 183 No
188 Table 6-5. Mean BI-RADS density as a f unction of compression thickness for CC and MLO views. Thickness (cm) Glandularity (% Â± ) Mammograms <3 50.09 Â± 14.14 73 3-5 48.07 Â± 18.54 476 5-7 45.86 Â± 18.03 420 >7 38.35 Â± 17.45 63
189 Table 6-6. Mean BI-RADS density as a function of age group. Age (years) Glandularity (% Â± ) Mammograms 30-39 51.84 Â± 23.86 53 40-49 51.28 Â± 17.54 364 50-59 47.35 Â± 20.04 274 60-69 40.88 Â± 12.79 229 70-79 42.21 Â± 18.24 76 >80 42.22 Â± 18.31 44
190 Table 6-7. Technologist applied compression pressure as a function of individual technologists. The mean value did not include the contributions of technologist 004 or 005 since both only had one patient. Technologist Pressure (daN Â± ) Mammograms 000 10.47 Â± 4.54 32 001 10.48 Â± 3.58 616 002 12.42 Â± 2.77 277 003 11.96 Â± 2.99 103 004 8.25 Â± 4.11 4 005 10.00 Â± 4.32 4 Mean 11.33 Â± 1.77 1032
191 Table 6-8. Technologist applie d compression pressure as a function of compressed breast thickness. Compression Thickness Pressure (daN) Mammograms < 3 cm 8.80 Â± 3.45 73 3 5 cm 10.56 Â± 3.03 515 5 7 cm 11.98 Â± 3.65 385 > 7 cm 13.32 Â± 3.27 59 Mean 11.13 Â± 3.48 1040
192Table 6-9. Mammography unit settings used for images inCC a nd MLO views as a function of target-filter combination. Position Target/Filter Compressed Breast Thickness (cm) Pressure (daN Â± ) Pressure Range (daN) Tube Potential (kVp Â± ) Tube Potential Range (kVp) Tube CurrentTime Product (mAs Â± ) Tube CurrentTime Product Range (mAs) n CC Mo/Mo 3.89 Â± 1.10 9.92 Â± 3.54 3 42 27.55 Â± 0.85 26 30 97.12 Â± 34.83 8 280 274 CC Mo/Rh 5.11 Â± 0.91 9.96 Â± 3.12 3 19 28.16 Â± 1.84 26 32 139.43 Â± 42.67 77 281 240 CC Rh/Rh 6.9 Â± 0.00 11.50 Â± 0.71 11 12 31.00 Â± 0.00 31 31 142.00 Â± 14.14 132 152 2 MLO Mo/Mo 4.28 Â±1.30 11.41 Â± 3.08 3 18 27.57 Â± 0.85 26 31 113.38 Â± 61.31 8 328 175 MLO Mo/Rh 5.64 Â±1.17 12.76 Â± 3.18 3 19 29.05 Â± 2.13 26 32 158.33 Â± 55.37 79 481 345 MLO Rh/Rh 8.0 Â±1.63 12.25 Â± 1.89 11 15 29.50 Â± 0.58 29 30 206.75 Â± 78.26 132 298 4
193Table 6-10. Study population CC breast measurements as a function of compressed breast thickness. Compressed Breast Thickness Range Measurement < 3 cm 3 5 cm 5 7 cm > 7 cm Mean Thickness (cm) 2.23 Â± 0.70 4.06 Â± 0.55 5.61 Â± 0.50 7.37 Â± 0.10 4.46 Â± 1.17 PNL (cm) 8.44 Â± 3.16 9.89 Â± 2.71 11.38 Â± 2.96 11.73 Â± 1.45 10.29 Â± 2.96 Width (cm) 18.55 Â± 3.55 20.74 Â± 2.65 22.38 Â± 2.94 24.05 Â± 2.79 21.14 Â± 3.05 Bedge (cm2) 19.41 Â± 11.42 27.79 Â± 10.91 39.5 Â± 14.85 55.78 Â± 8.72 31.31 Â± 14.21 Dense (cm2) 39.01 Â± 27.51 44.17 Â± 28.68 28.68 Â± 14.85 69.04 Â± 50.01 47.38 Â± 35.52 Uniform (cm2) 103.1 Â± 59.44 125.94 Â± 53.63 155.37 Â± 66.56 156.53 Â± 27.77 134.17 Â± 60.78 Whole (cm2) 122.51 Â± 67.52 153.73 Â± 60.68 194.83 Â± 77.27 212.62 Â± 34.06 165.48 Â± 70.93 Pressure (daN) 8.45 Â± 2.84 9.74 Â± 2.85 10.62 Â± 3.86 8.83 Â± 3.43 9.91 Â± 3.28 kVp 27.14 Â± 0.77 27.76 Â± 1.01 28.13 Â± 2.00 28.33 Â± 0.82 27.84 Â± 1.43 mAs 65.73 Â± 21.36 101.93 Â± 23.39 150.09 Â± 42.77 214.17 Â± 22.75 116.16 Â± 42.48 Glandularity (%) 49.11 Â± 14.96 46.85 Â± 18.36 46.12 Â± 19.04 37.5 Â± 0.00 46.69 Â± 18.23 N 44 283 168 6 503
194Table 6-11. Study population MLO breast measurements as a function of compressed breast thickness. Compressed Breast Thickness Range Measurements < 3 cm 3 5 cm 5 7 cm > 7 cm Mean Thickness (cm) 2.34 Â± 0.41 4.16 Â± 0.53 5.76 Â± 0.57 7.84 Â± 0.58 5.22 Â± 1.39 Axilla (cm) 6.23 Â± 1.41 7.46 Â± 1.43 8.15 Â± 3.14 8.69 Â± 1.53 7.87 Â± 2.48 CCD (cm) 20.37 Â± 1.42 21.66 Â± 2.18 22.94 Â± 7.9 23.58 Â± 2.32 22.41 Â± 5.76 PNL (cm) 8.72 Â± 3.31 10.09 Â± 2.45 12.49 Â± 4.71 15.27 Â± 3.90 11.71 Â± 4.23 PecMajD (cm) 12.83 Â± 2.44 13.83 Â± 2.38 14.57 Â± 5.59 13.45 Â± 4.02 14.1 Â± 4.40 Dense (cm2) 32.97 Â± 17.01 26.34 Â± 15.61 32.93 Â± 33.7 27.39 Â± 29.62 29.97 Â± 27.39 PecMaj (cm2) 35.92 Â± 12.09 38.61 Â± 12.52 38.5 Â± 14.11 34.74 Â± 16.88 38.05 Â± 13.82 Bedge (cm2) 23.67 Â± 10.61 38.56 Â± 9.82 50.12 Â± 12.67 69.7 Â± 24.05 46.66 Â± 16.87 Uniform (cm2) 87.23 Â± 45.6 111.98 Â± 47.80 148.92 Â± 53.95 197.87 Â± 83.26 137.65 Â± 61.95 Whole (cm2) 111.48 Â± 49.55 150.17 Â± 50.54 199.49 Â± 61.27 268.09 Â± 104.56 184.47 Â± 74.01 Pressure (daN) 10.18 Â± 3.35 11.67 Â± 2.85 12.85 Â± 3.01 13.63 Â± 3.12 12.38 Â± 3.06 kVp 27.41 Â± 0.73 27.81 Â± 1.08 28.86 Â± 2.22 30.24 Â± 1.50 28.53 Â± 1.90 mAs 72.68 Â± 16.18 107.71 Â± 27.35 151.66 Â± 42.44 249.43 Â± 62.98 141.56 Â± 58.47 Glandularity (%) 53.73 Â± 12.56 45.21 Â± 17.05 45.21 Â± 17.05 38.04 Â± 18.43 46.66 Â± 18.03 N 22 183 238 51 494
195 Table 6-12. The CC and MLO views PNL ratio calculated as a function of compressed breast thickness. CC MLO Compressed Thickness PNL Ratio Images PNL Ratio Images <3 cm 0.47 Â± 0.12 44 0.49 Â± 0.12 22 3-5 cm 0.45 Â± 0.14 282 0.50 Â± 0.14 183 5-7 cm 0.45 Â± 0.14 169 0.52 Â± 0.14 237 >7 cm 0.40 Â± 0.07 6 0.51 Â± 0.13 48 Mean 0.45 Â± 0.13 501 0.51 Â± 0.13 490
196 CHAPTER 7 ESTIMATING BREAST GLANDULARITY Introduction Breast parenchymal patterns, as depicted in mammography images, can be used to provide the probability of detecting an abnorma lity. As the parenchymal patterns become dense and complex, the ability or probability of the radiologist to detect an abnormality amongst the parenchymal pattern may be reduced . This study evalua ted the possibility that these patterns can also allow the calculation of the gl andular dose resulting from the mammography imaging. Breast parenchymal patterns, tissue density, composition, or glandularity are terms commonly used to descri be the fibroglandular content of a breast. Within the envelope of adipose tissue, fibr oglandular tissue is composed of glandular tissue, fibrous tissue, ligaments and thei r associated vasculature and enervation. Fibroglandular tissue is t ypically sequestered in 15 to 20 lobes. Each lobe is composed of lobules, whose functional unit for producing la ctate is called termin al duct lobular unit (TDLU). The terminal duct leading into th e grape-cluster like st ructures of TDLU, granular acini, is believed to be one of the sites for carcinogenesis.13 Johns and Yaffe showed that, radiographically, cancerous ti ssues attenuate x rays more than normal glandular tissues, and glandul ar tissues attenuate x rays more than adipose tissues.49 Medical professionals use th e parenchymal patterns in ma mmography images to estimate the distribution of adipose and fibrogl andular tissues within the breast. Radiologist and medical physicists are two groups of medical professionals who routinely use breast tissue gl andularity in their profession. The radiologist estimates
197 breast tissue glandularity qualitatively to determine the pr obability of dete cting a cancer or other breast abnormalities from the ma mmogram. The parenchymal patterns in a mammogram have been associated with an increase risk of breast cancer by population studies that have correlated can cer incidence with breast density.32-45 The medical physicist estimates breast tis sue glandularity quantitativ ely to determine average glandular dose (AGD). AGD is calculated usi ng a dose conversion fact or that correlates the breast entrance skin exposure to glandul ar tissue dose. The dose conversion factor varies with compressed breast thickness, breast tissue glandularity, ha lf-value layer, and tube potential. Over the year s, both groups of professionals have attempted to expedite the process of estimating glandularity. The only accurate method to quantify gla ndular tissue content in a breast is by biopsy. Biopsy ensures a quantifiable measure, but it is not practical , and a small tissues sample is not representative of the w hole breast. Mammography images provide a practical alternative for estimating the glandul ar tissue distribution of a breast. Numerous methods have been devised to ca tegorize parenchymal patterns. ACR BI-RADS Method Currently, the American College of Radiology (ACR) Breast Imaging Reporting and Data System (BI-RADS) provides a standardized mammography-reporting terminology for breast density. BI-RADS categor izes breast densities into four density ranges11; Â• Almost entirely fat (<25% fibroglandular); Â• Scattered fibroglandular dens ities (25-50% fibroglandular); Â• Heterogeneously dense ( 51-75% fibroglandular); Â• Extremely dense (>75% fibroglandular).
198 The interpreting radiologist reviews th e four images made during a typical mammography imaging session ( L, RCC, L, RM LO) and any previous images and then assigns the ACR BI-RADS density category to the individual patient. The BI-RADS categories were devised to assess qualita tively the sensitivity of the mammography image. Sensitivity is defined as the probability of detecting a cancer in a patient that has a cancer lesion.11 The BI-RADS category becomes part of the patient record, but it does not constitute a quantitative measurement of density by the interpreting radiologist. Given that glandularity is a biological measurem ent it would suggest that the distribution amongst categories should follow a normal di stribution. Ferris Hall, a physician at Harvard Medical School, suggested to his re sidents that the distribution of BI-RADS categories should be approximate ly 20% almost entirely fa tty, 30% scattered glandular tissue, 30% heterogeneously dense and 20% extremely dense.129 Recent population studies data have shown a wide range of distributions within each category, 4-43% almost entirely fatty, 27-46% scattered glandul ar tissue, 12-42% he terogeneously dense, and 2-18% extremely dense.119,125,130 This study found the distribution within each category to be 0-22% almost entirely fatt y, 23-68% scattered glandular tissue, 9-63% heterogeneously dense, and 0-18% extremely de nse, as described in Chapter 6. The BIRADS methodology is highly dependent on the interpreting radiologist training, professional experience and the f acility standard-of-care protocol s. As is shown in this study, an individual radiologist may have a wide range in the distributi on of densities, but group of radiologists interpreti ng a population of images ma y be close to the typical distribution suggested by Hall. The A CR BI-RADS method was never designed to quantify breast glandularity and has been succe ssful in providing interpreting radiologists
199 with a qualitative method for determining sens itivity. Application of this method to quantify breast density has inherent systemic problems of the interpreting radiologist's training, experience bias and cat egory assignment. The systemic issue is the systematic method of assigning an ACR BI-RADS category to an individual instead of by a breast. To address these criticisms, a planimetry method (PM) was introduced to quantify breast density based on the areas of dense tissue w ith respect to the whole breast area. Planimetry Method The simplest approach to quantify gla ndular tissue in a mammography image is to measure the area of the glandularity and divi de it by the whole breas t area. Wolfe and colleagues in 1976 used a computer-assisted planimetry method to measure breast density. They used an acetate overlay with a trace of the breast outline and parenchymal regions to identify the dense areas.36 Modern techniques have digitized the mammography images and then digitally se gmented the whole and the dense regions.46 The planimetry method is a binary system which treats breast tissues as either 100% adipose or 100% glandular, ignoring the complex nature of the three-dimensional geometry of these tissues. The advantage to th is system is that each breast is individually assessed allowing for biological variations between breasts. The two primary criticisms of this method arise from the binary deci sion of selecting tissues and the potential inclusion of adipose tissue with in the dense regions, both of wh ich affect the final breast density value. To address these criticis ms, a histogram threshold method (HTM) was introduced, which evaluates mammog rams by a pixel-by-pixel methodology. Histogram Threshold Method In 1994, Byng and colleagues developed an interactive threshol ding technique in which the observer selects two threshold pixel intensities.131,132 The first threshold pixel
200 intensity is used to segment th e whole breast. The second thre shold pixel value is used as to identify the edge of the dense tissue area. Using the mammography image pixelintensity histogram, the pixels exceeding the second threshold value are classified as glandular and included in dense area calcu lation. Glandular content is then the proportion of dense tissue area to whole breast area. The advantage to this method is that it reduces the bias of the classifier. Th e primary criticisms of this method arise from the binary decision of tissue assignment. To address the criticism of using a binary decision of tissue assignments, a method wa s introduced that utilized the mammography automatic exposure control (AEC) system to evaluate the percent glandularity. Tube Loading Method Geise and Palchevsky developed a method that does not rely on the mammography image but rather th e mammographic unit parameters.29 Geise and Palchevsky approached the problem of estim ating glandularity by using the AEC system of the mammography unit, thus eliminating the observer decision from the method. The authors reasoned that the ma mmographic unit's AEC system made a unique exposure for a breast based on compressed breast tissue thickness, glandula rity and scattering characteristics. A least-squa re fitted function was determin ed that correlated breast tissue glandularity to compressed breast thickness (tb), tube potential (k Vp) and tube-current time product (mAs). The system was calibrated using a tissue-equivale nt breast phantom, with glandularity rangin g from 0-100% glandular. The adva ntage of the system is that it eliminated the binary decision methods of the PM and HTM. The main criticism of this method is that the breast tissue area evaluate d by the AEC sensor is small in comparison to the whole breast. Another criticism is that the premise that each exposure is unique is not quite true because I found as part of th is study that the mammography units used in
201 this have an apparent default values for kVp and mAs for very thin breasts (< 2 cm) independent of breast tissue composition. I beli eve that the default value is a result of calibrating units that utilize a pre-exposure with acrylic sheets. Volumetric Method In 2003, Pawluczyk and colleagues devel oped a film-screen digitized method in which the imaging system was calibrated by using a tissue-equivalent step-wedge of breast densities.47 The authors used least squares to fit a three-dimensional plane function to compressed breast thickness, percent gla ndularity, and optical density. Based on the three-dimensional plane, the digitized image was evaluated pixel by pi xel, and a resultant aggregated breast density was determined. Th e advantage of the method was that it took into account the three-dimensional nature of the breast tissues. All the aforementioned methods present as reasonable attempts to quantify the glandular tissue content of a breast. Unfortunately, to date, no direct comparison has been made among these estimation met hods on a single clinical population. This study introduces a breast-tissue e quivalent modified phantom series thresholding method (BRTES-MOD TET) for es timating glandularity and compares it to ACR BI-RADS, PM, HTM, and TLM for a si ngle clinical mammography population. In addition, this study investigates how the three-di mensional nature of breast tissues affects the resulting radiographic image usi ng a simple Monte Carlo model. Method Precepts The BRTES-MOD TET method is introduced in this research as a glandularity referencing method for estimating patient gl andularity. The method is based on image characteristics of the BRTES-MOD phantom series. BRTES-MOD TET method uses a BRTES-MOD step phantom, described in chap ter 5, to establish a calibration curve for
202 glandularity to evaluate each mammogram's uniform region, de scribed in Chapter 6, on a pixel-by-pixel basis. The BRTES-MOD TET mo del is based on the generally-accepted premise in mammography that fibroglandular ti ssue attenuates x rays more than adipose tissue. The premise is based on a simple attenuation model in which the fluence ( ) of an impinging x-ray beam is modified by the attenuating proper ties of the breast.115 The attenuating properties of the breast result fr om the adipose and fibroglandular tissues (Âµa, Âµg; total linear attenuation coefficients for ad ipose and fibroglandular tissues) and their corresponding thicknesses (G , fraction associated with fibroglandular tissue; tb, compressed breast thickness). Equation (7 -1) shows the simplified model, ignoring scatter photons (that will be disc ussed later in this chapter.) b g at G Ge E E ) 1 ( 0 1 (7-1) The mammography spectrum is made up of polyenergic x rays. Therefore, glandularity cannot be easily calculated but can be estimated by using the BRTES-MOD step phantom. The BRTES-MOD step phant om has a homogeneous distribution of glandular and adipose tissue which ranges from 0 100% glandular tissue in onecentimeter incrementing steps as illustrated in figure 7-1 Using the phantom, a relationship between optical density of film and attenuated fluence can be developed in the following ma nner. The image detector transmittance (T) is a quantity that measures the level of image detector opacity and is proportional to the exposure received by that detector modified by the imaging system efficiencies. The image detector used in this study in ma mmography film and ther efore the detector opacity is the image optical density. The e fficiencies inherent in using a film-screen combination are quantum detection efficiency (QDE) and conversion efficiency (CE).
203 Quantum detection efficiency (QDE) is the frac tion of incident x-ray photons that interact with the screen.31 Conversion efficiency (CE) is th e ability of the sc reen to convert energy deposited onto the sc reen into film darkening.31 Digitizer efficiency (DE) is the ability of the film digitizer to translated film optical density to pixel intensity (PI). In mammography, the scatter-to-primary ratio (SPR ) remains relatively constant with kVp and increases with the geometric dimensions of the breast. 133,134 Optical density (OD) is related to the transmittance and SPR as shown in Equation (7-2 through 7-4). dE E E SPR E CE E QDE e E TkeV t G Gb g a 0 0 ) 1 ( 0)) ( 1 ( (7-2) T OD log (7-3) DE T PI log (7-4) Equation (7-4) shows that pixel intensity can be used as a surrogate for glandularity. Materials and Methods Study Population A three-month retrospective demographic study was conducted that compiled mammography data acquired from 253 patients seen at a U.S. Navy health care facility in Jacksonville, Florida, as described in Ch apter 6. One thousand and forty mammograms taken in the period from Ja nuary through March 2005 were reviewed and digitized for this study. The study population consisted of women that were seen for screening mammography or because of a primary care physic ian referral for a definitive diagnosis. Population Study Selection Criteria The patients that were selected for the study had an ACR BI-RADS assessment category of 1 (Negative) or 2 (Benign Fi ndings) assigned to them by the interpreting radiologist after diagnosis.11 Overall, BI-RADS assessment category 1 and 2 assessments
204 are indicative of a normal breast with no ev idence of malignancies. In evaluating methods to estimate breast glandularity, a subgroup of approximately 143 patients (CC view only) were selected that were assigned a BI-RADS assessment category 1. The selection of the BI-RADS a ssessment category 1 patients re duced potential influence from benign conditions on the inter comp arison of breast glandularity estimation methods. Mammography Imaging The typical mammography examination cons isted of four images made in two anatomical views, craniocaudal (CC) and medi olateral oblique (MLO), of the right and left breasts. The mammography units used to image patients utilize an automatic optimization parameter modality that optimizes unit parameters, by selecting the target and filter combination (T/F), t ube potential (kVp), and tube current-time product (mAs). A General Electric Senographe-DMR (G E-DMR) or a GE Senographe-800T (GE800T) provided the mammography x-ray fields used throughout the study period. The Senographe-DMR uses a single-pha se high-frequency generator and an x-ray tube, with a dual bi-metal track target of molybdenum (Mo) and rhodium (Rh). The x-ray spectra were filtered using Mo or Rh filters. The unit had a fixed source-to-image receptor distance of 660 mm and a focal spot of 0.3 mm. The Senographe-800T uses a highfrequency generator and an x-ray tube, with a single metal track target of Mo. The x-ray spectra were filtered using Mo or Rh filters . Both the GE-DMR and GE-800T utilize a focal spot of 0.3 mm in routine screening imaging. The GE-DMR and GE-800T mammographic units used an automatic optimi zation parameter (AOP) mode in clinical imaging which prioritizes to dose reducti on (AOP-D), contrast quality (AOP-C) or compromise between dose reduction and c ontrast quality (AOP-S) by selecting kVp,
205 mAs, T/F thru the use of a pre-exposure wh ich access the algorith m using unit parameter and the radiographic thickness of the breast. The radiation emerging from the breast passes through the patient support table, the antiscatter grid, the cassette front, film base before being absorbed by the in tensifying screen and film emulsion. Mammography Facility and Film Processing As a U.S. Food and Drug Administrati on certified and American College of Radiology (ACR) accredited facility, the studie d facility maintain its operations in accordance with the Mammography Quality Stan dards Act (MQSA) regulations and the ACR Mammography Accreditation Program (ACR-MAP) guidelines as specified.12,100 Throughout the study, the facility performed th e required daily quality control activities and ensured that the prescribed control limits were met prior to processing any patient's films as is required by MQSA and ACR-MAP.12 The Kodak Min-R 2000 scr een-film combination was used throughout the study. The film was developed using a Kodak XOM AT 5000 RA film processor. Processor quality control was performed prior to patient's films being developed in accordance with MQSA and ACR-MAP. Film Digitizer The mammograms along with the daily pr ocessor control film s were digitized using a Kodak LS-75 film digitizer with a limiting resolution of 5 line pairs per mm.121 Mammograms were digitized at a resolution width of 2048 pixels with the number of lines determined by the film size. The matrix size for 18x24 cm2 film was 2048 (0.009 cm) Â± 0 x 2811.85 (0.009 cm) Â± 114.58 pixels and 24x30 cm2 was 2048 (0.012 cm) Â± 0 x 2628.20 (0.011 cm) Â± 21.20. All digitized mammograms were stored for image processing in a Digital Imaging and Comm unications in Medicine (DICOM) format.
206 Image Segmentation The data collected for each mammogram ar e shown in Table 7-1. The data were compiled from each mammogram using paper fo rms that were subsequently transferred to a customized database program designed for this study22. The statistical analysis performed on the population data wa s done using SigmaStat (version 3.1)23. The two main statistical tests utilized for the study were the Mann-Whitney (MW) rank sum test and the Kruskal-Wallis (KW) one-way analysis of variance on ranks non-parametric statistical tests were utilized to evalua te differences between two or more groups. 135 Digitization of mammograms and daily pr ocessor control film s was done using a Kodak24) LS-75 film digitizer with a limiting re solution of 5 line pair per mm (0.1mm).121 All mammograms were digitized at a resoluti on width of 2048 pixels with the number of lines determined by the film size. The matrix sizes for the 18x24 cm2 and 24x30 cm2 were 2048 x 2812 and 2048 x 2628 pixels, respectively. The pixel size was approximately 0.1 mm x 0.1 mm, corresponding to th e limiting resolution of the scanner. All digitized mammograms were archived in a Digital Imaging and Communications in Medicine (DICOM) format for processing. Manual segmentation was performed using ImageJ25 (version 1.34i), image analysis software developed by the National Institutes of Health.122 Each CC view and MLO view of each mammogram was manually segmented into the regions and segments illustrated in figu re 7-2 and described in Chapter 6. In this 22 Microsoft Office Access 2003 , Microsoft, Redmond, WA 23 SigmaStat 3.1, Systat Software, Inc., Point Richmond, Ca 24 Eastman Kodak Company, Rochester, NY 25 ImageJ, National Institutes of Health, Bethesda, MD
207 research only CC views were used to estimat e patient glandularity and for the comparison between glandularity estimating methods. Monte Carlo Modeling Monte Carlo (MC) simulation was utilized to investigate whet her the effects of tissue layering and surrounding tis sues alter the OD of the mammographic image. The Monte Carlo Neutral Particle Code vers ion 5 (MCNP-5) developed by the U.S. Department of Energy was utilized in c onjunction with poly-energetic mammography spectra generated from the In stitute of Physics and Engin eering in Medicine Report 78 (IPEM-78) spectrum genera ting software (SRS-78).10,136 MCNP-5 accounts for coherent scattering, incoherent scattering, photoelectr ic effect absorption with florescence emission and pair production.10 The elemental composition of breast tissues (glandular, adipose) used with MCNP-5 were describe d in International Co mmission on Radiation Units and Measurements (ICRU) Report 44 (ICRU -44) and mean values in ICRU Report 46 (ICRU-46).107,114 Phantom specific glandular cont ent was derived using weight fractions of the mean tissues in ICRU-46. The geometry utilized in this study is shown in figure 7-3(A); spectra details are shown in Table 7-2. The compressed breast phantom was a 2 cm thick ellipse half with a minor half-axis of 8 cm and major axis of 18 cm, figure 7-3(B). The phantom contains a column of tissue made up of 1 cm x 1 cm x 0.1 cm tissue slices th at are either a 100% adipose or 100% glandular tissue. The tissue column is surrounded by either 100% adipose or 100% glandular tissue. The se quences of 100% adipose or 100% glandular tissues are listed in figure 7-3(C-1 through C-4) was intended to be a simplified model of the tissue layering that may be f ound near the lobules of the br east. The last layer in the column is composed of air a nd is intended to measure the exit dose from the column. A
208 change in dose in this layer would resu lt in a corresponding change in OD. Three separate MC runs were performed, tracing 100 million photon histories resulting in exit dose with 1% relative standard error ( x). The MCNP-5 spectrum was based on the spectra used to image a BRTES-MOD phantom of the same thickness and clinical geometry. The beam quality of the clinical spectrum were measured using narrow-beam geometry. Each MCNP-5 spectrum HVL was modified to match the clinical parameters by filtering the raw spectra generated from SRS-78 with beryllium (Be), molybdenum (Mo), aluminum (Al) and polymethyl methacrylate (acrylic), filter thickness are s hown in Table 7-2. The acrylic was utilized to simulate the compression paddle in the beam path. Once the correct HVL was determined using the filtered SRS-78 spectrum, the quality of the spectrum was verified using the MCNP-5 geometry with simulated Al filters. The MCNP-5 spectrum had an HVL within a half of a percent of the clini cally measured HVL as shown in Table 7-2. Breast Tissue Composit ion Estimating Methods Five methods of estimating glandularity (ACR BI-RADS, planimetry, histogram, tube loading and BRTES-MOD tissue equivale nt thresholding (TET)) were compared, based on the craniocaudal view of th e study population mammography images. ACR BI-RADS method The first method, the ACR BI-RADS method, was based the interpreting radiologist ACR BI-RADS dens ity categories assignment. The interpreting radiologist reviewed all four images taken during th e imaging session and any previous images before assigning the density category to th e individual patient. Th e retrospective study-
209 population density category was taken directly from the inte rpreting radiologist report filed along with the mammogram. Planimetry method In the planimetry method (PM), the digitized segmented mammograms were individually evaluated and th e resulting glandularity was ba sed on the ratio of the dense tissue area to the whole breas t area for each CC image. Histogram threshold method The histogram threshold method (HTM) us ed the histogram of the whole breast and evaluated each pixel for gla ndularity content. In order to determine the glandularity threshold for mammograms, the dense tissue regions of the left and right CC (LCC, RCC), and MLO (LMLO, RMLO) mammography im ages were evaluated separately for their mean pixel intensities. Th e mean threshold value ratio(PIRatio) is the mean pixel intensity of the dense region (PIDense) divided by the maximum pixel intensity of the image (4095)as shown in equation 75. The RCC and LCC population mean PIRatio was 0.64 0.05. 4095Dense ratioPI PI (7-5) Each RCC and LCC histogram was processed an d each pixel intensity that exceeded the dense tissue pixel intensity ( 0.64 = 2620.8) was identified as a dense tissue pixel. Breast glandularity was subsequently calculated by taking the ratio of the total pixel area corresponding to the glandular tissue (dense) to whole breast area. Tube Loading Method The tube loading method (TLM) was base d on the unit parameter response to a tissue-equivalent phantom series containing a range of compressed breast thicknesses ( tb)
210 and glandularities. The breast phantom used was the BRTES-MOD breast phantom, described in Chapter 5. Each of the co mpressed breast phantoms had adjustable compressed breast thickness from 1-8 cm and a posterior nipple line of 8 cm and a width of 18 cm composed of a homogeneous mixtur e of glandular and adipose tissue ranging from 16.2 67.8% glandular tissue. The GE-D MR and GE-800T were set up as though being used for clinical application when imaging the BRTES-MOD phantoms. The standard automatic optimizing parameter mode AOP-S, which was the typically-used modality at the clinic, was utilized to image BRTES-MOD phantom slices. The mammography unit parameter, ta rget/filter (T/F), kVp, mAs, tb and compression pressure (tP) were recorded for each BRTES-MOD phantom slice added. Subsequently, a correlation was derived betw een glandularity and unit parameters using Sigmaplot26 (version 9.0) which utilizes th e Marquardt-Levenberg algorith m (non-linear least squares) to derive the final regression equation parame ters iteratively. The general equation used in the regression algorith m was based on the work of Geise and Palchevsky.29 The mathematical form has no physical interpre tation. Breast glandularity was correlated to the unit parameters using th e equations (7-6), where GBRTES-MOD is the glandularity, tb is the compressed breast thickness, kVp is the tube potential, mAs is the tube current-time product and n has values that depend on th e T/F (Mo/Mo or Mo/Rh) combination. 2 2 0( lnb b n b b MOD BRTESt e t d c mAs kVp t b t a y G (7-6) BRTES-MOD tissue-equivalent threshold ethod The BRTES-MOD TET method uses a BRTES-MOD step phantom to determine a calibration curve relating th e reference glandularity of the mammogram to pixel 26 Sigmaplot version 9.0, Systat So ftware, Inc., Point Richmond, CA
211 intensity. The individual steps of the BR TES-MOD TET are illustrated in figure 7-4 and are discussed here. After reviewing the patient mammogram, the step phantom was centered on the chest wall as illustrated in figure 7-1(d). Using the same mammographic unit parameters of T/F and kVp, two step phant om images were taken that bracketed the mAs station used in the patient's mammogram as illustrated in fi gure 7-4. After having digitized the step phantom imag es, pixel intensities of two rows of equal thickness that bracketed the compressed breast thickness of the original mammogram in each previously taken step-phantom image were measured. The pixel intensities were measured at the center of each step using a region of interest of 75 x 75 pixels. A composite pixel intensity of th e step-phantom was derived us ing linear interpolation that matched the unit parameters and tb of the patient mammography image. A sensitometric adjustment was need ed because the mammography images and step phantom images were taken on differe nt days. An evaluation of the corresponding sensitometry was necessary to ensure that the optical range as a func tion of pixel intensity was similar to the step phantoms and orig inal mammogram. The subtle changes in sensitometry were within the control limits but needed to be adjusted otherwise, it would result in either over or under estimating glandularity. All ma mmogram-film and BRTESMOD step phantom daily sensitometry strips OD (21 steps) were measured using a densitometer. Then the measurement was re peated digitally using the digitized daily sensitometry strips with a pixe l region of interest (40 x 160 pixel) there by creating a correlation between OD to PI. A ratio (PIratio) of the mammogram-pixel intensity value (PImammo) to the step-phantom pixel intensity value (PIstep) of the same sensitometry step was calculated for all 21 steps of the optical st ep wedge using equation (7-7). Each step
212 of the optical step wedge increased optical density by 41% or 21/2. The form of equation (7-6) had no physical meaning and was used to obtain the best fitting parameters. The data were fitted to equation (7-8) for the two mammography units (GE-DMR and GE800T) using Sigmaplot27, (version 9.0) which utilizes the Marquardt-Levenberg algorithm (non-linear least square s) to iteratively derive th e final regression equation parameters. step mammo ratioPI PI PI (7-7) 4 4 3 3 2 2 1 0 step step step Step ratioPI B PI B PI B PI B B PI (7-8) Once the step phantom images were correct ed for sensitometry differences with the mammogram, the corrected pixel intensity of the step phantom corresponding to the mammogram compressed breast thickness were correlated to BRTES-MOD glandularity (Gpixel). The function was fitted using the nonlinear least-squares to equation (7-9), where the pixel intensity of each glandular ity corresponding to the compressed breast thickness (PIG ) and b and m are regression parameters derived for each mammogram. The form of the equation was used to provide the best fit to the data. ) ln( ) ln( ) ln( m b PI GG pixel (7-9) After having obtained a histogram of pi xel intensities for the uniform region, a pixel-intensity weight value was derived for each pixel intensity range using equation (710). i i i ifreq freq w (7-10) 27 Sigmaplot version 9.0, Systat So ftware, Inc., Point Richmond, CA
213 The final estimated glandularity (GBRTES-MOD) of the breast was cal culated for only those pixel intensity values that corresponding to the 0-100% glandular region of the uniform breast region using equation (7-11). i G i MOD BRTESm b PI w G ) ln( ) ln( ) ln( (7-11) No adjustments were made to the pixel intensities as a result of the heel effect. Measurements were made using narrow-beam geometry. The HVL measurement was made centered on chest wall and at 8 cm from th e chest wall to correspond to the size of the compressed breast and step phantoms. Both units showed no detectable difference in HVL at 8 cm from the chest wall when co mpared to the HVL measured centered and 4 cm from the chest wall. The effect of invers e square law was also evaluated and the ratio of exposures from the chest wall to 8 cm was 0.956. Results and Discussion Monte Carlo Modeling The Monte Carlo results shown in Table 73 indicate that the arrangement of the tissue layers does not influence the opacity of the film as dramatically as the surrounding tissue. Given the same pattern of tissue layering, changing the surrounding tissue can affect the exit dose from 10.6 to 12.9%. Whereas, given the same surrounding tissue, changing the tissue layering affects the exit dose only by 2.2 to 5.9%. The impact of this finding is that parenchymal patterns (op acities) seen on mammograms are only an estimate of the glandular tissu e content of the breast. The complexity of tissue layering may in fact mask the quantit y of the glandular tissue.
214 Breast Tissue Composit ion Estimating Methods In developing the rationale to evaluate the glandularity estim ation methods, several key factors were considered such as method reference, method reproducibility, method adherence to population tendencies, and demogr aphics. The method reference refers to a documented process or calibration phantom that can be used to estimate glandularity. The method reproducibility was estimated by evaluati ng the glandularity of the left and right breasts or bilateral symmetry of the method. The population tendency that was used in the evaluation was the increase of adipose tissu e content in the breast associated with increasing age or increasing compressed breast thickness.60,106,118,120 With regards to patient demographics, tw o histogram methods were utilized to display population distributions. The firs t method showed the general population distribution based on 10% incr easing increments of glandular tissue and the second method used the BI-RADS mean density cate gory values (12.50, 37.50, 63.0 and 87.50) used by interpreting radiologists to asse ss mammographic images. The glandularity distribution difference in median values for all five groups ar e statistically greater than would be expected by chance (p 0.001, KW), figures 7-5 and 7-6. However, within the group, the planimetry and histogram methods were indistinguishable from each other statistically (p=0.988, MW), as we re ACR BI-RADS, TM, and BRTES-MOD TET (p=0.268, KW). The general sta tistics of glandularity as show n in Table 7-4 and will be discussed within context of each glandularity estimation method. ACR BI-RADS Method The ACR BI-RADS method was based on a wr itten standard delineated in the ACR Breast Imaging Reporting and Data System book .11 Unfortunately, the standard lacks sufficient detail to prevent variation in inte rpretation of the categories by the radiologist.
215 The result of this variation is quite eviden t when data from indi vidual radiologist are viewed for their distribution of ACR BI-RADS density assignments as shown in Table 75. This type of variation has been demonstrated in previous studies.12,44,127 To combat this type of distribution varia tion requires the facility to es tablish strict criteria for the interpreting radiologist to follow. As an aggregate however, the ACR BI-RADS density category assigned by the radiol ogists, Table 7-4 was simila r to previously published data.60,120 The ACR BI-RADS method had the sec ond highest mean glandularity, with a median glandularity of 37.5%. In additi on, the method shows the population trend of increasing adipose tissue with age and tb as shown in figure 7-7 and figure 7-8. The one major advantage to this method is that anat omical features that could interfere with glandularity estimation is avoided because of human interaction. Another disadvantage to this method is that anatomical variation of parenchymal patterns in the left, and right breast are not taken into account, and therefor e should not be assigne d the same glandular value, Table 7-6. Planimetry Method The planimetry method was introduced to reduce variability s een in glandularity categorical methods such as ACR BI-RADS. The PM is not based on a standard but rather it relies on each reviewer's training a nd experience to define the dense tissue area. So in fact, the method traded one bias with another, for example, instead of interpreting delineated standards, the reviewer interprete d the image for dense tissue areas for which there are no standards. The variability in se lecting dense areas was seen in this study because each image was evaluated independent ly. Therefore, when evaluating left and right breasts one should see a small variation; however, th is method showed a difference of 10.92%, as shown in Table 7-6. The popula tion trend was not eviden t in PM as it was
216 seen in the ACR BI-RADS method. Another disadvantage to this method is that the dense area may incorporate small regions of adipose tissue, which would increase the glandular fractional area as determined from an image. The median value of glandularity using the planimetry method was statistical ly different from the ACR BI-RADS, Tube Loading or BRTES-MOD TET methods (p 0.001, MW). Histogram Thresholding Method The histogram method was introduced to re duce the variability seen in PM and ACR BI-RADS of evaluating glandularity. The HTM me thod does not incorporate a standard for defining the minimal pixel value fo r glandular tissue. In this study, the ratio of the dense pixel intensity to maximu m pixel intensity was 0.64 Â± 0.05. The standardization of determini ng the dense area reduced the glandularity variability as shown in Table 7-6. The population trends s een in ACR BI-RADS we re not present with the HTM method. The median value of gl andularity using the histogram method was statistically difference than ACR BI -RADS, Tube Loading or BRTES-MOD TET methods (p 0.001, MW). The great advantage to th is method over planimetry is that it does not incorporate adipose region s into the area calculation. Tube Loading Method The premise of the tube loading met hod is that a mammography unit performs uniquely to the combination of gl andularity of the breast and tb. In this study, 40 data points were used to evaluated to determine the parameters in equation (7-2) for Mo/Mo and Mo/Rh combinations of each mammographic unit. The fitted parameters are listed in Table 7-7. The ability of the model to predict the glandularity of the BRTES-MOD phantom had a range in R2 of 0.75 to 0.96, shown in figure 7-9 through 7-12.
217 Even though the reference phantom had a tb range of one cm to eight cm, the models had difficulty in predicting glandularit ies of thin (<3 cm) and thick breasts (>7 cm). The result of these difficulties was that glandularity exceeded 100% in small breast were less than 0% in large breasts were pred icted as shown in Tabl e 7-4. The trend is quite evident with the three-dimensional surface of the models shown in figure 7-13 through figure 7-16 and in the distribution of glandularities in figure 7-5 and 7-6. The trend was also noted in a previous study by Heggie.120 The fundamental problem with this method is the original premise, that ch ange in glandularity doe s not have a dramatic impact on automated algorithms used in de termining unit parameters. The AOP/S mode in the mammographic units GE-DMR and GE800T used this study were not able to distinguish between glandular ities for phantoms of 1-2 cm. The method did show the increasing trend in adipose with age and thickness. BRTES-MOD Tissue Equivalent Thresholding Method The BRTES-MOD TET method was developed to address issues brought up in previous methods. The method uses BRTES-M OD phantom as the reference tissue for glandularity the method calibrates each mammo gram image individually. The method is reproducible because it does not rely on the interpreting individual to decision points. The method does show the trend of decr easing glandularity with age and tb shown in figure 77 and 7-8. The difference between the left and right breasts, shown in Table 7-6, was comparable to PM, and HTM. The disadvantage to the method in this study was the sensitometric adjustments needed for the screen-film imaging which, in clinical practice could be time consuming and costly. However, this could be overcome by taking the step phantom images immediately after patient imaging. A calibration was performed using BRTES-MOD phantom of known glandularity at fixed compressed breast thickness,
218 figure 7-17. The BRTES-MOD TET tended to over respond with thinner breasts at higher glandularity. Adjustment to th e BRTES-MOD TET method would have to be made to adopt the system in MLO views becau se of the incorporation of the pectoral muscle. The median glandularity value fo r the population was higher than ACR BIRADS, PM and HTM. Conclusions In this study, BRTES-MOD TET method wa s introduced as a quantitative method of estimating patient glandularity and it wa s compared it to ACR BI-RADS, PM, HTM, and TLM on a single study population. The advantage of the BRTES-MOD TET met hod of over ACR BI-RADS is that it is quantitative and does not rely on the us er's training and ex perience to predict glandularity. Even though PM was designed to be a quantitative measure of the user's training and experience, it also relied on the user and reduced the range of glandularity to binary choice of 100% glandular or 100% adipose. HTM improved reproducibility over the PM method, but it still reduced the range of glandularity to a bi nary choice. TLM was the first non-user-dependent qua ntitative system, but in this study the underlying premise that AEC systems provide unique exposures based on glandularity and breast thickness turned out to be false. TLM was unable to reliably predict breast glandularities of small (<3cm) and large (>7 cm) compressed breasts. The BRTES-MOD TET method does require ca reful application when applied to screen-film mammography, in or der to ensure that sensitom etry does not interfere with the method results. The issue is avoided if the phantom images are taken immediately after patient imaging. Application of this method in a digital e nvironment would reduce the number of steps required to achieve a resu lt, since it would not require the user to
219 digitize the images. Also furt her development is needed in modifying the system to enable the application to MLO views, which include the pectoral muscle. Even though not directly compared in this study, the advantages of the BRTESMOD TET method over the volumetri c method proposed by Pawluczyk et al. that each mammogram is calibrated instead of using a single calibration curve. 47 The BRTESMOD TET method also did not require corr ection for field nonuniformity effects. Mammography dosimetry needs a reproducib le and reliable method of estimating breast glandularity to accurately reflect the patient-specific average glandular dose that can be documented. The BRTES-MOD TET method is proposed as the method that could provide an estimate of glandularity that is traceable to a standard tissue equivalent phantom.
220 1 0 0 . 0 % 8 3 . 4 % 6 7 . 8 % 5 4 . 2 % 4 2 . 6 % 3 3 . 0 % 2 5 . 4 % 1 9 . 8 % 1 6 . 2 % 0 . 0 % 0.0% 0.0%0.0% 8 cm 1 2 c m 9 cm(A) (C) 8 cm 1 cm 1 cm 8 cm 1 c m(B)100.0% RCC (D) Figure 7-1. (A) BRTES-MOD step phantom dime nsions (B) Individual step dimensions (C) BRTES-MOD step phantom glandularity content.(D) BRTES-MOD phantom geometry for clinical imaging.
221 Mammogram CC view Breast Edge Uniform Dense Whole Figure 7-2. Segmentation regions used in LCC and RCC mammographic views
222 66 cm X-ray Source 18 cm 8 cm 24 cm 18 cm 2 cm 1cm x 1cm x 0.1cm 1Air 4-adipose 12-glandular 4-adipose 1-Air 1Air 4-adipose 4-glandular 4-adipose 1-Air 4-glandular 4-adipose 1Air 4-glandular 4-adipose 4-glandular 1-Air 4-adipose 4-glandularGlandular or Adipose(A)(B) (C) (C-1) (C-2) (C-3) 1Air 3-adipose 7-glandular/7-adipose 3 1-Air -glandular(C-4) Patterns Evaluated Figure 7-3. (A) MCNP-5 geometry used in Monte Carlo (MC) simulations. (B) BRTESMOD phantom and image receptor physical measurements used in MC. (C) BRTES-MOD phantom tissue layer sequences.
223 Figure 7-4. BRTES-MOD TET me thod illustrated flowchart.
224 Glandularity (%) 0 . 0 0 1 0 . 0 0 2 0 . 0 0 3 0 . 0 0 4 0 . 0 0 5 0 . 0 0 6 0 . 0 0 7 0 . 0 0 8 0 . 0 0 9 0 . 0 0 1 0 0 . 0 0Frequency (%) 0 10 20 30 40 50 60 ACRBIRADS Planimetry Histogram TubeLoading BRTESMOD Figure 7-5. A histogram for the five glandular ity-estimating methods using a 10% increasing increment of glandularity.
225 Glandularity (%) <2525-5051-75>75Frequency (%) 0 20 40 60 80 ACR Planimetry Histogram Tube Loading BRTES-MOD Figure 7-6. A histogram for th e five glandularity-estimati ng methods using the ACR BIRADS density category as th e increment of glandularity.
226 Compressed Thickness Groups (cm) <33-55-7>7Glandularity (%) 0 50 100 150 200 250 300 ACR Tube Loading Planimetry Histogram BRTES-MOD Figure 7-7 Glandularity-estimating methods as a function of compressed breast thickness groups.
227 Age Groups (Years) 30-3940-4950-5960-6970-79>=80Glandularity (%) -20 0 20 40 60 80 100 120 140 160 ACR Tube Loading Planimetry Histogram BRTES-MOD Figure 7-8. Glandularity-estimating me thods as a function of age group.
228 BRTES-MOD Phantom Glandularity (fraction) 0.00.10.20.188.8.131.52.184.108.40.206Tube Loading Predicted Glandularity (fraction) 0.0 0.2 0.4 0.6 0.8 1.0 2 cm 3.3 cm 4 cm 6 cm 8 cm One-to-one Correlation Figure 7-9. A graph of the tube loading method model of GE-800T (Mo/Mo) predicted glandularity as a func tion of BRTES-MOD phantom glandularity. The line represents a one-to-one correlation be tween model prediction and phantom glandularity.
229 BRTES-MOD Phantom Glandularity (fraction) 0.00.10.20.220.127.116.11.18.104.22.168Tube Loading Predcited Glandularity (fraction) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 2 cm 3.3 cm 4 cm 6 cm 8 cm Figure 7-10. A graph of the tube loading method model of GE-800T (Mo/Rh) predicted glandularity as a func tion of BRTES-MOD phantom glandularity. The line represents a one-to-one correlation be tween model prediction and phantom glandularity.
230 BRTES-MOD Phantom Glandularity (fraction) 0.00.10.20.22.214.171.124.126.96.36.199Tube Loading Predicted Glandualrity (fraction) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 2 cm 3.3 cm 4 cm 6 cm 8 cm One-to-one Correlation Figure 7-11. A graph of the tube loading method model of GE-DMR (Mo/Mo) predicted glandularity as a func tion of BRTES-MOD phantom glandularity. The line represents a one-to-one correlation be tween model prediction and phantom glandularity.
231 BRTES-MOD Phantom Glandualrity (fraction) 0.00.10.20.188.8.131.52.184.108.40.206Tube Loading Predicted Glandularity (fraction) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 2 cm 3.3 cm 4 cm 6 cm 8 cm One-to-one correlation Figure 7-12. A graph of the tube loading method model of GE-DMR (Mo/Rh) predicted glandularity as a func tion of BRTES-MOD phantom glandularity. The line represents a one-to-one correlation be tween model prediction and phantom glandularity.
232 -100 0 100 200 300 400 500 10 20 30 40 4.5 5.0 5.5P r e d i c t e d G l a n d u l a ri t y ( fr a c t i o n )U n i t C o m p r e s s e d T h i c k n e s s ( m m )l n ( k V p0 . 3 5m A s ) Figure 7-13. A three-dimensional graph of the tube loading method model of GE-800T (Mo/Mo) as a function of ln(kVp0.35mAs), and compressed breast thickness.
233 -3 -2 -1 0 1 2 50 55 60 65 70 75 80 85 31.5 32.0 32.5 33.0 33.5P re d i c t e d G l a n d u l a r i t y ( f r a c t i o n )U n i t C o m p r e s s e d B r e a s t T h i c k n e s s ( m m )l n ( k V p8m A s ) Figure 7-14. A three-dimensional graph of the tube loading method model of GE-800T (Mo/Rh) as a function of ln(kVp8mAs), and compressed breast thickness.
234 -4 -2 0 2 4 6 8 10 12 5 10 15 20 25 30 35 40 45 21.5 22.0 22.5 23.0 23.5 24.0 24.5P r e d i c t e d G l a n d u l a r i t y ( f ra c t i o n )U n i t C o m p r es s ed B r ea s t T h i c k n e s s ( m m )l n ( k V p5 . 8 9* m A s ) Figure 7-15. A three-dimensional graph of the tube loading method model of GE-DMR (Mo/Mo) as a function of ln(kVp5.89mAs), and compressed breast thickness.
235 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 45 50 55 60 65 70 75 54.5 55.0 55.5 56.0 56.5P r e d i c t e d G l a n d u l a r i t y (f r a c t i o n )U n i t C o m p r es s e d B r e a s t T h i c k n es s ( m m )l n ( k V p1 5* m A s ) Figure 7-16. A three-dimensional graph of the tube loading method model of GE-DMR (Mo/Rh) as a function of ln(kVp15mAs), and compressed breast thickness.
236 BRTES-MOD Phantom Glandualrity (%) 1020304050607080BRTES-MOD TET Predicted Glandualrity (%) -60 -40 -20 0 20 40 60 80 100 120 140 GE-DMR GE-800T Figure 7-17. A graph of the BRTES-M OD TET model for GE-DMR and GE-800T predicted glandularity as a functi on of BRTES-MOD phantom glandularity. The line represents a one-to-one corr elation between m odel prediction and phantom glandularity.
237 Table 7-1. Data collected in the retrospective study. Collected Data Descriptor Characteristics BI-RADSÂ® Assessment Category Cat 1 or Cat 2 (only) Note: Assigned by interpreting radiologist BI-RADSÂ® Density Category <25%, 25-50%, 51-75%, >75% Note: Assigned by interpreting radiologist Mammography Technologist Code Anonymity maintained in study Interpreting Radiologist Code Anonymity maintained in study Individual Age If 89 years old of greater assigned " 89" Hormone replacement status Yes or No Radiographic Positioning Craniocaudal (CC) and mediolateral oblique (MLO) views, left (L), right (R) Codes: LCC, LMLO, RCC, RMLO Radiographic Unit GE-800T or GE-DMR Compressed Breast Thickness Distance in millimeters (mm) Compression Pressure Pressure in decaNewtons (daN) Radiographic Unit Operating Mode AOP-S, AOP-C, AOP-D, AEC Tube Potential kVp Tube Current-Time Product mAs Focal spot size 0.1 mm or 0.3 mm Target Mo or Rh Filter Mo or Rh Positioning Angle Degrees
238 Table 7-2. Spectral parameters used in MCNP5 simulations. Spectra Parameters Spectra source IPEM Report 78 Tube Potential 25 kVp 0.69 mm Be 0.03 mm Mo 2.5 mm PMMA Filters 0.1745 mm Al Half-Value Layer (HVL) measured 0.3476 mm Al HVL (MCNP5) simulated 0.3462 mm Al Difference in HVL (%) 0.41% Mean photon energy 16.6 keV Anode angle 20 degrees
239Table 7-3. The final exit doses from th ree patterns of tissue layers and surrounding tissues. Ti ssue patterns were illustrated in figure 3. Tissue Pattern Surrounding Tissue Dose (mGy, 10-13) Difference from Mean Tissue Column (%) Difference from Surrounding tissue (%) C-1 Adipose 2.02 Â± 0.025 2.17 Â± 1.41 11.41 Â± 1.68 C-2 Adipose 2.18 Â± 0.026 5.50 Â± 1.45 10.65 Â± 1.62 C-3 Adipose 1.99 Â± 0.025 3.33 Â± 1.40 10.95 Â± 1.69 C-4 Adipose 2.24 Â± 0.026 6.20 Â± 1.49 10.45 Â± 1.60 Mean Adipose 2.11 Â± 0.017 C-1 Glandular 1.79 Â± 0.023 2.64 Â± 1.43 12.89 Â± 1.68 C-2 Glandular 1.94 Â± 0.037 5.92 Â± 1.48 11.91 Â± 1.62 C-3 Glandular 1.78 Â± 0.026 3.28 Â± 1.42 12.30 Â± 1.69 C-4 Glandualr 2.00 Â± 0.024 6.68 Â± 1.52 11.67 Â± 1.60 Mean Glandular 1.84 Â± 0.013
240Table 7-4. Study population descriptive statistic s breast glandularity-estimating methods. Method Mean Â± (%) Â· images-(1/2) (%) Range (%) Min-Max (%) Median (%) ACR BI-RADS 46.70 Â± 18.40 1.08 75.0 12.5-87.5 37.50 Planimetry 30.10 Â± 16.10 0.95 78.3 0.9-79.2 28.30 Histogram 29.80 Â± 12.30 0.73 60.2 4.2-64.4 28.60 Tube Loading 47.90 Â± 66.60 3.92 648.90 -68.1-580.7 50.90 BRTES-MOD TET 41.20 Â± 21.70 1.29 77.8 0-77.8 44.90
Table 7-5. ACR BI-RADS assessment category 1 or 2 with associated density categories as a function of individual interpreting radiologists. BI-RADS Density Categories Interpreting Radiologist <25% 25-50% 51-75% >75% Subjects 001 0.00% 50.00% 50.00% 0.00% 27 002 5.26% 57.89% 34.21% 2.63% 38 004 21.74% 69.57% 8.70% 0.00% 23 005 8.77% 68.42% 17.54% 5.26% 57 006 8.00% 36.00% 48.00% 8.00% 25 007 12.20% 51.22% 24.39% 12.20% 41 008 9.09% 54.55% 18.18% 18.18% 11 009 3.33% 23.33% 63.33% 10.00% 30 Mean 8.37% 52.99% 32.27% 6.37% 251
Table 7-6. The mean glandularity calcul ated for LCC and RCC pairs in the study population. Glandularity Estimating Method View Glandularity Â± (%) Glandularity Difference (%) Images LCC 47.05 Â± 18.39 ACR BI-RADS RCC 47.05 Â± 18.39 0 LCC 35.56 Â± 16.30 Planimetry RCC 24.71 Â± 13.95 10.84 LCC 29.01 Â± 12.02 Histogram RCC 30.07 Â± 12.07 1.05 LCC 43.45 Â± 56.51 Tube Loading RCC 43.94 Â± 50.81 0.49 LCC 40.73 Â± 23.12 BRTES-MOD TET RCC 41.29 Â± 20.68 0.56 274
Table 7-7. Tube loading fitting parameters. Parameters GE-800T (Mo/Mo) GE-800T (Mo/Rh) GE-DMR (Mo/Mo) GE-DMR (Mo/Rh) n 0.35 8.00 5.89 15.00 y0 0.26 1.12 0.13 -3.96 a 27.67 15.68 36.22 648.65 b 181.21 -2544.93 -97.61 -23011.40 c 2.95 44.59 5.46 -207.75 d 88.50 -391.01 753.86 34577.92 e 777.43 -60742.77 -2047.07 -1237496.22 R2 0.91 0.96 0.82 0.75 2 2 0( lnb b n b b MOD BRTESt e t d c mAs kVp t b t a y G
244 CHAPTER 8 AVERAGE GLANDULAR DOSE BASE D ON HOMOGENEOUS PHANTOM Introduction Mammography provides the best opportunity today to detect th e early onset of breast cancer.5,44,137 The success of mammography has been in part due to the quality assurance oversight that has provided st andardization of breast imaging through accreditation and regulations.12,100 As part of that standardiz ation process, phantoms have played a key role in image quality and m onitoring the average glandular dose for a mammographic unit. This research affords the opportunity to go to the next step in diagnostic imaging, with individualized patient doses. The curr ent model for radiation induced cancer risk, from the Committee on the Biological Effects of Ionizing Radiations: Health effects of exposure to low levels of ionizing radiation (BEIR V) , uses in their foundation populations with relatively high dos es when compared to mammography.7 In vivo dosimetry has the possibility of allowing im provement to the current BEIR V model by providing for epidemiological studies a dose monitoring mechanism that takes into account population variability and imaging modality. Screening mammography patients would provide a population sufficien tly large to gain statistica l significance to allow clear determination of the radiati on risk associated with mammography without a need for model extrapolation. In vivo doses also provide the phys ician with the information needed to make an informed decision on that risk, as is recommended by the 2005 Recommendations of the International Co mmission on Radiological Protection (ICRP-
245 2005).86 The ICRP-2005 recommendati ons stated the following concerning diagnostic imaging "That exposure is not limited by any regula tory process, but is controlled by the physician, who therefore should be aware of the risks and benefits of the procedure involved." In vivo dosimetry can also provide regula tory bodies with realistic patient doses, which could drive cha nges in the current regulator y environment, potentially reducing the average glandular dose unit limits. The American College of Radiology has been FDA-approved to develop policies a nd procedures to ac credit mammography facilities. One of the ACR procedures de lineates the methodology for measuring average glandular dose. The dosimetric model used to measure average gl andular dose (Wu DM) was adapted from Wu et al. and Stanton et al.54,56,57 The Wu DM phantom model is based on a semielliptical-cylinder compressed-brea st geometry which c ontains a glandular region surrounded by a layer of adipose tissue, as illustrated in figure 8-1(A). The Wu DM has been adopted by the other three FDA-approved accreditation bodies, Arkansas, Iowa and Texas.138 The Wu DM average gl andular dose (AGD) is determined from the product of the normalized average glandular dose per unit entrance skin exposure and the measured entrance skin exposure (Xese), equation 8-1. DgN is a function of the incident spectru m, compressed breast thickness (tb), and breast fibroglandular content (G). The spectrum whic h can be characterized by the half-value layer (HVL), in turn, is a function of tube target and filter (T/F) materials, and tube potential (kVp). ese b gN gX kVp F T t HVL G D D , / , , , (8-1) The current dose conversion factors (DgN) used by the Wu DM were measured for breasts with a width of 18 cm and a poster ior nipple (PNL) dist ance of 8 cm, with
246 variable compressed breast thic knesses (2-8 cm), as illustra ted in figure 8-1(A). The phantom used to monitor the average gla ndular dose in mammography units is the rectangular ACR accreditation phantom that is intended to represent a 4.2 cm compressed breast composed of 50% fibr oglandular (glandula r) tissue. A common practice has been to use homogeneous tissue phantoms to eval uate patient glandularity and to calculate average glandular dose us ing equation (8-1) and DgN derived by Wu and colleagues.56,57 However, as shown in this study, the use of a homogeneous tissue phantom to evaluate breast glandularity may introduce systematic erro rs in the glandular content of the breast. The study further shows that an overestimation of gl andularity results in an underestimation of AGD. In this study, an accurate methodology to predict the average glandular dose (AGD) to a patient is developed that uses a homogeneous phantom to estimate glandularity. The homogeneous phantom used in this research is the modified breast tissue-equivalent phantom series (BRTES-M OD) described in Chapter 5. The AGD can then be customized to the glandularity and physical dimensions of the compressed breast of a specific patient. The study then compares the Dg derived from glandularity measurements using the BRTES-MOD met hod with those derived from using ACR breast imaging reporting and data system (ACR BI-RADS), histogr am threshold method (HTM), and modified averag e breast method (AVG). Method Precepts General Concepts The breast is made up of numerous ty pes of tissues; however, they can be categorized into two major cat egories: adipose and fibrogl andular. Both the Wu and BRTES-MOD dosimetry phantom models ar e composed of these two tissues.
247 The current model for calculating averag e glandular dose is the Wu DM. The BRTES-MOD dosimetric model (BRTES-MOD DM) is presented to facilitate the use of homogenous phantoms to estimate patient glan dularity and subsequently provide an average glandular dose. The Wu DM, as previously described, is co mposed of two distinct tissue layers, an adipose layer and homogenous fibroglandular region, illustrated in figure 8-1(A).56 The homogeneous fibroglandular re gion is composed of 100% ad ipose and 100% glandular tissue. Figure 8-1(B) illustrates the BRTES-MOD DM, which has overall lateral dimensions identical to th e Wu DM model but is com posed entirely of homogenous fibroglandular tissue. Th e fibroglandular content in both models is equal. In mammography, fibroglandular tissue attenuates x rays more than adipose tissue. In the BRTES-MOD DM, the digitized measurem ent (pixel intensity) of optical density (OD) of the film image of the phantom provides a surrogate measure for the fibroglandular content of the br east. Pixel intensity (PI) of the digitized image can be described using equation (8-2) whic h was derived in Chapter 7, where 0, 1 are the initial and attenuated fluences; Âµa, Âµg are the linear attenuation coefficients for adipose and glandular tissues respectively; G is the fibroglandular tissue fraction; tb is the compressed breast thickness; QDE is the qua ntum detection efficiency; CE is the conversion efficiency of the in tensifying screen; SPR is the scatter-to-prima ry ratio and DE is the digitizer conversion efficiency. DE dE E E SPR E CE E QDE e E PIkeV t G E G Eb g a 0 0 ) ) ( 1 ) ( ( 0) ( 1 log (8-2)
248 Due to the difficult nature of dealing with polyenergetic spectrum, an alternative method of estimating glandularity with pixel intensity can be used that utilizes the BRTES-MOD step phantom with a known glandul arity. Once glandularity is estimated using the BRTES-MOD TET method described in Chapter 7, average glandular dose (Dg) can be calculated using equation (8-3), where is the incident energy fluence; (Âµen/ )g, (Âµen/ )a are the energy absorption coefficients for glandular and adipose tissues; G and SPR is defined as in equation 8-2. dE E SPR G E G E E DkeV a en g en g ) ( 1 1 ) ( ) ( ) (0 (8-3) Again because of the complexity of dea ling with the polyenergetic spectrum of mammography, average glandular dose is typically not calcu lated using equation (8-3), but rather with equation (8-1). The general trend with the DgN, derived by Wu and colleague, is that DgN decreases as a function of in creasing glandularity, figure 8-2.56,57 Phantom Factor The breast glandularity measured usi ng the BRTES-MOD step phantom, figure 83, provides a reference to a known glandular ity of the BRTES-MOD phantom. In order to derive a dose, however, a c onversion factor must be used to compensate for the glandularity difference meas ured using a homogeneous phantom. The need for a conversion factor is best illustrated by comp aring the average glandular dose calculation based on an image of two phantoms whose fi broglandular regions, T/ F, tube potential, tube current-time product and compressed breas t thicknesses are equal, figure 8-4. One phantom is the Wu DM phantom, whereas the second is the homogeneous BRTES-MOD
249 DM phantom, figure 8-1(A and B). If th e homogeneous BRTES-MOD step phantom is used to measure the glandularity in both pha ntom images, the glandularity of the BRTESMOD DM phantom is higher than the glandul arity of the Wu DM phantom. The reason for this effect is that the adipose layer of the Wu DM phantom is primarily made of adipose tissue, which is less attenuating than the same layer in the BRTES-MOD DM phantom made of the homogenous fibrogla ndular tissues. The Wu DM phantom glandularity therefore indi cates a lower glandularity than the BRTES-MOD phantom. Given that DgN decreases with increasi ng glandularity, th e lower glandularity results in a DgN that is higher than expected for the fibrogl andular content in th e glandular region of the Wu DM phantom and therefore results in higher than expect ed average glandular dose. As a result, an adjustment must be ma de to compensate for this discrepancy. The phantom factor (fP) proposed is the ratio of Dg for the Wu DM to the BRTES-MOD DM phantom models derived using Monte Carlo simulations, Equation (8-4). ) ( ) (MOD BRTES g Wu g pD D f (8-4) Volumetric Factor In order to calculate an individual specific Dg, other factors also need to be evaluated, such as the compressed breast vo lume and anatomic features that affect average glandular dose. The Wu DM phantom had a predefined volume, which was not based on population study.56 Patient populations have a wider range of widths and PNL distances than those used by Wu DM as shown in Chap ter 6, so a volumetric factor (fv) is proposed to adjust the average glandular dose for the patient co mpressed breast volume. The measurements used for deriving the fv in this study were the compre ssed breast thickness as measured
250 with the mammography unit and measurements of breast width and PNL distances made on craniocaudal (CC) mammography images as described in Chapter 6. The fv is a ratio of the Dg based on patient measurements to the Dg based on Wu DM standard measurements, equation (8-5). ) , , , ( ) , , , () ( patient b bt PNL w Wu g t PNL w Patient g vD D f (8-5) Anatomical Factor Anatomical features that contribute to Dg are the pectoral muscle, intercostal muscles and ribs, which are typically directly pos terior to the breast. The anatomic factor fa is a ratio of the Dg based on the ACR dose model withou t anatomic features to the Dg based on the Wu DM dose model with anatomic features, equation (8-6). ) , , , ( ) , , , () ( patient b bt PNL w Wu g t PNL w Patient g aD D f (8-6) The Dg for an individual patie nt can then be derived using the following equation (8-7). a v p ESE b MOD BRTES gN gf f f X kVp F T t HVL G D D , / , , , (8-7) Materials and Methods Monte Carlo Simulations Monte Carlo simulations we re used to investigate the impact on the average glandular dose from the use of a homogene ous phantom, volumetric breast changes (individual changes in PNL, width and co mpressed breast thickness), and anatomic features of patients in th e following paragraphs.
251 The Monte Carlo Neutral Particle Code ve rsion 5 (MCNP-5) developed by the U.S. Department of Energy was utilized in c onjunction with polyenergetic mammography spectrum generated using the Institute of P hysics and Engineering in Medicine Report 78 (IPEM-78) spectrum generator software (SRS-78).10,136 MCNP-5 accounts for coherent scattering, incoherent scattering, photoel ectric effect absorption with florescence emission and pair production.10 Phantom Factor The comparison between the Wu phantom and the BRTES-MOD phantom was performed to derive the fp. The comparison was performed using the phantom geometry illustrated in figures 8-1(A and B) and setup geometry illustrated in figures 8-5(A and B). The Dg was based on the same fibr oglandular region in both phantoms, figure 8-1. The phantom factor fp was evaluated by varying breast tis sue glandularity from 0-100% in 10% increments and varying compressed breast thickness from 2 to 8 cm. In addition, a BR 12 phantom, representative of a 50% gl andular breast, was also evaluated for comparison to Wu DM and BRTES-MOD DM phantoms. The spectrum used for each glandularity corresponded to the spectrum used to image each of the BRTES-MOD phantoms of the same compressed breas t thickness and glandularity. The fp was calculated as the ratio of the Wu DM Dg to BRTES-MOD Dg at the same unit parameters and compressed breast thickness as de scribed in equation (8-4). Volumetric Factor The volumetric factor (fv) was a ratio of the Dg from Wu DM phantom standard geometry (18 cm width and 8 cm PNL) to the Dg of a Wu DM phantom with a variety of widths and PNLs. The comparison was performed using the phantom geometry illustrated in figure 8-1(A) and setup geometry illustrated in figures 8-5(A and C).
252 To evaluate the change in Dg and dose to the adipose region as a function of width, PNL, compressed breast thickness, the volumet ric factor was plotte d as a function of region tissue mass. Region mass was cal culated using equation (8-8). 4Re Re _ Re Re Re gion gion b gion gion giont width PNL Mass (8-8) The region mass is a function of PNL (PNLRegion), width (widthRegion), compressed breast thickness(tb_Region) and tissue density ( Region) associated with each tissue glandularity of the adipose and glandula r regions of the phantoms. Phantom glandularities evaluated were 67.8, 42.6, 25.4 and 16.2%, with compressed breast thicknesses of 2, 4, 6 and 8 cm, respectively. The impact of the heel effect would be greatest for large breasts extending beyond 8 cm PNL and 18 cm width. Therefore, the heel effect was evaluated by repeating the volumetric comparisons using a HVL increased by 5%. Terry et al. previously repo rted a range of 0.9 4.4 % increase in HVL for the GE Senographe DMR unit.139 Anatomical Factor The anatomic factor (fa) evaluated the effect on Dg of Wu DM phantom with and without the retromammary adipose, pectoral muscle, ribs and intercostal muscles. The comparison was performed using the phantom geometry illustrated in figure 8-1(A), anatomic features illustrated in figure 8-3( D) and setup geometry illustrated in figure 83(A). Phantom glandularities evaluated were 67.8, 42.6, 25.4 and 16.2% with compressed breast thicknesses of 2, 4, 6 and 8 cm, respectively. In addition, the effect on Dg from x-ray field misalignment of 0.5, 1, and 2% was evaluated for phantoms with compressed breast thicknesses of 2 and 4 cm and glandular content of 67.8 and 42.6%. The x-ray field misalignment was equal to or less than the
253 MQSA tolerance of 2% of the source-to-image receptor distance.12 The rectangular collimation was adjusted to place the x-ray fiel d on the anatomic features as illustrated in figures 8-3(D). Patient-specific Monte Carlo Simulations The phantom, volumetric and anatomic factors for the study population subjects were derived using a Monte Carlo simula tion that utilized the patient-specific mammography unit parameters and anthropomet ric measurements. The custom MCNP-5 simulation performed tracing of one million photon histories resulting in a glandular region dose with a rela tive standard error x/ less than 1%. Monte Carlo Model Elemental Composition The elemental composition of breast tissues (100% glandular, 100% adipose) used in MCNP-5 simulations were described in International Commission on Radiation Units and Measurements (ICRU) Publications 44 (ICRU-44) and mean values in ICRU Publication 46 (ICRU-46).107,114 Phantom-specific glandular content was derived using the per cent compositions of ICRU-46 100% adipose and 100% gla ndular tissues. Monte Carlo Spectrum The MCNP-5 spectrum was based on a clinical spectrum used to image a BRTESMOD phantom of the same compressed breas t thickness and clinical geometry. The quality of the clinical spectrum was meas ured using narrow-beam geometry. The 20Â° target angle used in the MCNP-5 simulations matched the characteristics described for General Electric mammography units. The MCNP -5 spectrum was modified to match the clinical spectrum by filtering the SRS-78 generated raw spectrum with 0.69 mm of beryllium (Be), 0.03 mm of molybdenum (Mo) , 2.5 mm of polymethyl methacrylate
254 (acrylic) and aluminum (Al). Al filter thickness and spectrum parameters are shown in Table 8-1. The acrylic filter was utilized to simulate the compression paddle in the beam path. After the MCNP-5 HVL matched the measured HVL by using the SRS-78 spectrum generator, the quality of the spect rum was verified using MCNP-5 geometry with Al filters. The MCNP-5 source used a 22.5Â° conical x-ray beam. In order to approximate the clinical conditions, such as the x-ray beam impinging normal to the chest wall edge of the image receptor, the x-ray sour ce and inverse-square intensity variations, the MCNP-5 x-ray focal spot was placed direc tly over the chest wall and collimated to a rectangular field, which eff ectively bisected the conical beam. MCNP-5 simulations were performed tracing 100 million photon historie s resulting in a glandular region dose with 0.1% relative standard error x/. Study Population and Selection Criteria A three-month (Jan-Mar 2005) retrospec tive demographic study evaluated data associated with 1040 mammography images from 253 patients seen at a U.S. Navy health care facility in Jacks onville, Florida. The study populati on consisted of women who were seen for screening mammography or because of a primary care physician referral for a definitive diagnosis. The patients selected for the study had an American College of Radiology (ACR) breast imaging reporting and data system BI-RADSÂ® (BI-RADS) assessment category of 1 (Negative) or 2 (Benign Findings) assigned to them by the interpreting radiologist.11 Overall, BI-RADSÂ® category 1 and 2 assessments are indicative of a normal breast with no evidence of malignancies. In evaluating methods to estimate breast glandularity, a subgroup of approximately 143 patients (CC view only) was selected who were assigned a BI-RADS a ssessment category 1. The selection of the
255 BI-RADS assessment category 1 patients re duced potential infl uence from benign conditions on the inter-comparison of breast glandularity es timation methods. Typical Mammographic Examination The typical screening mammography examina tion consisted of four images taken in two anatomical views, craniocaudal (CC) a nd mediolateral oblique (MLO), of the right and left breasts. The mammography units used to image patients ut ilize an automatic optimization parameters modality that optimizes unit parameters, by selecting the tube potential (kVp), tube current -time product (mAs), target-filter combination (T/F). Mammography Imaging A General Electric Senographe-DMR (G E-DMR) or a GE Senographe-800T (GE800T) mammography unit provided the mammogr aphy x-ray fields used throughout the study period. The Senographe-DMR uses a si ngle-phase high-frequency generator and an x-ray tube with a dual bi-metal track ta rget of molybdenum (Mo) and rhodium (Rh). The x-ray spectrum was filtered using Mo or Rh filters. The unit had a fixed source-toimage receptor SID distance of 660 mm and a focal spot of 0.3 mm. The Senographe800T uses a single-phase generator and an x-ra y tube with a single track Mo target. The x-ray spectrum was filtered using Mo or Rh filters. Both the GE-DMR and GE-800T utilize a focal spot of 0.3 mm in routine clinical imaging. Th e GE-DMR and GE-800T mammography units used an automatic optimization parameter (AOP) mode in clinical imaging which prioritizes for dose reduction (AOP-D), image contrast (AOP-C) or the standard, a compromise between dose reduction and contrast (AOP-S). The AOP selects T/F, kVp, and mAs and through the use of a pre-exposure, which access the algorithm using unit parameter and the radiographic thickn ess of the breast. The radiation emerging
256 from the breast passes through the patient s upport table, the antiscatt er grid, the cassette front, film base before being absorbed by the intensifying screen and film emulsion. Mammography Facility and Film Processing As a U.S. Food and Drug Administration certified and American College of Radiology (ACR) accredited facility, the studied facility maintained its operations in accordance with Mammography Quality St andards Act (MQSA) and the ACR Mammography Accreditation Program (ACR-MA P) requirements as described in the ACR Mammography Quality Control Manual and promulgated in MQSA statutes.12,100 The facility performed their da ily quality control activities and ensured that they were within prescribed control limits prior to processing any patient images as is required by MQSA and ACR-MAP.12 The Kodak Min-R 2000 screen -film system was used in combination with a Kodak XOMAT 5000 RA film processor for image processing. Digitization and Image Segmentation The mammography images, along with the daily processor control films, were digitized using a Kodak LS-75 (LS-75) film di gitizer with a limiting resolution of 5 linepairs per mm.121 Mammography images were digiti zed at a resolution width of 2048 pixels, with the number of lines determined by film size. The matrix size for 18 cm x 24 cm film was 2048 (0.009 cm) Â± 0 x 2811.85 (0 .009 cm) Â± 114.58 pixels and 24 cm x 30 cm film was 2048 ( 0.012 cm) Â± 0 x 2628.20 (0. 011 cm) Â± 21.20 pixe ls. All digitized mammography images were stored for imag e processing in a Digital Imaging and Communications in Medi cine (DICOM) format. The data collected on each mammography images are shown in Table 8-2. Tabulated data were compiled in paper fo rm and subsequently transferred to a
257 customized database program28). Statistical analysis was performed using SigmaStat version 3.129). The Mann-Whitney (MW) rank sum test and the Kruskal-Wallis (KW) one-way analysis of variance on ranks non-parame tric statistical test s were utilized to evaluate differences between two or more groups.135 Each digitized mammography image was manually segmented using ImageJ30) image analysis software, version 1.34i, de veloped by the National Institutes of Health.122 The digitized mammography images were ma nually segmented into the whole breast, breast edge, uniform and dense regions, illustrated in figure 8-1, as described in Chapter 6. Free-in-air Entrance Skin Exposure The ionization chamber used for freein-air entrance skin exposure (Xese) was a Victoreen NEROÂ®31) mAx model 8000 radiation mon itor with a mammography ion chamber model 6000-529. The mammography ionization chamber has a metal-coated polycarbonate window with a thickness of 9.5 mg cm-2 and an active volume of 3.3 cm3. Three free-in-air entrance skin exposures, m easurements were made 4 cm from the chest wall, 3 cm above the image receptor assemb ly and centered transversely using narrowbeam geometry for all clinical tube potentials and tube current-time stations used in the study-population patient mammograms. The averag e of the three free-i n-air entrance skin exposures (Xese(Ref)) as a function of tube current-tim e product was fitted to a linear equation, by least squares fitting with each clin ically used tube potential. Patient-specific 28 Microsoft Office Access 2003, Mi crosoft, Redmond, Washington 29 SigmaStat 3.1, Systat Software, Inc., Point Richmond, Ca 30 ImageJ, National Institutes of Health, Bethesda, MD 31 Inovision Radiation Measurements, 6045 Cochran Road, Cleveland, OH 44139-3303
258 Xese was estimated by correcting Xese(Ref) by inverse-square to the patient compressed breast thickness (tb). Average Glandular Dose The dose conversion factors (DgN) were based on a Sobol and Wu parametrization algorithm.140 Dg was calculated for the three methods of determining per cent glandularity (ACR BI-RADS, HTM and AVG) using equation (8-1). The ACR BIRADS, HTM and BRTES-MOD glandularity met hods were previously described in Chapter 7. The AVG method is based on the ACR method of calculating average glandular dose that assigns a value of 50% to glandularity and 4.2 cm to compressed breast thickness. The BRTES-MOD DM Dg, study-population Dg for a particular patient was estimated using equation (8-9). a v p ESE b MOD BRTES gN gf f f X kVp F T t HVL G D D , / , , , (8-9) All four methods utilize the HVL, tb, T/F, kVp used for the mammography image of the study-population and the glandularity estimated by the method; except that AVG method has glandularity of 50% and tb of 4.2 cm. The fp, fv and fa were evaluated for each study-popul ation patient using Monte Carlo simulation and study findings. The ACR BI-RADS Dg calculation used the patient compressed breast thickness and the radiol ogist-assigned tissue de nsity category, as described in Chapter 7. The HTM Dg calculation used the patient compressed breast thickness and the estimated gla ndularity using the HTM method described in Chapter 7.
259 Results and Discussion Monte Carlo Simulation HVL Good agreement was observed between the HVLs determined for the MCNP-5 spectra and the measured HVLs, as illustrated in figure 8-6. For each pair, the HVL of the MCNP-5 spectrum matched within 1.68% of th e clinically measured HVL, as shown in Table 8-1. Heel Effect Impact of Dg The heel effect was evaluated using na rrow-beam geometry to measure spectrum HVL at 4 cm centered on the chest wall and at 8 cm from the chest wall with no detectable difference in HVL. The measuremen ts were made to evaluate the magnitude of the heel effect on the Wu DM phantom. As a result, no corrections were made for fp, because the phantoms did not extend beyond 8 cm PNL. The impact on Dg from the heel effect on breasts with lateral dimensions exceeding those of the Wu DM phantom were examined for the calculation of fv. MCNP-5 simulations were performed for all the breast sizes listed in Table 83 with an HVL increase of 5%. The percent difference in Dg resulting from an increase of 5% in HVL was less than 3% as shown in figure 8-7 for all the breast masses corresponding to dimensions listed in Table 8-3. Even though the MCNP-5 simulation did not have a gradual incr ease in HVL from the chest wall as would be found with the heel effect in a clini cal environment, the 5% increase in HVL simulation shows that the impact of the heel effect would be at most, 3%. Phantom Factor In the MCNP-5 evaluation, the dose to th e glandular region in the Wu DM phantom increased as a function of gl andularity, whereas the dose to the adipose region decreased as a function of glandularity, figure 8-8 th rough figure 8-12. The dose to the adipose
260 region fell below the dose to the glandular region for glandularities above 60% for 2 cm compressed breast thickness as shown in fi gure 8-8. In the BRTES-MOD DM phantom, the dose to the outer region of the same thickness as Wu increased with increasing glandularity, and the dose to the inner regi on decreased with incr easing glandularity, for all compressed breast thickness tested. The findings for the Wu DM phantom indicat e that the as glandul arity increases in the glandular region, the addi tional attenuation decreases the dose to the adipose region which surrounds the glandular region. The ef fect seen in the BRTES-MOD DM phantom shows that the increasing glandularity in the outer region reduces the dose absorbed in the glandular region. The over all effect is that the fp for all evaluated compressed breast thicknesses show an increase fo r all glandularities above 0% glandular content, indicating that the Dg for the Wu DM model is always higher than that of a homogeneous phantom Dg. The clinical impact of this effect is that using a homogeneous phantom to evaluate dose must have a correction factor to compensate for the Dg effect seen in this study. If the BRTES-MOD phantom is used in estimating glandularity, it would underestimate the glandularity of the Wu DM phantom because of the lower attenuation properties of the adipose layers as compared with the BRTES-MOD phantom. BR 12 (50% glandularity) was used as a comparison to both the Wu DM and BRTES-MOD DM phantoms. The BR 12 adipose-region dose consistently was higher than either the BRTES-MOD or Wu DM phantom s, which is consistent with the higher effective Z ( 7.14) of the BR 12 tissue composition, figure 8-8 through figure 8-12, because of increased photoelectric intera ctions. The BR 12 glandular-region dose was consistently higher than the BRTES-MOD and lower than the Wu DM phantoms, figure
261 8-8 through figure 8-12. All compressed breast thicknesses evaluated indicated that the Wu DM phantom had a higher dose than the homogeneous BRTES-MOD or BR 12 models, primarily because of the attenuating effect of the adipose region. Phantom homogeneity had the largest impact on the sm allest compressed breast thickness of 2 cm, which showed that the homogenous phantom under-responds by as much as 12% with increasing glandularity, as il lustrated in figure 8-13, when compared to the Wu DM phantom. As a comparison, fp(BR 12) (Dg(Wu DM) Dg(BR 12)-1) was also calculated for BR 12 tissue composition. The Dg difference between Wu DM and BR 12 was less than 2% for the compressed breast thicknesses of 2 to 8 cm as illustrated in figure 8-14. Overall, the difference in gl andular region doses between the Wu DM phantom and the BRTES-MOD phantom were sufficiently signi ficant to warrant adjustment, given a threshold of 5%. Volumetric Factor The volumetric factor (fv), or the compressed breast size effect, on Dg was examined by utilizing the st udy population anthropometric m easurements of the breast PNL and width shown in Table 8-3. The Wu DM phantom geometry (PNL, width) was consistent with the study-popul ation compressed breast thickness of less than 3 cm as shown in Table 8-3. Dg varied as a function of breast ma ss (equation 8-8) as illustrated in figure 8-15 through figure 8-18 for a given comp ressed breast thickness. Overall, the Dg increase to a peak then decrease with increasing breast mass. The change is most dramatic in compressed breast thickness of 2 cm. The primary effect in the large breasts is attributed to the decrease of intensity in the x-ray field. The effect seen in the small breasts is due to the loss of scattered radi ation as the volume decreases. In general, fv ranges from a 0.28% increase to a 2.68% d ecrease as a function of breast mass over the
262 range of 37 2700 gm, illustrated in figur e 8-19 through figure 8-22. The largest decrease (2.68%) occurred with the 2 cm compressed breast size having a 2 cm PNL and 12 cm width. The largest increase (0.28%) occurred with 8 cm compressed breast size having a 12 cm PNL and 24 cm width. The change in Dg is consistent with the data reported by Wu et al.56 The clinical implication of this finding is that there is a decrease in Dg for small breasts and an increase in Dg for large breasts when compared to the standard size phantom used by Wu DM. Cu rrent mammography systems do not adjust for breast dimensions other than compressed breast thickness. Anatomical Factor The presence of anatomic features such adipose layer, pectoral muscle and ribs, illustrated in figure 8-3(D) with proper x-ray field alignment resulted in a fa of only 1.01 Â± 0.002 which was independent of the evaluated kVp, HVL, tb, or glandularity. However, it was when the x-ray field was misaligned by 2.0% as shown in figure 8-23 that the greatest impact on Dg from the anatomic features was seen with a change of fa as much as 4%. The clinical impact of this finding is th at anatomic features outside of the breast do not contribute to the Dg unless there is x-ray field misalignment greater than that prescribed by the MQSA. In th e calculation of BRTES-MOD Dg, fa is therefore assigned a value of one, indicating no significant cont ribution resulting from the adipose layer, pectoral muscle, and ribs of the patient. Free-in-air Entrance Skin Exposure The entrance skin exposure was calculated from equation (8-11) with the regression equation parameters listed in Table 8-4. The Xese(Ref) as a function of mAs was linear regardless of kVp, with R2 >0.999. The Xese for patients was determined by inversesquare correction to tb.
263 Average Glandular Dose Application of the BRTES-MOD dose calcula tion, equation (814), was performed on the study population by determining fp and fv factors in MCNP-5 simulation for each individual patient. Because the anatomic factor had less than 1% impact on Dg, it was assigned a value of 1.0 for all patients in the study population. The average glandular dose distribution for all four glandularity methods is illustrated in figure 8-24. Descriptive statistics for Dg are shown in Table 8-5. The BRTES-MOD DM had the second highest mean Dg at 2.4 Â± .61 mGy. The differences in median values among BRTES-MOD, histogram and AVG methods were not statistically significant, (p=0.056, KW). The trend of the Dg as a function of compressed breast thickness, illustrated in figure 8-25, show that the AVG method underestimates doses for compressed breast thicknesses of less th an 4.8 cm and overestimates doses for compressed breast thickness greater than 4. 8 cm. The AVG method may be sufficient for evaluating mammography units and mammogra phy facilities in their compliance with MQSA regulations, but it is clea rly insufficient for estimating Dg to individual patients because of the trend seen in the scatter plot shown in figure 8-5. The differences amongst the remaining three methods of estimating Dg are based on the inaccuracies of their methods to estimate glandularity. The BRTES-MOD DM methodology bases its glandularity on a standard tissueequivalent-phantom series, mammography unit a nd image detector. The tissue equivalent phantom series is traceable to elemental co mpositions of adipose and glandular tissues defined in ICRU-46. The BRTES-MOD DM method does not rely on the user for any decision to estimate glandularity, unlike ACR BI-RADS and HTM methods. The ACR BI-RADS relies on the radiologist's interpreta tion of the ACR density categories. The
264 HTM method relies on the user to decide th e threshold difference between 100% adipose and 100% glandular tissue. The best met hod to illustrate the BRTES-MOD DM method is by providing an example for a patient in th e study population, shown in Table 8-6. The fp, fv, fa product relating to the use of a homogeneous phantom, breast size and anatomic features accounted for approximately 6% increase the Dg for in this patient. The clinical application of the BRTES-MOD methodology would provide patients with average glandular doses that are patient -specific and documentable. Conclusions This study describes a method to utilize a homogeneous phantom such as the BRTES-MOD phantom to determine the average glandular dose Dg of the breast. The method adjusts for the use of the homogeneous phantom (fp), breast volume (fv), and anatomic features (fa) that may contribute to Dg. The study shows th at the use of a homogeneous phantom fp to measure glandularity can underestimate the Dg by as much as 11% for a compressed breast thickness of 2 cm and glandularity of 90%. Clinically, given a glandularity range of 10-90%, th e use of a homogeneous phantom could underestimate the Dg by a range of 1-11%. Th e impact of breast size fv on Dg had a range of 0.3% increase to 2.7% decrease as a f unction of compressed breast thickness, PNL and width. The anatomic features fa such as subcutaneous adipos e, pectoral muscle and ribs would be pertinent only if mi salignment of x-ray field is beyond acceptable standards; otherwise, it affects Dg by less than 1%. In evaluati ng the clinical population, only fp, and fv were utilized because the fa effect on Dg was negligible. Even though the BRTES-MOD DM method Dg was found to have a similar trend in Dg as the ACR BI-RADS and HTM methods, it provides a quantitative method that eliminates individual judgment variation inherent to these other systems.
265 The application of the BRTES-MOD protocol in a clinical setting would proceed as follows: 1. The patient arrives and receives the typical mammographic examination. 2. Dosimeters are placed on the underside of the compression paddle adjacent to the breast to measure the entrance skin exposure. 3. While the patient waits, the technologist images the BRTES-MOD step phantom and processes all images. 4. The digital version of the mammogram is segmented as described in Chapter 6 by the technologist using a automated computer routine. 5. Glandularity is estimated from the di gitized mammogram and step phantom images as described in Chapter 7 by the technologist using a computer program. 6. A MCNP-5 simulation using the measur ements of estimated glandularity, compressed breast thickness, PNL, width and mammography unit parameters is used to estimate the fa, fv, and fa as describe in this Chapter to determine Dg for the patient by the technologist using a computer program. 7. The Dg calculated using equation (8-14) is documented as part of the mammogram record. Thus, clinical application of the BRTES -MOD system would provide patients with average glandular doses that are patient-sp ecific and documentable. Additionally, clinical application of this system would provide the information needed to improve existing radiation-risk models , so that they would be ba sed on actual patient population data.
266 XY YZ 8 cm 18 cm XYYZ 8 cm 18 cm XYZ Adipose (Outer, 0.4 cm) (A) (B) XZ XZ XYZ Fibroglandular (Inner) Fibroglandular (Inner) Fibroglandular (Outer) Figure 8-1. (A) Wu DM breast model showing the adipose and fibrogl andular regions and dimensional measurements (B) BRTES-MOD breast model showing the dimensional measurements. Dg is measured in the inner region shown in both models, which represents the homogene ous fibroglandular region of the breast
267 Glandularity (fraction) 0.00.20.40.60.81.0DgN(mGy/mGy) 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 Mo/Mo, 25 kVp, HVL: 0.35 mm Al, 2cm Mo/Mo, 27 kVp, HVL: 0.36 mm Al, 4cm Mo/Rh, 27 kVp, HVL: 0.42 mm Al, 6cm Mo/Rh, 30 kVp, HVL: 0.46 mm Al, 8cm Figure 8-2. Average glandul ar dose conversion factors DgN as a function of breast glandularity for the spectrum used to image the BRTES-MOD phantom.
268 1 0 0 . 0 % 8 3 . 4 % 6 7 . 8 % 5 4 . 2 % 4 2 . 6 % 3 3 . 0 % 2 5 . 4 % 1 9 . 8 % 1 6 . 2 % 0 . 0 % 0.0% 0.0%0.0% 8 cm 1 2 c m 9 cm(A) (C) 8 cm 1 cm 1 cm 8 cm 1 c m(B)100.0% RCC (D) Figure 8-3. (A) BRTES-MOD step phantom measurements. (B) BRTES-MOD phantom wedge measurements. (C) BRTES-MOD st ep phantom glandularity. (D) The typical clinical geometry used for the BRTES-MOD step phantom.
269 G Wu BRTES-MOD GBRTES-MOD BRTES-MOD G GG DgN(BRTES-MOD)DgN(Wu) DgN(Wu) DgN (BRTES-MOD) ÂƒpÂƒ=Dg(Wu)/Dg(BRTES-MOD)pGBRTES-MOD T/F, kVp, mAs, tbT/F, kVp, mAs, tb ODWuODBRTES-MOD GG Figure 8-4. Flowchart rationa le for homogeneous phantom factor used with BRTESMOD.
270 66 cm X-ray S ource(A) 24 cm 18 cm 24 cm 30 cm BRT ES-MOD Homogeneous MQSA HeterogeneousF i b r o g l a n d u l a r r e g i o nPhantom F i b r o g l a n d u l a r r e g i o n Adipose Muscle Bone fibroglandula r (B)(C)(D)fibroglandular Adipose Adipose XY YZ xyz YZ 2-8 cm Or Figure 8-5 (A) MCNP-5 geometry used in th e simulations. (B) Tissue region descriptions for phantom factor phantoms (C), volumetric factor phantoms (D) and anatomic phantom.
271 Measured HVL(mm Al) 0.340.360.380.400.420.440.460.48MCNP-5 HVL (mm Al) 0.34 0.36 0.38 0.40 0.42 0.44 0.46 0.48 GE-800T GE-DMR Figure 8-6. The clinically m easured half-value layer as function of MCNP-5 HVL. The solid line represents a oneto-one correspondence of HVLs.
272 Mass (gm) 05001000150020002500Dg Difference (%) -3 -2 -1 0 1 2 3 4 2 cm 4 cm 6 cm 8 cm Figure 8-7. The percent differe nce of average glandular dose, Dg, as a function of mass resulting from 5% increase in half-value layer.
273 Glandularity (%) 020406080100Dose (mGy) 10-8 particle histories-1 3.20e-13 3.52e-13 3.85e-13 4.17e-13 4.49e-13 4.81e-13 5.13e-13 5.45e-13 Wu Adipose Region Wu Glandular Region BRTES-MOD Outer Region BRTES-MOD InnerRegion BR 12 Outer Region BR 12 Inner Region Figure 8-8. Adipose and gl andular region doses of MQSA, BRTES-MOD and BR 12 phantoms as a function of breast tissue glandularity for a compressed breast of 2 cm with a Mo/Mo spectrum at 25 kVp.
274 Glandularity (%) 020406080100Dose (mGy) 10-8 particle histories-1 2.56e-13 2.88e-13 3.20e-13 3.52e-13 3.85e-13 4.17e-13 4.49e-13 Wu Adipose Region Wu Glandular Region BRTES-MOD Outer Region BRTES-MOD Inner Region BR 12 Outer Region BR 12 Inner Region Figure 8-9. Adipose and gl andular region doses of MQSA, BRTES-MOD and BR 12 phantoms as a function of breast tissue glandularity for a compressed breast of 3.3 cm with a Mo/Mo spectrum at 26 kVp.
275 Glandularity (%) 020406080100Dose (mGy) 10-8 particle histories-1 2.24e-13 2.56e-13 2.88e-13 3.20e-13 3.52e-13 3.85e-13 4.17e-13 4.49e-13 Wu Adipose Region Wu Glandular Region BRTES-MOD Outer Region BRTES-MOD Inner Region BR 12 Outer Region BR 12 Inner Region Figure 8-10. Adipose and glandular regi on doses of MQSA, BRTES-MOD and BR 12 phantoms as a function of breast tissue glandularity for a compressed breast of 4 cm with a Mo/Mo spectrum at 27 kVp.
276 Glandularity (%) 020406080100120Dose (mGy) 10-8 particle histories-1 1.60e-13 1.92e-13 2.24e-13 2.56e-13 2.88e-13 3.20e-13 3.52e-13 3.85e-13 Wu Adipose Region Wu Glandular Region BRTES-MOD Outer Region BRTES-MOD Inner Region BR 12 Outer Region BR 12 Inner Region Figure 8-11. Adipose and glandular regi on doses of MQSA, BRTES-MOD and BR 12 phantoms as a function of breast tissue glandularity for a compressed breast of 6 cm with a Mo/Mo spectrum at 27 kVp.
277 Glandularity (%) 020406080100Dose (mGy) 10-8 particle histories-1 1.28e-13 1.60e-13 1.92e-13 2.24e-13 2.56e-13 2.88e-13 3.20e-13 3.52e-13 Wu Adipose Region Wu Glandular Region BRTES-MOD Outer Region BRTES-MOD Inner Region BR 12 Outer Region BR 12 Inner Region Figure 8-12. Adipose and glandular regi on doses of MQSA, BRTES-MOD and BR 12 phantoms as a function of breast tissue glandularity for a compressed breast of 8 cm with a Mo/Mo spectrum at 30 kVp.
278 Glandularity (%) 0102030405060708090100fp (DoseWu DoseBRTES-MOD -1) 0.98 1.00 1.02 1.04 1.06 1.08 1.10 1.12 1.14 2 cm, Mo/Mo, 25 kVp 3.3 cm, Mo/Mo, 26 kVp 4.0 cm, Mo/Mo, 27 kVp 6.0 cm, Mo/Rh, 27 kVp 8.0 cm, Mo/Rh, 30 kVp Figure 8-13. Phantom factor for 2, 3.3, 4, 6, and 8 cm compressed breast phantoms as a function of breast tissue glandularity.
279 Phantom Thickness (cm) 123456789fp (DoseWu DoseBR 12 -1) 1.00 1.01 1.02 1.03 1.04 1.05 1.06 Figure 8-14. Phantom factor, BR 12, as a function of compressed breast thickness.
280 Mass(gm) 050100150200250300350Dose (mGy) 4.13e-13 4.17e-13 4.20e-13 4.23e-13 4.26e-13 4.29e-13 4.33e-13 4.36e-13 4.39e-13 Adipose Region Glandular Region Figure 8-15. Adipose and glandul ar region doses of Wu phantom as a function of breast mass for a compressed breast of 2 cm, 67.8% glandularity, using a Mo/Mo spectrum at 25 kVp.
281 Mass (gm) 02004006008001000Dose (mGy) 1.79e-13 1.83e-13 1.86e-13 1.89e-13 2.40e-13 2.56e-13 Adipose Region Glandular Region Figure 8-16. Adipose and glandul ar region doses of Wu phantom as a function of breast mass for a compressed breast of 4 cm, 42.6% glandularity, using a Mo/Mo spectrum at 27 kVp.
282 Mass (gm) 0200400600800100012001400160018002000Dose (mGy) 1.41e-13 1.44e-13 1.47e-13 1.51e-13 2.08e-13 2.24e-13 2.40e-13 Adipose Region Glandular Region Figure 8-17. Adipose and glandul ar region doses of Wu phantom as a function of breast mass for a compressed breast of 6 cm, 25.4% glandularity, using a Mo/Rh spectrum at 27 kVp.
283 Mass (gm) 05001000150020002500Dose (mGy) 1.12e-13 1.15e-13 1.19e-13 1.22e-13 2.08e-13 2.24e-13 Adipose Region Glandular Region Figure 8-18. Adipose and glandul ar region doses of Wu phantom as a function of breast mass for a compressed breast of 8 cm, 16.2% glandularity, using a Mo/Rh spectrum at 30 kVp.
284 Mass (gm) 050100150200250300350fv (DoseWu (size) DWu (std size)-1) 0.965 0.970 0.975 0.980 0.985 0.990 0.995 1.000 1.005 Figure 8-19. Volumetric factor as a function of breast mass for a compressed breast of 2 cm.
285 Mass (gm) 02004006008001000fv (DoseWu (size) DWu (std size)-1) 0.984 0.986 0.988 0.990 0.992 0.994 0.996 0.998 1.000 1.002 Figure 8-20. Volumetric factor as a function of breast mass for a compressed breast of 4 cm.
286 Mass (gm) 200400600800100012001400160018002000fv (DoseWu (size) DWu (std size)-1) 0.984 0.986 0.988 0.990 0.992 0.994 0.996 0.998 1.000 1.002 1.004 Figure 8-21. Volumetric factor as a function of breast mass for a compressed breast of 6 cm.
287 Mass (gm) 60080010001200140016001800200022002400fv (DoseWu (size) DWu (std size)-1) 0.988 0.990 0.992 0.994 0.996 0.998 1.000 1.002 1.004 1.006 Figure 8-22. Volumetric factor as a function of breast mass for a compressed breast of 8 cm.
288 (A)X-ray Field Misallignment Percent 0.00.51.01.52.02.5fa(DosewanaDosewoana -1) 0.90 0.92 0.94 0.96 0.98 1.00 1.02 1.04 2 cm Phantom 4 cm Phantom X-ray field Patient Misaligned X-ray field (B) Figure 8-23. A) Anatomic factor fa as a function of percent x-ray field misalignment for compressed breast thicknesses of 2 and 4 cm. B) Schematic of x-ray field misalignment.
289 Dg(mGy) 0 . 0 1 . 0 1 . 0 1 . 5 1 . 5 2 . 0 2 . 0 2 . 5 2 . 5 3 . 0 3 . 0 3 . 5 3 . 5 4 . 0 4 . 0 4 . 5 4 . 5 5 . 0 >5 . 0Frequency (%) 0 10 20 30 40 50 BRTES-MOD ACR BI-RADS HTM AVG Figure 8-24. Histogram distribut ion of average glandular dose Dg.
290 Compressed Thickness (cm) 12345678Dg (mGy) 0 1 2 3 4 5 6 7 8 HTM AVG BRTES-MOD ACR BI-RADS Figure 8-25. Average glandular dose trends of average glandular dose as a function of compressed breast thickness for HTM, AVG, BRTES-MOD, and ACR BIRADS methods. The AVG dose is measured for the 4.2 cm, 50 % glandularity phantom, and th en the patient-specific te chniques factors are used to determine the dose-conversion factor for the actual compressed breast thickness of the patient.
291 Table 8-1. Spectra used in population evaluation and MCNP-5 simulations. Mammography Unit Tube Target Filter kVp HVL Measured (mm Al) HVL MCNP-5 (mm Al) Diff(%) Al Filter GE800T Mo Mo 26 0.3577 0.3562 0.42 0.1690 GE800T Mo Mo 27 0.3578 0.3575 0.08 0.1304 GE800T Mo Mo 28 0.3680 0.3668 0.33 0.1330 GE800T Mo Mo 29 0.3867 0.3819 1.24 0.1702 GE800T Mo Rh 27 0.3946 0.3884 1.57 0.0610 GE800T Mo Rh 28 0.4239 0.4223 0.38 0.1270 GE800T Mo Rh 29 0.4356 0.4313 0.99 0.1382 GE800T Mo Rh 30 0.4480 0.4448 0.71 0.1570 GE800T Mo Rh 32 0.4653 0.4636 0.37 0.1755 GEDMR Mo Mo 26 0.3594 0.3559 0.97 0.1755 GEDMR Mo Mo 27 0.3578 0.3575 0.08 0.1304 GEDMR Mo Mo 28 0.3679 0.3667 0.33 0.1325 GEDMR Mo Rh 26 0.4179 0.4109 1.68 0.1852 GEDMR Mo Rh 27 0.4158 0.4125 0.79 0.1338 GEDMR Mo Rh 28 0.4481 0.4444 0.83 0.2210 GEDMR Mo Rh 30 0.4562 0.4516 1.01 0.2015
292 Table 8-2. Data collected in the retrospective study. Collected Data Descriptor Characteristics BI-RADSÂ® Assessment Category Cat 1 or Cat 2 (only) Note: Assigned by interpreting radiologist BI-RADSÂ® Density Category <25%, 25-50%, 51-75%, >75% Note: Assigned by interpreting radiologist Mammography Technician Code Anonymity maintained in study Interpreting Radiologist Code Anonymity maintained in study Individual Age If 89 years old of greater assigned " 89" Hormone replacement status Yes or No Radiographic Positioning Cranialcaudal (CC) and mediolateral oblique (MLO) views, left (L), right (R) Codes: LCC, LMLO, RCC, RMLO Radiographic Unit GE-800T or GE-DMR Compressed Breast Thickness Distance in millimeters (mm) Compression Pressure Pressure in decaNewtons (daN) Radiographic Unit Operating Mode AOP-S, AOP-C, AOP-D, AEC Tube Potential kVp Tube Current-Time Product mAs Focalspot size 0.1 mm or 0.3 mm Target Mo or Rh Filter Mo or Rh Positioning Angle Degrees
293Table 8-3. Craniocaudal population demographics of breast volumetric parameters and PN L, width and volume selected for MCNP-5 simulations. Compressed Breast Thickness (CBT) Range (cm) < 3 3-5 5-7 >7 MCNP-5 CBT (cm) 2 4 6 8 Population CBT (cm) 2.23 Â± 0.70 4.06 Â± 0.55 5.61 Â± 0.50 7.37 Â± 0.10 Population PNL (cm) 8.44 Â± 3.16 9.89 Â± 2.71 11.38 Â± 2.96 11.73 Â± 1.45 Population width (cm) 18.55 Â± 3.55 20.74 Â± 2.65 22.38 Â± 2.94 24.05 Â± 2.79 N 44 283 168 6 MCNP-5 Simulation Parameters PNL (cm) Width (cm) Volume (cm3) PNL (cm) Width (cm) Volume (cm3) PNL (cm) Width (cm) Volume (cm3) PNL (cm) Width (cm) Volume (cm3) 2 2 12 37.70 5 16 251.33 6 17 480.66 9 19 1074.42 1 5 15 117.81 7 18 395.84 8 19 716.28 10 21 1319.47 Mean 8 19 238.76 10 21 659.73 11 22 1140.4012 24 1809.56 1 12 22 414.69 13 23 939.34 14 25 1649.3413 27 2205.40 2 15 25 589.05 15 26 1225.2217 28 2243.1 15 30 2827.43
294 Table 8-4. Fitting parameters for entran ce skin exposure reference measurement. b mAs a kVp Xf ese ) (Re Mammographic unit T/F Tube Potential (kVp) a b Rsqr GE800T Mo/Mo 26 8.02 -2.19 0.99999 GE800T Mo/Mo 27 9.05 2.00 0.99999 GE800T Mo/Mo 28 10.19 2.67 0.99990 GE800T Mo/Mo 29 11.29 12.03 0.99997 GE800T Mo/Rh 27 7.57 11.69 0.99997 GE800T Mo/Rh 28 8.67 2.34 0.99988 GE800T Mo/Rh 29 9.70 8.78 0.99979 GE800T Mo/Rh 30 10.76 13.37 0.99990 GE800T Mo/Rh 32 13.30 -11.22 0.99999 GEDMR Mo/Mo 26 8.95 -3.39 1.00000 GEDMR Mo/Mo 27 10.13 -5.02 0.99997 GEDMR Mo/Mo 28 11.44 -6.54 0.99987 GEDMR Mo/Rh 26 7.02 -2.13 0.99996 GEDMR Mo/Rh 27 8.01 -1.12 1.00000 GEDMR Mo/Rh 28 9.15 -18.58 0.99998 GEDMR Mo/Rh 30 11.32 -12.94 0.99996
295Table 8-5. Study population descriptiv e statistics of Dg for the four me thods of estimating glandularity. Mean Std Dev Std. Error Range Max Min Median Method Size mGy mrad mGymrad mGymradmGymrad mGymrad mGymrad mGymrad BRTES-MOD 282 2.40 240.430.61 60.630.04 3.61 4.41 440.67 5.42 541.981.01 101.322.31 230.54 ACR BI-RADS 282 2.24 223.570.53 52.890.03 3.15 3.39 339.01 4.36 436.010.97 97.01 2.19 219.17 Histogram 282 2.42 242.230.63 62.630.04 3.73 4.69 469.46 5.72 571.961.02 102.492.36 236.24 ACR Standard 282 2.37 237.190.91 90.910.05 5.41 6.38 638.41 6.98 697.700.59 59.29 2.21 220.90
296 Table 8-6. The table lists the Parameters and Dg calculation using the BRTES-MOD DM method using equation (8-14). Description Equation (14) Symbol Mammogram Data Radiographic Position LCC RCC Age (years) 56 Mammography Unit GE-800T Target/Filter T/F Mo/MoMo/Mo Tube Potential (kVp) kVp 28 28 HVL (mm Al) HVL 0.37 0.37 Tube-current time product (mAs) mAs 90 99 Compressed Breast Thickness (cm) tb 3.3 3.4 Skin entrance exposure (mGy) Xese 8.02 8.85 Fibroglandular content (%) G 51.38 55.74 DgN (mGy mGy-1) DgN 0.26 0.27 PNL (cm) 6.15 6.51 Width (cm) 18.32 18.18 Factors product fpÂ·fvÂ·fa fpÂ·fvÂ·fa 1.05 1.06 BRTES-MOD Dg (mGy) Dg 2.15 2.50 Total CC Dg 4.65
297 CHAPTER 9 CONCLUSIONS AND FUTURE WORK Conclusions In this dissertation a dosimetry protocol was developed to provide patient-specific average glandular dose for clinical mammogr aphy. The protocol was designed to be performed in the period of time that it ta kes for a patient to go through a typical diagnostic mammography imaging session. Chapters 3 and 4 provided a comprehe nsive evaluation of MOSFET and FOCD dosimeters for use in measuring the entrance skin exposure, Xese, during mammography imaging. Both dosimeter types are clini cally applicable in the mammography energy range. However, FOCD performance was supe rior to that of the MOSFET in angular response, sensitivity, reproducibility, and lifes pan. The axial angular response in the MOSFET dosimeter showed reduced signal between 120Â° to 150Â° and 190Â° to 220Â° while the FOCD dosimeter showed no significant anisotropy. The MOSFET also displayed an unexpected behavior in which a device would randomly fail, to providing a measurement or display an extremely low measurement wh en compared to adjacent dosimeters of the same model. MOSFET dosimeter sensitivity showed a 9.52 Â± 8.28 % energy dependence for all target/filter combinations and dosimeter models tested while the FOCD dosimeter showed a 4.62 Â± 2.10 % energy dependence. The MOSFET dosimeter had an overall reproducibility of 25.95 Â± 8.28 % for all targe t/filter combinations and dosimeter models tested while the FOCD dosimeter had an ove rall reproducibility 1.07 Â± 0.68 %. Lastly,
298 the MOSFET dosimeter has a maximum lifespan of 20,000 mV accumulated dose, whereas the FOCD is reusable. Chapter 5 presented the homogeneous BRTES-MOD phantom, a breast tissue equivalent series of phantoms that closely mimic the radiological characteristics of ICRU-44 referenced tissues and anthropometri c dimensions of a compressed breast in mammography applications. The modifications performed to the BRTES phantom series improved the overall homogeneity by changi ng the type of microspheres used and reducing the incorporated air by mixing unde r vacuum. The BRTES-MOD also had the added benefit of having the same geometric dimensions of an Wu dosimetric phantom with a range of glandularities . Doses predicted on this series of phantoms are more representative of the patient populati on. The BRTES-MOD phantom series outperformed either acrylic or BR 12 when evaluated on Âµ -1 and Âµen -1 ratios to ICRU-44 reference tissues. Further refinement is n eeded for the 100% glandular and 100% adipose BRTES-MOD to further reduce the differen ces to ICRU-44 reference tissues. The BRTES-MOD phantom series provides the pl atform to perform patient glandularity referenced measurements and individual patien t doses in the clinical environment. A potential use for the BRTES-MOD phantom is to supplement the ACR phantom to evaluate AGD for a variety of glandularitie s providing a better assessment of the AGD that may be delivered to a variety of patients. In Chapter 6, a retrospective study was c onducted to determine the anthropometric measurements of an average breast using a mammogram. The goal of the retrospective study was to define the average breast and the distribution of parameters for a clinical population, the results of which are shown in ta bles 6-10 and 6-11. The data provided the
299 anthropometric data for developing realistic phantoms. In addition, the data provided evidence that the current NCRP, MQSA and ACR dosimetry-model linear dimensions are too small to accurately describe the study population. The retrospective study did not take into account several fact ors that might influence the anthropometric measurements of a population, such as race, menopausal status and body index measurements. The anthropometric measurements of th is study population did clearly show that adipose tissue increases with age and compre ssed breast thickness. The study presented evidence that CC and MLO are two distinct image groups. This makes it important to distinguish the two views in population studies in contrast with previous studies that combined the views as a single aggregate group.29,30,60,81,119,128 In this study, the CC and MLO views constituted two st atistically separate groups in compression pressure, compressed breast thickness, tube potent ial and tube current-time product. The underlying reason for these differences is the anatomical exclusion or inclusion of the pectoral muscle in CC and MLO views respectively. The current ACR model estimates the ge ometry of the breast in CC view as semielliptical, which was consistent with the data in this study. Equations 6-3 and 6-4 were derived to estimate the whole br east area of CC and MLO views based on measurements of the PNL, width, CCD, a nd axilla. The equations facilitated the development of anatomically correct mammography phantoms. The estimated measurement of the skin la yer indicates that current models that assume a 0.4 cm skin layer need to be revised. The smaller skin layer of 0.2 cm found in this study would result in lower glandular dos es when current dose conversion factors are used.
300 In addition, the assessment of the most probable location of the dense part of the breast allows the determination of the optimal position for the AEC sensor with respect to the dense tissue for the patients for whom previous films do not exist or cannot be accessed. The most probable location was defi ned with respect to the PNL. The PNL Ratio was derived by dividing the PNL distan ce by the distance of the dense tissue region centroid from the chest wall. The use of the PNL Ratio could directly contribute to reducing retake rates for new patients. The retrospective study clearly demonstrated the potential benefit of characterizing patient populations to define anthropometric parameters for the manufacturing of anatomically-correct phantoms. Chapter 7 compared several methods of estimating glandularity using the study population of Chapter 6 and the BRTES-MOD phan tom series of Chapter 5. In this study, the BRTES-MOD TET method was introd uced as a quantitative method of estimating patient glandularity. This method was compared to the ACR BI-RADS, PM, HTM, and TLM methods for a single study population. Mammography dosimetry requires a reproducib le and reliable method of estimating breast glandularity to establish a docum entable standard. The BRTES-MOD TET method is proposed as the method that could provide an estimate of glandularity that is traceable to a standard tissue equivalent phantom. The advantage of the BRTES-MOD TET met hod of over ACR BI-RADS is that it is quantitative and does not rely on the us er's training and ex perience to predict glandularity. Even though pl animetry method (PM) was designed to be a quantitative measure of the user's training and experience, it also relied on th e user and reduced the range of glandularity to a binary choice of 100% glandular or 100% adipose. The HTM
301 improved reproducibility over the PM met hod, but it still redu ced the range of glandularity to a binary choice. The TLM wa s the first non-user-dependent quantitative system, but in this study the underlying pr emise that AEC systems provide unique exposures based on glandularity and breast thic kness turned out to be false. TLM was unable to reliably predict breast glandular ities of small (<3cm) and large (>7 cm) compressed breasts. The BRTES-MOD TET method does require care ful application when applied to screen-film mammography, in or der to ensure that sensitom etry does not interfere with the method results. The issue is avoided when the phantom images are taken immediately after patient imaging. Applica tion of this method in a digital environment would reduce the number of step s required to achieve a result , since it would not require the user to digitize the images. Also furt her development is needed in adapting the method to enable the applica tion to MLO views, which in clude the pectoral muscle. Even though not directly compared in th is study, the advantages of the BRTESMOD TET method over the volumetri c method proposed by Pawluczyk et al. is that each mammogram has an individual calibration inst ead of using a single calibration curve as proposed by Pawluczyk et al. 47 The BRTES-MOD TET method also did not require correction for field nonuniformity effects. Chapter 8 integrates the BRTES-MOD method steps from the previous chapters to evaluate the CC-view average glandular dos e which was then compared to commonly used methods of determining AGD. In this ch apter, a method was described that utilizes a homogeneous phantom, the BRTES-MOD, and Monte Carlo simulations to determine the average glandular dose of the breast . The method adjusts for the use of the
302 homogeneous phantom (fp), breast volume (fv), and anatomical features (fa) that may contribute to Dg. The study showed that the use of a homogeneous tissue phantom to measure glandularity can result in underestimating Dg by as much as 11% for a compressed breast thickness of 2 cm and glandul arity of 90%. The range of the impact of breast volume had on Dg was 0.3% decrease to 2.7% increase as a function of compressed breast thickness when compared to the Dg of the Wu DM phantom. Anatomical features such as retromammary ad ipose, pectoral muscle, ribs and intercostal muscles would affect Dg only if misalignment of the x-ray fi eld is an issue; otherwise, it would affect Dg by approximately 1%. In evalua ting the clinical population, only fp, and fv were utilized because the fa effect on Dg was negligible in magnitude to other uncertainties. Even though the BRTES-MOD method Dg was found to have similar trend in Dg as the ACR BI-RADS and HTM methods, it provides a quantitative method that eliminates the variations resulting individua l judgment that is i nherent to the other systems. The application of the BRTES-MOD protocol in a clinical setting would proceed in the following manner. The patient arrive s and receives the typical mammographic examination. During the positioning of th e patient, the technologist would place dosimeters on the underside of the compression paddle, adjacent to but not in contact with breast to measure the entr ance skin exposure as described in Chapters 3 and 4. After the patients examination; the technologist images the BRTES_MOD step phantom and processes all images. The digitalized ma mmograms would be automatically segmented as described in Chapter 6. Glandularity would be estimated from the digitized
303 mammogram and step phantom images as described in Chapter 7. Tabulated values or a MCNP-5 simulation using the patient-specific measurements of estimated glandularity, compressed breast thickness, PNL, width and mammography unit parameters would be used to estimate the fa, fv, and fa as describe in Chapter 8. The Dg would be calculated using equation (14) and documented as part of the mammogram record. The collected information would provide the data needed for epidemiological studies that would improve BEIR V radiation-risk models, gi ven that mammography represents a healthy population that receive a re occurring x-ray exposure. Future Work The protocol presented in this dissertation provides the framework necessary to measure individual average glandular dos e during a clinical mammography imaging session. This method has been developed and demonstrated for two models of GE mammography units (800T and DMR), and future work can assess the applicability to other models of mammography units. The logical progression of the protocol development will be its applicability in a digital environment and evaluating it in a prospective clinical study. Developments in the areas of dosimetry, population studies, phantom construction, estimating glandularity and average glandular dose calculation will further build upon the protocol developed in this dissertation in order to make it applicable in the digital mammography environment. Dosimetry The dosimeters evaluated in Chapters 3 and 4 provide a platform to implement the protocol in a clinical application. However, with the emergence of digital mammography, further studies are needed to ev aluate the capability of using the exposure to the computed radiography plates or the di rect digital radiography image receptors to
304 measure the entrance skin exposure. One a dvantage to the method described in this dissertation would be not havi ng a foreign object in the im age field; a second would be the linear dose-response of the detector. Phantom Manufacturing The phantom developed in Chapter 5 provided a reliable methodology to manufacture tissue-equivalent ph antoms. However, further res earch is needed in scaling up the manufacturing techniques to reduce manufacturing times. Population Demographic and Anthropometric Studies The retrospective study clearly demonstrated the potential benef it of characterizing patient populations to define anthropometric parameters for manufacturing anatomicallycorrect phantoms and to provide insights into the effect anatomical features have on dosimetry. However, the retrospective study conducted in Chapter 6 lacked the demographic information that might influen ce some anthropometric measurements of the breast, such as body mass index and race. A future prospective study would enable the collection of the demographic data in conjunc tion with anthropometric measurements to enable an average breast to be defined for more speci fic mammography populations. The demographic data would provide the informa tion necessary to estimate the applicability of phantoms to screening mammography populati ons nationwide. The prospective study size needs to ensure that sufficient numbers of large (>7 cm) and small (<3 cm) breasts are included to improve on the current retr ospective study anthropometric measurements standard deviations. In addition, the current dosimetric models used by the NCRP and ACR were developed with assumptions that have not been verified in population studies prior to this study. Some of the important dosimetry fact ors include lateral skin layer thickness and
305 breast area measurements. Further research is needed to explore the use of magnetic resonance mammography or dedicated breast CT to measure the biological structures in the breast to develop a thr ee-dimensional anatomy that c ould be adopted in epoxy-resin phantoms and dosimetric models. These biological structures could provide the ability to refine measurements of glandularity and calculate dose down to the radiosensitive glandular unit, TDLU. Breast Tissue Fibroglandular Content The fibroglandular content estimates used in Chapter 7 were based on digitized screen-film mammography. Exte nsion of this protocol to the purely digital environment would enable the adoption of adjustable contrast, linear exposure response, and correction for image inhomogeneities relating to heel effect, inverse-square law effects and varying compressed thickness of the brea st. An added benefit of the digital environment would be to eliminate the film processing issues discussed in Chapter 7. Individual Average Glandular Dose The average glandular dose calculated in Chapter 8 showed the potential application of Monte Carlo simulations for clinical mammography. Monte Carlo model development of the fine structures in the breast would enable dose calculations down to the basic radiosensitive unit of the breast, for example, the fibroglandular tissues. Another aspect on this resear ch that has a direct impact on the AGD is the skin layer thickness. Given the measurements made in this study, new dose conversion tables will be needed. Dissertation Derived Peer Reviewed Journal Articles The dissertation will be publis hed in five separate peer reviewed journal articles. The status of each journal article is as follows:
306 1. Characterization of MOSFET Dosimeters for Application in Clinical Mammography journal article was submitted to Medical Physics and currently in review process. 2. Characterization of Fiber-Optic-Coupled Detector for Dosimetry in Clinical Mammography journal article has been written and is currently being reviewed by the authors. 3. Anthropometric Variations in Mammography journal article has been written and is currently being review ed by the authors. 4. Estimating Breast Glandularity journal article has been written and is currently being reviewed by the authors. 5. Average Glandular Dose Based on Homogeneous Phantom journal article has been written and is currently being reviewed by the authors. The soon-to-be-published BEIR VII: Health Risks Exposures from Exposure to Low Levels of Ionizing Radiation still indicates that it will rely on small studies (16-17,202 cases) that have relatively high average doses (0.11 5.8 Gy) when compared to screening mammography.141 This dissertation provides the ability to document the individual specific doses providing future ep idemiological studies the data needed to improve the methods of estimating the risk a ssociated with ionizing radiation used in mammography.
307 APPENDIX MCNP-5 INPUT FILES This appendix delineates the MCNP-5 input files used thorough the current study to evaluate spectra half-value laye r, effects of tissue layering, fp, fv, and fa . The specific details of the MCNP-5 input file is delineate d in the chapter 3 of user's manual that is furnished along with the software.10 MCNP-5 was operated on Windows XP operating system32 . The general format used for MCNP-5 input files is listed below; Title Card-File description Cell CardsMaterial Ca rd and Geometry Reference Blank Line Delimiter-required Surface Cards-Geometry Blank Line Delimiter-required Data Cards X-ray Source Definition X-ray Source Spectra, derived from SRS-78 Vol Cardvolume cards TallyMeasurements such as fluence and dose MeshPhoton fluence measured in structured mesh Material Card-elemental compositions of materials Nps-histories to evaluate Blank Line Terminator-required The following pages list representative sa mple of all input files used in this research followed by the x-rayenergy spectra histograms and material cards from used 32 Microsoft Corporation, One Microsoft Way, Redmond, Washington 98052-6399
308 Free in Air Measurement MCNP-5 Input File Figure B-1 illustrates geometry us ed for free-in-air measurements. C ****Cell Cards**Free in air Evaluation *************** C Cylinder Encircling World 5 2 -0.001205 -21 #10 #11 #12 #13 IMP:P 1 C Void Outside World 7 0 21 IMP:P 0 C Collimator 10 0 -20 #11 IMP:P 0 11 2 -0.001205 -19 IMP:P 1 C Compression Paddle 12 2 -0.001205 -31 IMP:P 1 C Detector 13 2 -0.001205 -32 IMP:P 1 C Surface Cards C Collimator 19 rpp -5 5 -12.5 -5.19 47.3625 47.68 20 rcc 0 0 47.3625 0 0 0.3175 30 C Cylinder Emcopassing World 21 rcc 0 0 -10 0 0 110 30 C Compression Paddle 31 rcc 0 0 45 0 0 0.3175 30 C Free in Air detect or 6cc 0.00025in 0.7mg/cm^2 32 rcc 0 -11.1180234 0 0 0 1 1.381976598 SDEF DIR d1 POS 0 -12.0 66.00 ERG d2 PAR 2 Vec 0 0 -1 mode p C -1, Cos of angle, 1 (angle is 22.5 deg) Si1 -1 0.923879533 1 C 0, (1+costheta)/2, (1-costheta)/2 Sp1 0 0 1 # Si2 Sp2 C Spectra Energy Histogram Derived From SRS-78 C Talley Cards Vol 5J 6.0 C Talley for Dose in F6-MeV/gm F6:p 13 C Material Cards nps 100000000
309 Free in Air Measurement with Al uminum MCNP-5 Input File Figure B-1 illustrates geometry us ed for free-in-air measurements. C ****Cell Cards**Free in Air w/Al*************** C Cylinder Encircling World 5 2 -0.001205 -21 #10 #11 #12 #13 #14 IMP:P 1 C Void Outside World 7 0 21 IMP:P 0 C Collimator 10 0 -20 #11 IMP:P 0 11 2 -0.001205 -19 IMP:P 1 C Compression Paddle 12 2 -0.001205 -31 IMP:P 1 C Detector 13 2 -0.001205 -32 IMP:P 1 C HVL Filter 14 8 -2.7 -33 IMP:P 1 C Surface Cards C Collimator 19 rpp -5 5 -12.5 -5.19 47.3625 47.68 20 rcc 0 0 47.3625 0 0 0.3175 30 C Cylinder Emcopassing World 21 rcc 0 0 -10 0 0 110 30 C Compression Paddle 31 rcc 0 0 45 0 0 0.3175 30 C Free in Air detect or 6cc 0.00025in 0.7mg/cm^2 32 rcc 0 -11.1180234 0 0 0 1 1.381976598 C HVL Al Filter in mm thickness above Cell 12 33 rcc 0 0 46 0 0 0.03900 30 SDEF DIR d1 POS 0 -12.0 66.00 ERG d2 PAR 2 Vec 0 0 -1 mode p C -1, Cos of angle, 1 (angle is 22.5 deg) Si1 -1 0.923879533 1 C 0, (1+costheta)/2, (1-costheta)/2 Sp1 0 0 1 # Si2 Sp2 C Spectra Energy Histogram Derived From SRS-78 C Talley Cards Vol 5J 6.0 C Talley for Dose in F6-MeV/gm F6:p 13 C Material Cards nps 10000000
310 Phantom Factor MCNP-5 Input for Wu and BRTES-MOD phantoms Figure B-2 illustrates geometry used for phantom factor evaluations. C ****Cell Cards**Phantom Factor*************** C Cylinder Encircling World 5 2 -0.001205 -21 #8 #9 #10 #11 #12 #13 IMP:P 1 C Void Outside World 7 0 21 IMP:P 0 C Breast Simulator 8 5 -0.9500 -1 3 -5 4 #9 IMP:P 1 9 7 -0.9964 -6 3 -9 8 IMP:P 1 C Collimator 10 0 -20 #11 IMP:P 0 11 2 -0.001205 -19 IMP:P 1 C Compression Paddle 12 2 -0.001205 -31 IMP:P 1 C Detector 13 2 -0.001205 -32 IMP:P 1 C Surface Cards C Breast Simullator 2cm C Exterior 0.40 cm layer usually 100pct adipose 1 sq 64 81 0 0 0 0 -5184 0 -12 0 3 py -12.0 4 pz -2.2 5 pz -0.2 C Interior main phantom homogenous contents 6 sq 57.76 73.96 0 0 0 0 -4271.9296 0 -12 0 8 pz -1.8 9 pz -0.6 C Collimator 19 rpp -5 5 -12.5 -5.19 47.3625 47.68 20 rcc 0 0 47.3625 0 0 0.3175 30 C Cylinder Emcopassing World 21 rcc 0 0 -10 0 0 110 30 C Compression Paddle 31 rcc 0 0 45 0 0 0.3175 30 C Free in Air detect or 6cc 0.00025in 0.7mg/cm^2 32 rcc 0 -8 -3.2 0 0 1 1.381976598 SDEF DIR d1 POS 0 -12.0 63.80 ERG d2 PAR 2 Vec 0 0 -1 mode p C -1, Cos of angle, 1 (angle is 22.5 deg) Si1 -1 0.923879533 1 C 0, (1+costheta)/2, (1-costheta)/2 Sp1 0 0 1
311 # Si2 Sp2 C Spectra Energy Histogram Derived From SRS-78 C Talley Cards Vol 2J 102.99 123.20 3J 6.0 C Talley for Dose in F6-MeV/gm, F4-mean Photon/cm^2 F6:p 8 9 13 F4:p 13 C Material Cards nps 100000000
312 Volumetric Factor MCNP-5 Input Figure B-2 illustrates geom etry used for volumetric factor evaluations. C ****Cell Cards**BreastPhantom_v02242004*************** C Cylinder Encircling World 5 2 -0.001205 -21 #8 #9 #10 #11 #12 #13 IMP:P 1 C Void Outside World 7 0 21 IMP:P 0 C Breast Simulator 8 5 -0.9500 -1 3 -5 4 #9 IMP:P 1 9 7 -0.9964 -6 3 -9 8 IMP:P 1 C Collimator 10 0 -20 #11 IMP:P 0 11 2 -0.001205 -19 IMP:P 1 C Compression Paddle 12 2 -0.001205 -31 IMP:P 1 C Detector 13 2 -0.001205 -32 IMP:P 1 C Surface Cards C Breast Simullator 2cm C Exterior 0.40 cm layer usually 100pct adipose 1 sq 64 81 0 0 0 0 -5184 0 -12 0 3 py -12.0 4 pz -2.2 5 pz -0.2 C Interior main phantom homogenous contents 6 sq 57.76 73.96 0 0 0 0 -4271.9296 0 -12 0 8 pz -1.8 9 pz -0.6 C Collimator 19 rpp -5 5 -12.5 -5.19 47.3625 47.68 20 rcc 0 0 47.3625 0 0 0.3175 30 C Cylinder Emcopassing World 21 rcc 0 0 -10 0 0 110 30 C Compression Paddle 31 rcc 0 0 45 0 0 0.3175 30 C Free in Air detect or 6cc 0.00025in 0.7mg/cm^2 32 rcc 0 -8 -3.2 0 0 1 1.381976598 SDEF DIR d1 POS 0 -12.0 63.80 ERG d2 PAR 2 Vec 0 0 -1 mode p C -1, Cos of angle, 1 (angle is 22.5 deg) Si1 -1 0.923879533 1 C 0, (1+costheta)/2, (1-costheta)/2 Sp1 0 0 1
313 # Si2 Sp2 C Spectra Energy Histogram Derived From SRS-78 C Talley Cards Vol 2J 102.99 123.20 3J 6.0 C Talley for Dose in F6-MeV/gm, F4-mean Photon/cm^2 F6:p 8 9 13 C Material Cards nps 100000000
314 Anatomical Factor MCNP-5 Input Figure B-3 illustrates geom etry used for anatomical factor evaluations. C ****Cell Cards**Anantomy Factor*************** C Cylinder Encircling World 5 2 -0.001205 -21 #8 #9 #10 #11 #12 #13 #14 #15 #16 #17 #18 #19 #20 #21 IMP:P 1 C Void Outside World 7 0 21 IMP:P 0 C Breast Simulator 8 5 -0.9500 -1 3 -5 4 #9 IMP:P 1 9 7 -0.9964 -6 3 -9 8 IMP:P 1 C Collimator 10 0 -20 #11 IMP:P 0 11 2 -0.001205 -19 IMP:P 1 C Compression Paddle 12 2 -0.001205 -31 IMP:P 1 C Detector 13 2 -0.001205 -32 IMP:P 1 C Tissue Slab 14 9 -1.050 -33 IMP:P 1 15 10 -3.225 -34 IMP:P 1 16 like 15 but trcl=(0 0 2.6) 17 9 -1.050 -35 IMP:P 1 18 like 17 but trcl=(0 0 2.6) 19 like 15 but trcl=(0 0 5.2) 20 like 17 but trcl=(0 0 5.2) C Subcutaneous fat 21 5 -0.9500 -36 IMP:P 1 C Surface Cards C Breast Simullator 2cm C Exterior 0.40 cm layer usually 100pct adipose 1 sq 64 81 0 0 0 0 -5184 0 -12 0 3 py -12.0 4 pz -2.2 5 pz -0.2 C Interior main phantom homogenous contents 6 sq 57.76 73.96 0 0 0 0 -4271.9296 0 -12 0 8 pz -1.8 9 pz -0.6 C Collimator 19 rpp -5 5 -12.5 -4.69 47.3625 47.68 20 rcc 0 0 47.3625 0 0 0.3175 30 C Cylinder Emcopassing World 21 rcc 0 0 -10 0 0 110 30
315 C Compression Paddle 31 rcc 0 0 45 0 0 0.3175 30 C Free in Air detect or 6cc 0.00025in 0.7mg/cm^2 32 rcc 0 -8 -3.2 0 0 1 1.381976598 C Anatomical Slab of Tissue C xx rpp Xmin Xmax Ymin Ymax Zmin Zmax 33 rpp -20 10 -13.4 -12.4 -2.2 5.6 C Second slab made of bone 34 rpp -20 10 -14.1 -13.4 -1.2 0.4 C Intercostal Muscle 35 rpp -20 10 -14.1 -13.4 -2.2 -1.2 C Subcutaneous adipose 0.4 cm 36 rpp -20 10 -12.4 -12 -2.2 5.6 SDEF DIR d1 POS 0 -12.0 63.80 ERG d2 PAR 2 Vec 0 0 -1 mode P C -1, Cos of angle, 1 (angle is 22.5 deg) Si1 -1 0.923879533 1 C 0, (1+costheta)/2, (1-costheta)/2 Sp1 0 0 1 # Si2 Sp2 C Talley Cards Vol 2J 102.99 123.20 3J 6.0 7J C Talley for Dose in F6-MeV/gm, F4-mean Photon/cm^2 F6:p 8 9 13 F4:p 8 9 13 C Material Cards nps 10000000
316 Energy Spectra Histogram for MCNP-5 Input # Si2 Sp2 X-ray energy spectra used in simulations was placed after "# Si2 Sp2 " section in the MCNP-5 input files. The following pages lists the spectra shown in Table A-1. Table A-1. Spectra used in population evaluation and MCNP-5 simulations. Mammographic Unit Tube Target Filter kVp HVL Measured (mm Al) HVL MCNP-5 (mm Al) Diff(%) GE800T Mo Mo 26 0.3577 0.3562 0.42 GE800T Mo Mo 27 0.3578 0.3575 0.08 GE800T Mo Mo 28 0.3680 0.3668 0.33 GE800T Mo Mo 29 0.3867 0.3819 1.24 GE800T Mo Rh 27 0.3946 0.3884 1.57 GE800T Mo Rh 28 0.4239 0.4223 0.38 GE800T Mo Rh 29 0.4356 0.4313 0.99 GE800T Mo Rh 30 0.4480 0.4448 0.71 GE800T Mo Rh 32 0.4653 0.4636 0.37 GEDMR Mo Mo 25 0.3476 0.3462 0.40 GEDMR Mo Mo 26 0.3594 0.3559 0.97 GEDMR Mo Mo 27 0.3578 0.3575 0.08 GEDMR Mo Mo 28 0.3679 0.3667 0.33 GEDMR Mo Rh 26 0.4179 0.4109 1.68 GEDMR Mo Rh 27 0.4158 0.4125 0.79 GEDMR Mo Rh 28 0.4481 0.4444 0.83 GEDMR Mo Rh 30 0.4562 0.4516 1.01
317 C **************************************************** C * Source = IPEM Rep 78_ 026K20D0M2* Mo/Mo 26 kVp * C * Tube Potential = 26 kVp * C * Filter(mm)=>0.69 Be_ 0.03 Mo_ 2.5 PMMA_ 0.1690 AL* C * Mean Photon = 16.8 kev * C * Air Kerma = 4.499E01 uGy per mAs @ 750mm * C * HVL = 0.3577 mm Al * C * Anode Angle = 20 degrees * C **************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 3.6573100000E-35 0.0040 3.7560000000E-23 0.0045 1.4994800000E-15 0.0050 1.6371700000E-10 0.0055 4.7164500000E-07 0.0060 1.3545300000E-04 0.0065 8.3045800000E-03 0.0070 1.7318400000E-01 0.0075 1.6994600000E+00 0.0080 9.5790000000E+00 0.0085 3.8683400000E+01 0.0090 1.1252300000E+02 0.0095 2.6545300000E+02 0.0100 5.3234100000E+02 0.0105 9.2431100000E+02 0.0110 1.4462100000E+03 0.0115 2.0740100000E+03 0.0120 2.7876200000E+03 0.0125 3.5464000000E+03 0.0130 4.3268000000E+03 0.0135 5.0583400000E+03 0.0140 5.7303500000E+03 0.0145 6.3590400000E+03 0.0150 6.9110300000E+03 0.0155 7.3392800000E+03 0.0160 7.6936100000E+03 0.0165 7.9371500000E+03 0.0170 8.0957200000E+03 0.0175 4.7047900000E+04 0.0180 8.0735000000E+03
318 0.0185 7.9337400000E+03 0.0190 7.7382600000E+03 0.0195 1.5146300000E+04 0.0200 8.7274500000E+02 0.0205 8.1385200000E+02 0.0210 7.7995600000E+02 0.0215 7.5666700000E+02 0.0220 7.3592900000E+02 0.0225 7.1454300000E+02 0.0230 6.8122600000E+02 0.0235 6.3238400000E+02 0.0240 5.6553100000E+02 0.0245 4.7132800000E+02 0.0250 3.4624000000E+02 0.0255 1.8937100000E+02 0.0260 1.7496600000E+01
319 C **************************************************** C * Source = IPEM Rep 78_ 027K20D0M2* 27 kVp Mo/Mo * C * Tube Potential = 27 kVp * C * Filtered => 0.69 Be_ 0.03 Mo_ 2.5 PMMA_ 0.1304 Al* C * Mean Photon = 16.9 kev * C * Air Kerma = 5.563E01 uGy per mAs @ 750mm * C * HVL = 0.3578 mm Al * C * Anode Angle = 20 degrees * C * *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 7.9815700000E-33 0.0040 1.5681600000E-21 0.0045 2.2248400000E-14 0.0050 1.2200700000E-09 0.0055 2.2210100000E-06 0.0060 4.5642100000E-04 0.0065 2.2125800000E-02 0.0070 3.8664200000E-01 0.0075 3.3183400000E+00 0.0080 1.6869200000E+01 0.0085 6.2855300000E+01 0.0090 1.7145900000E+02 0.0095 3.8416600000E+02 0.0100 7.3868100000E+02 0.0105 1.2390200000E+03 0.0110 1.8844500000E+03 0.0115 2.6415200000E+03 0.0120 3.4797000000E+03 0.0125 4.3563500000E+03 0.0130 5.2397700000E+03 0.0135 6.0561800000E+03 0.0140 6.7981800000E+03 0.0145 7.4898800000E+03 0.0150 8.0901000000E+03 0.0155 8.5502900000E+03 0.0160 8.9310100000E+03 0.0165 9.1935300000E+03 0.0170 9.3655500000E+03 0.0175 6.2273700000E+04 0.0180 9.3600400000E+03
320 0.0185 9.2196500000E+03 0.0190 9.0267100000E+03 0.0195 1.9057600000E+04 0.0200 1.0226000000E+03 0.0205 9.6048300000E+02 0.0210 9.3301700000E+02 0.0215 9.2066400000E+02 0.0220 9.1518300000E+02 0.0225 9.1401300000E+02 0.0230 9.0184800000E+02 0.0235 8.8088100000E+02 0.0240 8.3999000000E+02 0.0245 7.7806400000E+02 0.0250 6.8755000000E+02 0.0255 5.6987600000E+02 0.0260 4.1766000000E+02 0.0265 2.2465400000E+02 0.0270 1.7094100000E+01
321 Mo/Mo C *************************************************** C * Source = IPEM Rep 78_ 028K20D0M2 28 kVp Mo/Mo * C * Tube Potential = 28 kVp * C * Filters(mm)=>0.69 Be_0.03 Mo_2.5 PMMA_0.1330 Al * C * Mean Photon = 17.1 kev * C * Air Kerma = 6.218E01 uG y per mAs @ 750mm * C * HVL = 0..3680 mm Al * C * Anode Angle = 20 degrees * C *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 5.4238000000E-33 0.0040 1.1949100000E-21 0.0045 1.8302400000E-14 0.0050 1.0608100000E-09 0.0055 2.0112600000E-06 0.0060 4.2672300000E-04 0.0065 2.1186800000E-02 0.0070 3.7738800000E-01 0.0075 3.2884000000E+00 0.0080 1.6916500000E+01 0.0085 6.3659600000E+01 0.0090 1.7502800000E+02 0.0095 3.9474700000E+02 0.0100 7.6318500000E+02 0.0105 1.2861000000E+03 0.0110 1.9643900000E+03 0.0115 2.7621800000E+03 0.0120 3.6501700000E+03 0.0125 4.5822200000E+03 0.0130 5.5271400000E+03 0.0135 6.4038200000E+03 0.0140 7.2064600000E+03 0.0145 7.9563000000E+03 0.0150 8.6179900000E+03 0.0155 9.1360800000E+03 0.0160 9.5676300000E+03 0.0165 9.8841000000E+03 0.0170 1.0108500000E+04 0.0175 7.5781600000E+04
322 0.0180 1.0186600000E+04 0.0185 1.0088300000E+04 0.0190 9.9372100000E+03 0.0195 2.2490100000E+04 0.0200 1.1293200000E+03 0.0205 1.0699400000E+03 0.0210 1.0508500000E+03 0.0215 1.0504100000E+03 0.0220 1.0604100000E+03 0.0225 1.0787600000E+03 0.0230 1.0891800000E+03 0.0235 1.0949900000E+03 0.0240 1.0877200000E+03 0.0245 1.0570100000E+03 0.0250 1.0039100000E+03 0.0255 9.2420200000E+02 0.0260 8.1197300000E+02 0.0265 6.6070100000E+02 0.0270 4.7838300000E+02 0.0275 2.5661000000E+02 0.0280 2.1258500000E+01
323 C **************************************************** C * Source = IPEM Rep 78_ 029K20D0M2* 29 kVp Mo/Mo * C * Tube Potential = 29 kVp * C * Filtered => 0.69 Be_ 0.03 Mo_ 2.5 PMMA_ 0.1702 Al* C * Mean Photon = 17.3 kev * C * Air Kerma = 6.425E01 uGy per mAs @ 750mm * C * HVL = 0.3867 mm Al * C * Anode Angle = 20 degrees * C * ************************************************** C Mev Probability* 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 2.8944500000E-35 0.0040 3.1578400000E-23 0.0045 1.3267900000E-15 0.0050 1.5187300000E-10 0.0055 4.5622400000E-07 0.0060 1.3595200000E-04 0.0065 8.5961600000E-03 0.0070 1.8386300000E-01 0.0075 1.8437300000E+00 0.0080 1.0577700000E+01 0.0085 4.3351100000E+01 0.0090 1.2766700000E+02 0.0095 3.0426700000E+02 0.0100 6.1574600000E+02 0.0105 1.0771000000E+03 0.0110 1.6963500000E+03 0.0115 2.4487700000E+03 0.0120 3.3091400000E+03 0.0125 4.2336000000E+03 0.0130 5.1907000000E+03 0.0135 6.1019700000E+03 0.0140 6.9481400000E+03 0.0145 7.7593600000E+03 0.0150 8.4877300000E+03 0.0155 9.0794400000E+03 0.0160 9.5955700000E+03 0.0165 9.9902800000E+03 0.0170 1.0286800000E+04 0.0175 8.6062400000E+04 0.0180 1.0523500000E+04
324 0.0185 1.0506500000E+04 0.0190 1.0435400000E+04 0.0195 2.5215900000E+04 0.0200 1.1934700000E+03 0.0205 1.1381800000E+03 0.0210 1.1304300000E+03 0.0215 1.1419000000E+03 0.0220 1.1687300000E+03 0.0225 1.2076800000E+03 0.0230 1.2390600000E+03 0.0235 1.2725300000E+03 0.0240 1.2902600000E+03 0.0245 1.2983400000E+03 0.0250 1.2757500000E+03 0.0255 1.2394600000E+03 0.0260 1.1689000000E+03 0.0265 1.0616100000E+03 0.0270 9.1816900000E+02 0.0275 7.4362600000E+02 0.0280 5.2684200000E+02 0.0285 2.7881300000E+02 0.0290 1.6950600000E+01
325 C ***************************************************** C * Source = IPEM Rep 78_ 027K20D0M2* 27 kVp Mo/Rh * C * Tube Potential = 27 kVp * C * Filtered => 0.69 Be_ 0.025 Rh_ 2.5 PMMA_ 0.0610 Al* C * Mean Photon = 17.7 kev * C * Air Kerma = 5.802E01 uGy per mAs @ 750mm * C * HVL = 0.3946 mm Al * C * Anode Angle = 20 degrees * C * *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 1.8715000000E-32 0.0040 1.8655500000E-21 0.0045 2.2793300000E-14 0.0050 1.0816400000E-09 0.0055 1.8471500000E-06 0.0060 3.5685900000E-04 0.0065 1.7358700000E-02 0.0070 3.0358900000E-01 0.0075 2.7008400000E+00 0.0080 1.4471300000E+01 0.0085 5.2155400000E+01 0.0090 1.4540400000E+02 0.0095 3.2975500000E+02 0.0100 6.4288600000E+02 0.0105 1.0865800000E+03 0.0110 1.6655800000E+03 0.0115 2.3642400000E+03 0.0120 3.1383400000E+03 0.0125 3.9608700000E+03 0.0130 4.7872000000E+03 0.0135 5.5762300000E+03 0.0140 6.3138800000E+03 0.0145 6.9845900000E+03 0.0150 7.5430900000E+03 0.0155 8.0314100000E+03 0.0160 8.4110500000E+03 0.0165 8.6928200000E+03 0.0170 8.8694400000E+03 0.0175 5.9311300000E+04 0.0180 8.9437800000E+03
326 0.0185 8.8432300000E+03 0.0190 8.6708700000E+03 0.0195 1.8339100000E+04 0.0200 7.5858500000E+03 0.0205 7.1874200000E+03 0.0210 6.8772700000E+03 0.0215 6.4892900000E+03 0.0220 6.0595100000E+03 0.0225 5.6023100000E+03 0.0230 5.0712300000E+03 0.0235 9.0868900000E+02 0.0240 8.5689400000E+02 0.0245 7.8035300000E+02 0.0250 6.7794700000E+02 0.0255 5.4902500000E+02 0.0260 3.9436200000E+02 0.0265 2.0724900000E+02 0.0270 1.5546800000E+01
327 C *************************************************** C * Source = IPEM Rep 78_ 028K20D0M2 28 kVp Mo/Rh * C * Tube Potential = 28 kVp * C * Filters(mm)=>0.69 Be_0.025 Rh_2.5 PMMA_0.1270 Al * C * Mean Photon = 18.0 kev * C * Air Kerma = 5.757E01 uG y per mAs @ 750mm * C * HVL = 0.4239 mm Al * C * Anode Angle = 20 degrees * C *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 1.7654400000E-36 0.0040 3.0027600000E-24 0.0045 2.1924900000E-16 0.0050 3.4600000000E-11 0.0055 1.3287000000E-07 0.0060 4.6624400000E-05 0.0065 3.4610700000E-03 0.0070 8.3383400000E-02 0.0075 9.4743900000E-01 0.0080 6.1349000000E+00 0.0085 2.5657000000E+01 0.0090 8.0580400000E+01 0.0095 2.0106200000E+02 0.0100 4.2424100000E+02 0.0105 7.6480800000E+02 0.0110 1.2372800000E+03 0.0115 1.8356500000E+03 0.0120 2.5338600000E+03 0.0125 3.3011600000E+03 0.0130 4.1052800000E+03 0.0135 4.9013800000E+03 0.0140 5.6654900000E+03 0.0145 6.3836100000E+03 0.0150 7.0122600000E+03 0.0155 7.5831700000E+03 0.0160 8.0510600000E+03 0.0165 8.4308700000E+03 0.0170 8.7117700000E+03 0.0175 6.6180500000E+04 0.0180 8.9862300000E+03
328 0.0185 8.9885700000E+03 0.0190 8.9154500000E+03 0.0195 2.0314500000E+04 0.0200 7.8986200000E+03 0.0205 7.5785400000E+03 0.0210 7.3594900000E+03 0.0215 7.0574500000E+03 0.0220 6.7132900000E+03 0.0225 6.3395100000E+03 0.0230 5.8872500000E+03 0.0235 1.0882000000E+03 0.0240 1.0713600000E+03 0.0245 1.0255100000E+03 0.0250 9.5920900000E+02 0.0255 8.6426800000E+02 0.0260 7.4533700000E+02 0.0265 5.9335700000E+02 0.0270 4.2412700000E+02 0.0275 2.2388100000E+02 0.0280 1.8359500000E+01
329 C ***************************************************** C * Source = IPEM Rep 78_ 029K20D0M2* 29 kVp Mo/Rh * C * Tube Potential = 29 kVp * C * Filtered => 0.69 Be_ 0.025 Rh_ 2.5 PMMA_ 0.1382 Al* C * Mean Photon = 18.1 kev * C * Air Kerma = 6.317E01 uGy per mAs @ 750mm * C * HVL = 0.4356 mm Al * C * Anode Angle = 20 degrees * C * *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 3.5981300000E-37 0.0040 9.9237300000E-25 0.0045 9.8529600000E-17 0.0050 1.9191200000E-11 0.0055 8.5160200000E-08 0.0060 3.3292600000E-05 0.0065 2.6725000000E-03 0.0070 6.8331200000E-02 0.0075 8.1324300000E-01 0.0080 5.4604600000E+00 0.0085 2.3493800000E+01 0.0090 7.5508400000E+01 0.0095 1.9196300000E+02 0.0100 4.1136600000E+02 0.0105 7.5113400000E+02 0.0110 1.2277100000E+03 0.0115 1.8387000000E+03 0.0120 2.5574500000E+03 0.0125 3.3554500000E+03 0.0130 4.1970700000E+03 0.0135 5.0380700000E+03 0.0140 5.8488400000E+03 0.0145 6.6216300000E+03 0.0150 7.3029400000E+03 0.0155 7.9283000000E+03 0.0160 8.4561700000E+03 0.0165 8.8891800000E+03 0.0170 9.2148700000E+03 0.0175 7.7880200000E+04 0.0180 9.5926200000E+03
330 0.0185 9.6486200000E+03 0.0190 9.6282600000E+03 0.0195 2.3375700000E+04 0.0200 8.5512500000E+03 0.0205 8.2456100000E+03 0.0210 8.0847000000E+03 0.0215 7.8243600000E+03 0.0220 7.5362800000E+03 0.0225 7.2207500000E+03 0.0230 6.8068600000E+03 0.0235 1.2841300000E+03 0.0240 1.2892700000E+03 0.0245 1.2769100000E+03 0.0250 1.2347800000E+03 0.0255 1.1733200000E+03 0.0260 1.0854600000E+03 0.0265 9.6395900000E+02 0.0270 8.2259200000E+02 0.0275 6.5527900000E+02 0.0280 4.5932900000E+02 0.0285 2.4050300000E+02 0.0290 1.4506800000E+01
331 C ***************************************************** C * Source = IPEM Rep 78_ 027K20D0M2* 30 kVp Mo/Rh * C * Tube Potential = 30 kVp * C * Filtered => 0.69 Be_ 0.025 Rh_ 2.5 PMMA_ 0.1570 Al* C * Mean Photon = 18.3 kev * C * Air Kerma = 6.789E01 uGy per mAs @ 750mm * C * HVL = 0.4480 mm Al * C * Anode Angle = 20 degrees * C * *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 2.5300500000E-38 0.0040 1.5675700000E-25 0.0045 2.5965200000E-17 0.0050 7.1528700000E-12 0.0055 4.0202900000E-08 0.0060 1.8728900000E-05 0.0065 1.7050800000E-03 0.0070 4.7984300000E-02 0.0075 6.1459500000E-01 0.0080 4.3698500000E+00 0.0085 1.9671400000E+01 0.0090 6.5576300000E+01 0.0095 1.7164400000E+02 0.0100 3.7700500000E+02 0.0105 7.0206000000E+02 0.0110 1.1670800000E+03 0.0115 1.7718000000E+03 0.0120 2.4949300000E+03 0.0125 3.3065200000E+03 0.0130 4.1724800000E+03 0.0135 5.0488600000E+03 0.0140 5.9023700000E+03 0.0145 6.7229200000E+03 0.0150 7.4563200000E+03 0.0155 8.1379200000E+03 0.0160 8.7176800000E+03 0.0165 9.2082400000E+03 0.0170 9.5988300000E+03 0.0175 8.9116300000E+04 0.0180 1.0088500000E+04
332 0.0185 1.0187700000E+04 0.0190 1.0217500000E+04 0.0195 2.6328800000E+04 0.0200 9.0806100000E+03 0.0205 8.7941400000E+03 0.0210 8.6860500000E+03 0.0215 8.4877500000E+03 0.0220 8.2493300000E+03 0.0225 7.9844900000E+03 0.0230 7.6324500000E+03 0.0235 1.4563600000E+03 0.0240 1.4896100000E+03 0.0245 1.5000900000E+03 0.0250 1.4899800000E+03 0.0255 1.4535800000E+03 0.0260 1.3917000000E+03 0.0265 1.3062000000E+03 0.0270 1.1926400000E+03 0.0275 1.0526800000E+03 0.0280 8.9148400000E+02 0.0285 7.0133300000E+02 0.0290 4.8860400000E+02 0.0295 2.4975500000E+02 0.0300 1.6948800000E+01
333 C ***************************************************** C * Source = IPEM Rep 78_ 032K20D0M2* 32 kVp Mo/Rh * C * Tube Potential = 32 kVp * C * Filtered => 0.69 Be_ 0.025 Rh_ 2.5 PMMA_ 0.1755 Al* C * Mean Photon = 18.6 kev * C * Air Kerma = 7.968E01 uGy per mAs @ 750mm * C * HVL = 0.4653 mm Al * C * Anode Angle = 20 degrees * C * *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 0.0000000000E+00 0.0040 2.4748900000E-26 0.0045 6.8029100000E-18 0.0050 2.6505200000E-12 0.0055 1.8945900000E-08 0.0060 1.0578400000E-05 0.0065 1.0981900000E-03 0.0070 3.4215300000E-02 0.0075 4.7414000000E-01 0.0080 3.5870100000E+00 0.0085 1.6965300000E+01 0.0090 5.8847000000E+01 0.0095 1.5908800000E+02 0.0100 3.5891400000E+02 0.0105 6.8351000000E+02 0.0110 1.1572800000E+03 0.0115 1.7846100000E+03 0.0120 2.5480200000E+03 0.0125 3.4143600000E+03 0.0130 4.3538700000E+03 0.0135 5.3151200000E+03 0.0140 6.2634300000E+03 0.0145 7.1817600000E+03 0.0150 8.0203400000E+03 0.0155 8.8068500000E+03 0.0160 9.4974000000E+03 0.0165 1.0083300000E+04 0.0170 1.0573600000E+04 0.0175 1.1500700000E+05 0.0180 1.1238000000E+04
334 0.0185 1.1420400000E+04 0.0190 1.1529800000E+04 0.0195 3.2955500000E+04 0.0200 1.0172200000E+04 0.0205 9.8969900000E+03 0.0210 9.8900500000E+03 0.0215 9.7733300000E+03 0.0220 9.6180400000E+03 0.0225 9.4386200000E+03 0.0230 9.1635500000E+03 0.0235 1.7832800000E+03 0.0240 1.8576800000E+03 0.0245 1.9196500000E+03 0.0250 1.9551800000E+03 0.0255 1.9711600000E+03 0.0260 1.9646700000E+03 0.0265 1.9290300000E+03 0.0270 1.8752200000E+03 0.0275 1.7957100000E+03 0.0280 1.6906700000E+03 0.0285 1.5572300000E+03 0.0290 1.4001600000E+03 0.0295 1.2197400000E+03 0.0300 1.0116600000E+03 0.0305 7.8577700000E+02 0.0310 5.4406200000E+02 0.0315 2.7884000000E+02 0.0320 2.2338600000E+01
335 C *************************************************** C * Source = IPEM Rep 78_ 025K20D0M2 25 kVp Mo/Mo * C * Tube Potential = 25 kVp * C * Filters(mm)=>0.69 Be_0.03 Mo_2.5 PMMA_0.1745 Al * C * Mean Photon = 16.6 kev * C * Air Kerma = 3.872E01 uG y per mAs @ 750mm * C * HVL = 0.3476 mm Al * C * Anode Angle = 20 degrees * C *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 1.7339100000E-35 0.0040 2.2437600000E-23 0.0045 1.0305200000E-15 0.0050 1.2302500000E-10 0.0055 3.7500100000E-07 0.0060 1.1208900000E-04 0.0065 7.0571900000E-03 0.0070 1.4999900000E-01 0.0075 1.4929400000E+00 0.0080 8.5048200000E+00 0.0085 3.4625700000E+01 0.0090 1.0137500000E+02 0.0095 2.4025400000E+02 0.0100 4.8371400000E+02 0.0105 8.4208500000E+02 0.0110 1.3206500000E+03 0.0115 1.8970300000E+03 0.0120 2.5530100000E+03 0.0125 3.2505900000E+03 0.0130 3.9649400000E+03 0.0135 4.6359900000E+03 0.0140 5.2464600000E+03 0.0145 5.8128700000E+03 0.0150 6.3077700000E+03 0.0155 6.6801400000E+03 0.0160 6.9840800000E+03 0.0165 7.1758800000E+03 0.0170 7.2834100000E+03 0.0175 3.5911000000E+04 0.0180 7.1713800000E+03
336 0.0185 6.9819200000E+03 0.0190 6.7434500000E+03 0.0195 1.2086500000E+04 0.0200 7.4918400000E+02 0.0205 6.8696400000E+02 0.0210 6.4294100000E+02 0.0215 6.0673300000E+02 0.0220 5.6911800000E+02 0.0225 5.2614900000E+02 0.0230 4.6784300000E+02 0.0235 3.8950000000E+02 0.0240 2.8884700000E+02 0.0245 1.5944400000E+02 0.0250 1.3193000000E+01
337 C **************************************************** C * Source = IPEM Rep 78_ 026K20D0M2* Mo/Mo 26 kVp * C * Tube Potential = 26 kVp * C * Filter(mm)=>0.69 Be_ 0.03 Mo_ 2.5 PMMA_ 0.1755 AL* C * Mean Photon = 16.8 kev * C * Air Kerma = 4.4335E01 uGy per mAs @ 750mm * C * HVL = 0.3594 mm Al * C * Anode Angle = 20 degrees * C **************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 1.4712000000E-35 0.0040 1.9973400000E-23 0.0045 9.5029300000E-16 0.0050 1.1669400000E-10 0.0055 3.6378100000E-07 0.0060 1.1070500000E-04 0.0065 7.0705100000E-03 0.0070 1.5207200000E-01 0.0075 1.5278200000E+00 0.0080 8.7697100000E+00 0.0085 3.5922900000E+01 0.0090 1.0569200000E+02 0.0095 2.5162200000E+02 0.0100 5.0842800000E+02 0.0105 8.8822400000E+02 0.0110 1.3968400000E+03 0.0115 2.0116600000E+03 0.0120 2.7138000000E+03 0.0125 3.4627800000E+03 0.0130 4.2359100000E+03 0.0135 4.9634000000E+03 0.0140 5.6332600000E+03 0.0145 6.2617500000E+03 0.0150 6.8152100000E+03 0.0155 7.2467900000E+03 0.0160 7.6053100000E+03 0.0165 7.8537600000E+03 0.0170 8.0178500000E+03 0.0175 4.6631400000E+04 0.0180 8.0076500000E+03
338 0.0185 7.8739900000E+03 0.0190 7.6842700000E+03 0.0195 1.5048300000E+04 0.0200 8.6749300000E+02 0.0205 8.0928200000E+02 0.0210 7.7587500000E+02 0.0215 7.5295900000E+02 0.0220 7.3255500000E+02 0.0225 7.1146600000E+02 0.0230 6.7847100000E+02 0.0235 6.2997000000E+02 0.0240 5.6350100000E+02 0.0245 4.6972700000E+02 0.0250 3.4512400000E+02 0.0255 1.8879400000E+02 0.0260 1.7446100000E+01
339 C *************************************************** C * Source = IPEM Rep 78_ 028K20D0M2 28 kVp Mo/Mo * C * Tube Potential = 28 kVp * C * Filters(mm)=>0.69 Be_0.03 Mo_2.5 PMMA_0.1325 Al * C * Mean Photon = 17.1 kev * C * Air Kerma = 6.223E01 uG y per mAs @ 750mm * C * HVL = 0.3679 mm Al * C * Anode Angle = 20 degrees * C *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 5.7364300000E-33 0.0040 1.2422600000E-21 0.0045 1.8823400000E-14 0.0050 1.0831400000E-09 0.0055 2.0436600000E-06 0.0060 4.3205400000E-04 0.0065 2.1397600000E-02 0.0070 3.8041900000E-01 0.0075 3.3100200000E+00 0.0080 1.7008600000E+01 0.0085 6.3950300000E+01 0.0090 1.7570400000E+02 0.0095 3.9604900000E+02 0.0100 7.6534700000E+02 0.0105 1.2892500000E+03 0.0110 1.9685900000E+03 0.0115 2.7673700000E+03 0.0120 3.6562000000E+03 0.0125 4.5889600000E+03 0.0130 5.5343700000E+03 0.0135 6.4112900000E+03 0.0140 7.2140400000E+03 0.0145 7.9638500000E+03 0.0150 8.6254000000E+03 0.0155 9.1432100000E+03 0.0160 9.5744300000E+03 0.0165 9.8905300000E+03 0.0170 1.0114500000E+04 0.0175 7.5823100000E+04 0.0180 1.0191700000E+04
340 0.0185 1.0093000000E+04 0.0190 9.9414900000E+03 0.0195 2.2499100000E+04 0.0200 1.1297400000E+03 0.0205 1.0703100000E+03 0.0210 1.0511900000E+03 0.0215 1.0507300000E+03 0.0220 1.0607100000E+03 0.0225 1.0790400000E+03 0.0230 1.0894500000E+03 0.0235 1.0952500000E+03 0.0240 1.0879600000E+03 0.0245 1.0572300000E+03 0.0250 1.0041100000E+03 0.0255 9.2437500000E+02 0.0260 8.1211700000E+02 0.0265 6.6081300000E+02 0.0270 4.7846000000E+02 0.0275 2.5665000000E+02 0.0280 2.1261600000E+01
341 C **************************************************** C * Source = IPEM Rep 78_ 026K20D0M2* Mo/Rh 26 kVp * C * Tube Potential = 26 kVp * C * Filter(mm)=>0.69 Be_ 0.025 Rh_ 2.5 PMMA_ 0.1852 AL* C * Mean Photon = 17.8 kev * C * Air Kerma = 3.979E01 uGy per mAs @ 750mm * C * HVL = 0.4179 mm Al * C * Anode Angle = 20 degrees * C **************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 0.0000000000E+00 0.0040 1.0918400000E-26 0.0045 3.7830200000E-18 0.0050 1.6800200000E-12 0.0055 1.2835100000E-08 0.0060 7.4300300000E-06 0.0065 7.8314100000E-04 0.0070 2.4545500000E-02 0.0075 3.4037100000E-01 0.0080 2.5697300000E+00 0.0085 1.2107400000E+01 0.0090 4.1828200000E+01 0.0095 1.1262400000E+02 0.0100 2.5292900000E+02 0.0105 4.7976000000E+02 0.0110 8.0900000000E+02 0.0115 1.2418600000E+03 0.0120 1.7656100000E+03 0.0125 2.3549700000E+03 0.0130 2.9889600000E+03 0.0135 3.6289700000E+03 0.0140 4.2496200000E+03 0.0145 4.8403600000E+03 0.0150 5.3614900000E+03 0.0155 5.8335700000E+03 0.0160 6.2238100000E+03 0.0165 6.5303000000E+03 0.0170 6.7505300000E+03 0.0175 3.9858200000E+04 0.0180 6.9257100000E+03
342 0.0185 6.8887300000E+03 0.0190 6.7785700000E+03 0.0195 1.3381200000E+04 0.0200 5.9795400000E+03 0.0205 5.6548300000E+03 0.0210 5.3653200000E+03 0.0215 4.9992800000E+03 0.0220 4.5864700000E+03 0.0225 4.1377000000E+03 0.0230 3.6315600000E+03 0.0235 6.2030400000E+02 0.0240 5.5022500000E+02 0.0245 4.5199600000E+02 0.0250 3.2719500000E+02 0.0255 1.7525400000E+02 0.0260 1.5902800000E+01
343 C ***************************************************** C * Source = IPEM Rep 78_ 027K20D0M2* 27 kVp Mo/Rh * C * Tube Potential = 27 kVp * C * Filtered => 0.69 Be_ 0.025 Rh_ 2.5 PMMA_ 0.1338 Al* C * Mean Photon = 17.9 kev * C * Air Kerma = 5.027E01 uGy per mAs @ 750mm * C * HVL = 0.4158 mm Al * C * Anode Angle = 20 degrees * C * *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 6.9614400000E-37 0.0040 1.5810100000E-24 0.0045 1.3780600000E-16 0.0050 2.4387800000E-11 0.0055 1.0079000000E-07 0.0060 3.7249200000E-05 0.0065 2.8642100000E-03 0.0070 7.0787000000E-02 0.0075 8.1963000000E-01 0.0080 5.3845900000E+00 0.0085 2.2760700000E+01 0.0090 7.2102400000E+01 0.0095 1.8110200000E+02 0.0100 3.8421400000E+02 0.0105 6.9557700000E+02 0.0110 1.1287700000E+03 0.0115 1.6796500000E+03 0.0120 2.3235600000E+03 0.0125 3.0319800000E+03 0.0130 3.7741700000E+03 0.0135 4.5100200000E+03 0.0140 5.2140600000E+03 0.0145 5.8768700000E+03 0.0150 6.4511500000E+03 0.0155 6.9679900000E+03 0.0160 7.3909300000E+03 0.0165 7.7229600000E+03 0.0170 7.9594300000E+03 0.0175 5.3689100000E+04 0.0180 8.1598900000E+03
344 0.0185 8.1253200000E+03 0.0190 8.0170000000E+03 0.0195 1.7053200000E+04 0.0200 7.0900300000E+03 0.0205 6.7480800000E+03 0.0210 6.4848700000E+03 0.0215 6.1419300000E+03 0.0220 5.7554800000E+03 0.0225 5.3379400000E+03 0.0230 4.8461900000E+03 0.0235 8.7058600000E+02 0.0240 8.2306400000E+02 0.0245 7.5116700000E+02 0.0250 6.5387500000E+02 0.0255 5.3057200000E+02 0.0260 3.8178300000E+02 0.0265 2.0095400000E+02 0.0270 1.5098200000E+01
345 C *************************************************** C * Source = IPEM Rep 78_ 028K20D0M2 28 kVp Mo/Rh * C * Tube Potential = 28 kVp * C * Filters(mm)=>0.69 Be_0.025 Rh_2.5 PMMA_0.2210 Al * C * Mean Photon = 18.2 kev * C * Air Kerma = 4.860E01 uG y per mAs @ 750mm * C * HVL = 0.4481 mm Al * C * Anode Angle = 20 degrees * C *************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 0.0000000000E+00 0.0040 3.2440800000E-28 0.0045 2.9946000000E-19 0.0050 2.5856500000E-13 0.0055 3.1082800000E-09 0.0060 2.5202700000E-06 0.0065 3.3792800000E-04 0.0070 1.2723500000E-02 0.0075 2.0317000000E-01 0.0080 1.7116800000E+00 0.0085 8.7947500000E+00 0.0090 3.2575700000E+01 0.0095 9.2740500000E+01 0.0100 2.1824700000E+02 0.0105 4.2995800000E+02 0.0110 7.4869600000E+02 0.0115 1.1805400000E+03 0.0120 1.7187800000E+03 0.0125 2.3377700000E+03 0.0130 3.0200400000E+03 0.0135 3.7266400000E+03 0.0140 4.4249900000E+03 0.0145 5.1077800000E+03 0.0150 5.7301900000E+03 0.0155 6.3125400000E+03 0.0160 6.8131900000E+03 0.0165 7.2365900000E+03 0.0170 7.5753300000E+03 0.0175 5.8194700000E+04 0.0180 7.9825100000E+03
346 0.0185 8.0577300000E+03 0.0190 8.0570500000E+03 0.0195 1.8494400000E+04 0.0200 7.2384700000E+03 0.0205 6.9858000000E+03 0.0210 6.8218600000E+03 0.0215 6.5735100000E+03 0.0220 6.2815800000E+03 0.0225 5.9559200000E+03 0.0230 5.5521200000E+03 0.0235 1.0296400000E+03 0.0240 1.0170600000E+03 0.0245 9.7625600000E+02 0.0250 9.1546100000E+02 0.0255 8.2694700000E+02 0.0260 7.1478200000E+02 0.0265 5.7018900000E+02 0.0270 4.0839100000E+02 0.0275 2.1595800000E+02 0.0280 1.7741300000E+01
347 C **************************************************** C * Source = IPEM Rep 78_ 030K20D0M2* 30 kVp Mo/Rh * C * Tube Potential = 30 kVp * C * Filter(mm)=> 0.69 Be_ 0.025 Rh_ 2.5 Al_ 0.2015 Al* C * Mean Photon = 18.4 kev * C * Air Kerma = 6.417E01 uGy per mAs @ 750mm * C * HVL = 0.4562 mm Al * C * Anode Angle = 20 degrees * C * ************************************************** C Mev Probability 0.0005 0.0000000000E+00 0.0010 0.0000000000E+00 0.0015 0.0000000000E+00 0.0020 0.0000000000E+00 0.0025 0.0000000000E+00 0.0030 0.0000000000E+00 0.0035 0.0000000000E+00 0.0040 6.6658200000E-27 0.0045 2.6544400000E-18 0.0050 1.3159900000E-12 0.0055 1.0974400000E-08 0.0060 6.8296200000E-06 0.0065 7.6279900000E-04 0.0070 2.5050000000E-02 0.0075 3.6090400000E-01 0.0080 2.8105400000E+00 0.0085 1.3585300000E+01 0.0090 4.7946100000E+01 0.0095 1.3135200000E+02 0.0100 2.9960000000E+02 0.0105 5.7530000000E+02 0.0110 9.8101100000E+02 0.0115 1.5210200000E+03 0.0120 2.1816200000E+03 0.0125 2.9346500000E+03 0.0130 3.7522900000E+03 0.0135 4.5925000000E+03 0.0140 5.4190000000E+03 0.0145 6.2241200000E+03 0.0150 6.9535500000E+03 0.0155 7.6379300000E+03 0.0160 8.2287400000E+03 0.0165 8.7345500000E+03 0.0170 9.1459700000E+03 0.0175 8.5241000000E+04 0.0180 9.6837500000E+03
348 0.0185 9.8098600000E+03 0.0190 9.8660800000E+03 0.0195 2.5488000000E+04 0.0200 8.8106900000E+03 0.0205 8.5499700000E+03 0.0210 8.4612000000E+03 0.0215 8.2818200000E+03 0.0220 8.0619200000E+03 0.0225 7.8140300000E+03 0.0230 7.4793400000E+03 0.0235 1.4287700000E+03 0.0240 1.4630700000E+03 0.0245 1.4747800000E+03 0.0250 1.4661300000E+03 0.0255 1.4315600000E+03 0.0260 1.3717000000E+03 0.0265 1.2883300000E+03 0.0270 1.1771500000E+03 0.0275 1.0396500000E+03 0.0280 8.8098900000E+02 0.0285 6.9344200000E+02 0.0290 4.8336000000E+02 0.0295 2.4718300000E+02 0.0300 1.6781600000E+01
349 Material Cards for MCNP-5 Input Material Cards used in simulations placed af ter "C Material Cards" section. C Elemental Lead Rho-11.4 gm/cm^3_12-Pb m1 82000 1 C Dry Air Rh0-0.001205 gm/cm^3_7-N, 8-O, 18-Ar m2 7000 -0.755 8000 -0.232 18000 -0.013 C Elemental Calcium Rho-1.5500 gm/cm^3_20-Ca m3 20000 1 C Methyl-methacrylate Rho-1.17 gm/cm^3_1-H, 6-C, 8-O C ICRU44, Acrylic m4 1000 -0.08 6000 -0.6 8000 -0.32 C Tissue Glandularity in percent C 0pct Rho-0.9500 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S C ICRU44-100pct Adipose m5 1000 -0.1140 6000 -0.5980 7000 -0.0070 8000 -0.2780 17000 -0.0010 & 11000 -0.0010 16000 -0.0010 C 100pct Rho-1.020 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S C ICRU44-Glandular m6 1000 -0.1060 6000 -0.3320 7000 -0.0300 8000 -0.5270 17000 -0.0010 & 11000 -0.0010 15000 -0.0010 16000 -0.0020 C 10pct Rho-0.9566 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1132 6000 -0.5714 7000 -0.0093 8000 -0.3029 11000 -0.0010 15000 -0.0001 16000 -0.0011 17000 -0.0010 C 16.2pct Rho-0.9607 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1127 6000 -0.55491 7000 -0.01073 8000 -0.31834 11000 -0.0010 15000 -0.00016 16000 -0.00116 17000 -0.0010 C 20pct Rho-0.9632 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1124 6000 -0.5448 7000 -0.0116 8000 -0.3278 11000 -0.0010 15000 -0.0002 16000 -0.0012 17000 -0.0010 C 25.4pct Rho-0.9669 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.11197 6000 -0.53044 7000 -0.01284 8000 -0.34125 11000 -0.0010 15000 -0.00025 16000 -0.00125 17000 -0.0010 C 30pct Rho-0.9700 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1116 6000 -0.5182 7000 -0.0139 8000 -0.3527 11000 -0.0010
350 15000 -0.0003 16000 -0.0013 17000 -0.0010 C 40pct Rho-0.9768 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1108 6000 -0.4916 7000 -0.0162 8000 -0.3776 11000 -0.0010 15000 -0.0004 16000 -0.0014 17000 -0.0010 C 42.6pct Rho-0.9786 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.11059 6000 -0.48468 7000 -0.01680 8000 -0.38407 11000 -0.0010 15000 -0.00043 16000 -0.00143 17000 -0.0010 C 50pct Rho-0.9838 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1100 6000 -0.4650 7000 -0.0185 8000 -0.4025 11000 -0.0010 15000 -0.0005 16000 -0.0015 17000 -0.0010 C 60pct Rho-0.9908 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1092 6000 -0.4384 7000 -0.0208 8000 -0.4274 11000 -0.0010 15000 -0.0006 16000 -0.0016 17000 -0.0010 C 67.8pct Rho-0.9964 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.10858 6000 -0.41765 7000 -0.02259 8000 -0.44682 11000 -0.0010 15000 -0.00068 16000 -0.00168 17000 -0.0010 C 70pct Rho-0.9979 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1084 6000 -0.4118 7000 -0.0231 8000 -0.4523 11000 -0.0010 15000 -0.0007 16000 -0.0017 17000 -0.0010 C 80pct Rho-1.0052 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1076 6000 -0.3852 7000 -0.0254 8000 -0.4772 11000 -0.0010 15000 -0.0008 16000 -0.0018 17000 -0.0010 C 90pct Rho-1.0125 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S m7 1000 -0.1068 6000 -0.3586 7000 -0.0277 8000 -0.5021 11000 -0.0010 15000 -0.0009 16000 -0.0019 17000 -0.0010 C BR-12 50pct Rho-0.9700 gm/cm^3_1-H, 6-C, 7-N, 8-O, 17-Cl, 20-Ca C ICRU44,Br-12 m7 1000 -0.0870 6000 -0.6990 7000 -0.02400 8000 -0.1790 17000 -0.0010 20000 -0.0100 C Parafin Rho-0.9300 gm/cm^3_1-H, 6-C C ICRU44,Br-12 m7 1000 -0.1500 6000 -0.8500 C Elemental Aluminum Rho-2.700 gm/cm^3_13-Al m8 13000
351 C Adult Muscle ICRU46 Rho-1.050 gm/cm^3 C 1-H, 6-C, 7-N, 8-O, 17-Cl, 11-Na, 15-P, 16-S, 19-K m9 1000 -0.1020 6000 -0.1430 7000 -0.0340 8000 -0.7100 11000 -0.0010 15000 -0.0020 16000 -0.0030 17000 -0.0010 19000 -0.0040 C Nylon-6 ICRU44 Rho-1.130 gm/cm^3 C 1-H, 6-C, 7-N, 8-O M9 1000 -0.0980 6000 -0.6370 7000 -0.1240 8000 -0.1410 C Al2O3 simulate specks in ACR phantom Rho-3.65 gm/cm^3 m9 13000 -0.5293 8000 -0.4707 m8 13000 1 $Aluminum,2.7gm/cm^3 1
352 Source Collimator Al or air Ionization Chamber (6cc) Compression Paddle YZ X Y XY Figure A-1. Free-in-air geometry for HVL measurements.
353 Collimator Compression Paddle Ionization Chamber (6cc) Phantom Y Z XY Adipose Glandular Source Figure A-2. Phantom and volumetric factors geomet ry for Wu and BRTES-MOD phantoms.
354 Source Collimator Compression Paddle Ionization Chamber (6cc) Phantom Muscle Rib Glandular Adipose XY XZ YZ Figure A-3. Anatomical factor geometry for Wu phantom.
355 LIST OF REFERENCES 1D. M. Parkin, "Global cancer statistics in the year 2000," Lancet Oncol 2, 533-43 (2001). 2ACS, Cancer facts & figures 2003. Corp Au thor(s): American Cancer Society, (American Cancer Society, Atlanta, GA, 2003). 3ACS, Cancer facts & figures 2005. Corp Au thor(s): American Cancer Society, (American Cancer Society, Atlanta, GA, 2005). 4L. L. Humphrey, M. Helfand, B. K. Chan, a nd S. H. Woolf, "Breast cancer screening: a summary of the evidence for the U.S. Preven tive Services Task Force," Ann Intern Med 137, 347-60 (2002). 5ACS, Breast Cancer facts & figures 2003-2004. Corp Author(s): American Cancer Society, (American Cancer Society, Atlanta, GA, 2003). 6National Council on Radiation Pr otection and Measurements., Mammography--a user's guide: recommendations of the National Council on Radiation Protection and Measurements. Corp Author(s): National Council on Radiation Protection and Measurements.; National Council on Radiati on Protection and Measur ements.; Scientific Committee 72 on Radiation Protection in Mammography, (The Council, Bethesda, MD., 1987). 7A. C. Upton, National Research Council (U.S .). Committee on the Biological Effects of Ionizing Radiations., National Research C ouncil (U.S.). Board on Radiation Effects Research., and National Research Council (U.S.). Commission on Life Sciences., Health effects of exposure to low levels of ionizing radiation, (National Academy Press, Washington, D.C., 1990). 8M. Frankenberg-Schwager, I. Garg, D. Fran-K enberg, B. Greve, E. Severin, D. Uthe, and W. Gohde, "Mutagenicity of low-filte red 30 kVp X-rays, ma mmography X-rays and conventional X-rays in cultured mammalian ce lls," Int J Radiat Biol 78, 781-9 (2002). 9National Council on Radiation Pr otection and Measurements., Mammography, recommendations of the National Council on Radiation Protection and Measurements, (NCRP, Washington, D.C., 1980).
356 10Los Alamos Scientific Laboratory. Group X-6 ., and Los Alamos Scientific Laboratory., MCNP: a general Monte Carlo c ode for neutron and photon transport, (Dept. of Energy Los Alamos Scientific Laboratory; for sa le by the National Technical Information Service, Los Alamos, N.M. Springfield, Va., 1979). 11ACR, Breast imaging reporting and data system (BI-RADS). Corp Author(s): American College of Radiology, 3rd (American College of Radiology, Reston, VA, 1998). 12ACR, Mammography quality control manual: radiol ogist's section, clinical image quality, radiologic technologist's section, me dical physicist's section Corp Author(s): American College of Radiology. Committ ee on Quality Assurance in Mammography, Rev. (American College of Radiology, Rest on, VA (1891 Preston White Drive, Reston, 20191-4397), 1999). 13D. B. Kopans, Breast imaging, 2nd (Lippincott Williams & Wilkins, Philadelphia, 1998). 14K. L. Mossman, and W. A. Mills, The Biological basis of r adiation protection practice, (Williams & Wilkins, Baltimore, 1992). 15B. H. Lerner, ""To see today with the ey es of tomorrow": A history of screening mammography," Canadian bulletin of medical history = Bulletin canadien d'histoire de la medecine 20, 299-321 (2003). 16J. Gershon-Cohen, H. Ingleby, and L. Moor e, "Can mass x-ray surveys be used in detection of early cance r of the breast?" J Am Med Assoc 161, 1069-71 (1956). 17L. W. Bassett, "The regulation of ma mmography," Semin Ultrasound CT MR 17, 41523 (1996). 18R. H. Gold, "The evolution of mammography, " Radiol Clin North Am 30, 1-19 (1992). 19M. Segal, "Mammography Facilities Must Meet Quality Standards," U.S. Food and Drug Administration Consumer Magazine FDA 94-8284, (1994). 20"Making Sense of the Mammography C ontroversy: What Women should Know," in Committee on Health, education, labor, and pensions, United States Senate and Committee on Appropriations United States Senate, edited by US Printing Office, 2002), pp. 109. 21P. C. Gotzsche, and O. Olsen, "Is sc reening for breast can cer with mammography justifiable?" The Lancet 129-133 (2000). 22A. B. Miller, T. To, C. J. Baines, and C. Wall, "Canadian National Breast Screening Study-2: 13-Year Results of a Randomized Tr ial in Women Aged 50-59 Years," Journal of the National Cancer Institute 92, no 18, 1490 (10 pages) (2000).
357 23D. A. Freedman, D. B. Petitti, and J. M. R obins, "On the efficacy of screening for breast cancer," Int J Epidemiol 33, 43-55 (2004). 24L. Nystrom, "Long-term effects of ma mmography screening," The Lancet 909-919 (2002). 25L. Tabar, "Mammography service screening and mortality in br east cancer patients," The Lancet 1405-1410 (2003). 26S. J. Otto, "Initiation of populationbased mammography screening in Dutch municipalities and effect on breast-cancer mortality," The Lancet 1411-1417 (2003). 27R. A. Smith, S. W. Duffy, R. Gabe, L. Ta bar, A. M. Yen, and T. H. Chen, "The randomized trials of breast cancer screening: what have we learned?" Radiol Clin North Am 42, 793-806, v (2004). 28G. Hammerstein, D. Miller, D. White, M. Masterson, H. Woodard, and J. Laughlin, "Absorbed radiation dose in mammo graphy," Radiology 130, 485-491 (1979). 29R. A. Geise, and A. Palchevsky, "Composition of Mammographic Phantom Materials," Radiology 198, 347 (4 pages) (1996). 30D. R. Dance, C. L. Skinner, K. C. Young, J. R. Beckett, and C. J. Kotre, "Additional factors for the estimation of mean gla ndular breast dose us ing the UK mammography dosimetry protocol," Physics in medi cine & biology 45 Part 11, 3225-3240 (2000). 31J. T. Bushberg, The essential physics of medical imaging, 2nd (Lippincott Williams & Wilkins, Philadelphia, 2002). 32J. A. Harvey, and V. E. Bovbjerg, "Quan titative assessment of mammographic breast density: relationship with breast cancer risk," Ra diology 230, 29-41 (2004). 33G. Maskarinec, and L. Meng, "A case-c ontrol study of mammographic densities in Hawaii," Breast Cancer Re s Treat 63, 153-61 (2000). 34P. B. Lam, P. M. Vacek, B. M. Geller, and H. B. Muss, "The association of increased weight, body mass index, and tissue density with the risk of breast carcinoma in Vermont," Cancer 89, 369-75 (2000). 35J. Brisson, N. L. Sadowsky, J. A. Twaddle, A. S. Morrison, P. Cole, and F. Merletti, "The relation of mammographic features of the breast to brea st cancer risk factors," Am J Epidemiol 115, 438-43 (1982). 36J. N. Wolfe, "Breast patterns as an index of risk for de veloping breast cancer," Am J Roentgenol 126, 1130-7 (1976).
358 37N. F. Boyd, B. O'Sullivan, J. E. Campbell, E. Fishell, I. Simor, G. Cooke, and T. Germanson, "Mammographic signs as risk f actors for breast cancer," Br J Cancer 45, 185-93 (1982). 38J. N. Wolfe, A. F. Saftlas, and M. Sala ne, "Mammographic pare nchymal patterns and quantitative evaluation of mammographic dens ities: a case-control study," AJR Am J Roentgenol 148, 1087-92 (1987). 39J. Brisson, R. Verreault, A. S. Morris on, S. Tennina, and F. Meyer, "Diet, mammographic features of breas t tissue, and breast cancer risk," Am J Epidemiol 130, 14-24 (1989). 40R. L. Egan, and R. C. Mosteller, "Breas t cancer mammography patterns," Cancer 40, 2087-90 (1977). 41A. F. Saftlas, and M. Szklo, "Mammogr aphic parenchymal patterns and breast cancer risk," Epidemiol Rev 9, 146-74 (1987). 42A. M. Oza, and N. F. Boyd, "Mammographi c parenchymal patterns: a marker of breast cancer risk," Epidemiol Rev 15, 196-208 (1993). 43A. F. Saftlas, R. N. Hoover, L. A. Brinton, M. Szklo, D. R. Olson, M. Salane, and J. N. Wolfe, "Mammographic densities and risk of breast cancer," Can cer 67, 2833-8 (1991). 44J. J. Heine, and P. Malhotra, "Mammographi c tissue, breast cancer risk, serial image analysis, and digital mammography. Part 2. Serial breast tissue change and related temporal influences," Acad Radiol 9, 317-35 (2002). 45J. J. Heine, and P. Malhotra, "Mammographi c tissue, breast cancer risk, serial image analysis, and digital mammography. Part 1. Tiss ue and related risk factors," Acad Radiol 9, 298-316 (2002). 46Y. H. Chang, X. H. Wang, L. A. Hardesty, T. S. Chang, W. R. Poller, W. F. Good, and D. Gur, "Computerized assessment of ti ssue composition on digitized mammograms," Academic radiology 9, 899-905 (2002). 47O. Pawluczyk, B. J. Augustine, M. J. Yaffe, D. Rico, J. Yang, G. E. Mawdsley, and N. F. Boyd, "A volumetric method for estimation of breast density on digitized screen-film mammograms," Medical physics 30, 352 (13 pages) (2003). 48R. P. Highnam, J. M. Brady, and B. J. Shepstone, "Estimation of compressed breast thickness during mammography," Br J Radiol 71, 646-53 (1998). 49P. C. Johns, and M. J. Yaffe, "X-ray char acterisation of normal and neoplastic breast tissues," Physics in medici ne and biology 32, 675-95 (1987).
359 50D. R. Dance, C. L. Skinner, and G. Alm Carlsson, "Breast dosimetry," Applied radiation and isotopes: including data, instrume ntation and methods for use in agriculture, industry, and medicine 50, 185 (1999). 51D. R. White, R. J. Martin, and R. Darlison, "Epoxy resin based tissue substitutes," The British journal of radiology 50, 814-21 (1977). 52M. Rosenstein, L. W. Andersen, and G. G. Warner, Handbook of Glandular Tissue Doses in Mammography, Reprint 1987 (Food and Drug Ad ministartion, Rockville, Maryland, 1985). 53B. o. R. Health, Symposium on Biological Effects, Imaging Techniques, and Dosimetry of Ionizing Radiations, Rockvill e, Maryland, June 6-8, 1979, (U.S., Rockville, Md., 1980). 54L. Stanton, T. Villafana, J. L. Day, and D. A. Lightfoot, "Dosage evaluation in mammography," Radiology 150, 577-584 (1984). 55D. R. Dance, "Monte Carlo calculation of co nversion factors for the estimation of mean glandular breast dose," Physics in me dicine and biology 35, 1211-9 (1990). 56X. Wu, G. Barnes, and D. Tucker, "Spectra l dependence of glandular tissue dose in screen-film mammography," Radiology 179, 143-148 (1991). 57X. Wu, E. L. Gingold, G. T. Barnes, and D. M. Tucker, "Normalized Average Glandular Dose in Molybdenum Target-R hodium Filter and Rhodium Target-Rhodium Filter Mammography," Radiology 193, 83 (1994). 58R. D. Rosenberg, C. A. Kelsey, M. R. W illiamson, J. D. Houston, and W. C. Hunt, "Computer-based collection of mammographi c exposure data for quality assurance and dosimetry," Medical physics 28, 1546 (6 pages) (2001). 59R. L. Kruger, and B. A. Schueler, "A surv ey of clinical factors and patient dose in mammography," Medical physics 28, 1449 (6 pages) (2001). 60N. Jamal, K. H. Ng, D. McLean, L. M. Looi, and F. Moosa, "Mammographic breast glandularity in Malaysian women: data derived from radiography," AJR Am J Roentgenol 182, 713-7 (2004). 61A. Burch, and D. A. Goodman, "A pilot su rvey of radiation doses received in the United Kingdom Breast Screening Program me," Br J Radiol 71, 517-27 (1998). 62R. Klein, H. Aichinger, J. Di erker, J. T. Jansen, S. Joite-Barfuss, M. Sabel, R. SchulzWendtland, and J. Zoetelief, "Determinati on of average glandular dose with modern mammography units for two large groups of pa tients," Phys Med Biol 42, 651-71 (1997). 63I. Fife, "The physical dimensions of the compressed breast," Br J Radiol 64, 73-4 (1991).
360 64R. Kienboch, "On the quantimetric met hod," Arch Roentgen Ray 11, 17 (1906). 65A. P. Council, "The history of plastic," edited 2005). 66A. Holmes-Siedle, "The space-charge dosimet er: General principl es of a new method of radiation detection," Nuclear Inst ruments and Methods 121, 169-179 (1974). 67M. W. Bower, and D. E. Hintenlang, "The Characterization of a Commercial MOSFET Dosimeter System for Use in Diagnostic X Ray," Health Phys 75, 197 (8 pages) (1998). 68D. J. Gladstone, and L. M. Chin, "Real-Tim e, In Vivo Measurement of Radiation Dose during Radioimmunotherapy in Mice Using a Miniature MOSFET Dosimeter Probe," Radiation research 141, 330 (1995). 69R. Ramani, S. Russell, and P. O'Brien, "Clinical Dosimetry Using MOSFETS," International journal of radi ation oncology, biology, physic s 37, 959 (6 pages) (1997). 70D. J. Peet, and M. D. Pryor, "Evaluati on of a MOSFET radiation sensor for the measurement of entrance surface dose in dia gnostic radiology," The British journal of radiology 72, 562 (1999). 71B. D. Pomije, C. H. Huh, M. A. Tressl er, D. E. Hintenlang, and W. E. Bolch, "OPERATIONAL TOPICS Comparison of Angul ar Free-in-Air and Tissue-Equivalent Phantom Response Measurements in p-MOSFET Dosimeters," Health Physics 80, 497 (9 pages) (2001). 72J. N. Roshau, and D. E. Hintenlang, "Chara cterization of the angular response of an "isotropic" MOSFET dosimeter," Health Phys 84, 376-9 (2003). 73P. Scalchi, and P. Francescon, "Calibrati on of a MOSFET Detection System for 6-MV in vivo Dosimetry," Internati onal journal of radiation oncol ogy, biology, physics 40, 987 (8 pages) (1998). 74S. L. Dong, T. C. Chu, J. S. Lee, G. Y. Lan, T. H. Wu, Y. H. Yeh, and J. J. Hwang, "Estimation of mean-glandular dose from monito ring breast entrance skin air kerma using a high sensitivity metal oxide semiconductor fi eld effect transistor (MOSFET) dosimeter system in mammography," Applied radi ation and isotopes: including data, instrumentation and methods for use in agri culture, industry, a nd medicine 57, 791-799 (2002). 75L. A. Benevides, and D. E. Hintenlang, "Characterization of MOSFET dosimeters for applications in mammography dosim etry," Med Phys 30, 1422 (2003). 76L. A. Benevides, and D. E. Hintenlang, "C haracterization of Very High Bias Supply for MOSFET Dosimeters in Mammography," in Radiological Society of North America Annual Meeting, edited by RSNA (RSNA, Chicago, 2004), pp. 1.
361 77B. L. Justus, P. Falkenstein, A. L. Huston, M. C. Plazas, H. Ning, and R. W. Miller, "Gated fiber-optic-coupled detector for in vi vo real-time radiation dosimetry," Appl Opt 43, 1663-8 (2004). 78S. Pllana, "History of Mont e Carlo Method," edited 2005). 79J. M. Boone, M. H. Buonocore, and V. N. Cooper, 3rd, "Monte Carlo validation in diagnostic radiological imagi ng," Med Phys 27, 1294-304 (2000). 80D. E. Peplow, and K. Verghese, "Ima ging Physics Digital mammography image simulation using Monte Carlo," Medical physics 27, 568 (12 pages) (2000). 81L. E. Wilkinson, J. C. Heggie, and P. N. J ohnston, "An investigati on into the impact of anatomical variation upon m ean glandular dose produced wi thin a standard breast," Australas Phys Eng Sci Med 22, 53-63 (1999). 82J. R. Beckett, and C. J. Kotre, "Dosimet ric implications of age related glandular changes in screening mammography," Phys Med Biol 45, 801-13 (2000). 83G. J. Tortora, and N. P. Anagnostakos, Principles of anatomy and physiology, 3rd (Harper & Row, New York, N.Y., 1981). 84M. J. Berger, Hubbell, J.H., Seltzer, S. M., Coursey, J.S., and Zucker, D.S., "XCOM: Photon Cross Section Database (version 1.2)," in edited by (National Institute of Standards and Technology, 1999). 85Thomson Nielsen Electronics Ltd., Technical Note #4: Intr oduction to The MOSFET Dosimeter, Ottawa, Canada, 1996). 86International Commission on Radiological Pr otection., "2005 Recommendations of the International Commission on Radiological Protection," in edited by (ICRP, 2004). 87H. M. Warren-Forward, and L. Duggan, "Towards in vivo TLD dosimetry in mammography," Br J Radiol 77, 426-32 (2004). 88Thomson Nielsen Electronics Ltd., Technical Note #7: MOSFET Dosimeter Specifications, Ottawa, Canada, 2004). 89A. R. No.87, Diode in vivo dosimetry for patient s receiving externa l beam radiation therapy, 87. (American Association of Physic ists in Medicine, Madison, 2005). 90A. B. de Gonzalez, and S. Darby, "Risk of cancer from diagnostic X-rays: estimates for the UK and 14 other countries," The lancet 345-351 (2004). 91P. Herzog, and C. T. Rieger, "Risk of cancer from diagnostic X-rays," The lancet 340 (2004).
362 92D. J. Brenner, and E. J. Hall, "Risk of cancer from diagnostic X-rays," Lancet 363, 2192; author reply 2192-3 (2004). 93E. Picano, "Risk of cancer from diagnostic X-rays," Lancet 363, 1909-10; author reply 1910 (2004). 94C. Cousins, and C. Sharp, "Medical inte rventional procedures-reducing the radiation risks," Clinical radiology 59, 468-73 (2004). 95D. J. Brenner, and C. D. Elliston, "Estimated radiation risks potentially associated with full-body CT screening," Radiology 232, 735-8 (2004). 96D. J. Brenner, "Estimating cancer risks from pediatric CT: going from the qualitative to the quantitative," Pediatric radiol ogy 32, 228-3; discus sion 242-4 (2002). 97C. E. Land, "Estimating cancer risks from low doses of ionizing radiation," Science 209, 1197-203 (1980). 98J. Law, and K. Faulkner, "Cancers detect ed and induced, and associated risk and benefit, in a breast screening programme," The British journal of radiology 74, 1121-7 (2001). 99K. C. Young, "Radiation doses in the UK tr ial of breast screening in women aged 40-48 years," The British journal of radiology 75, 362-70 (2002). 100United States. Congress. House. Committ ee on Commerce. Subcommittee on Health and the Environment., "Reauthorization of the Mammography Quality Standards Act of 1998 and 2004," in edited by (U.S. G.P.O.: For sale by the U.S. G.P.O., Supt. of Docs., Congressional Sales Office, 2004), pp. 27 p. 101M. C. Aznar, B. Hemdal, J. Medin, C. J. Marckmann, C. E. Andersen, L. BotterJensen, I. Andersson, and S. Mattsson, "In vivo absorbed dose measurements in mammography using a new real-time lumines cence technique," Br J Radiol 78, 328-34 (2005). 102M. C. Aznar, C. E. Andersen, L. Botter-Je nsen, S. A. Back, S. Mattsson, F. KjaerKristoffersen, and J. Medin, "Real-time optical-fibre luminescence dosimetry for radiotherapy: physical characte ristics and applications in photon beams," Phys Med Biol 49, 1655-69 (2004). 103A. L. Huston, B. L. Justus, P. L. Falkenstein, R. W. Miller, H. Ning, and R. Altemus, "Optically stimulated luminescent glass opti cal fibre dosemeter," Radiat Prot Dosimetry 101, 23-6 (2002). 104A. L. Huston, B. L. Justus, P. L. Falkenstein, R. W. Miller, H. Ning, and R. Altemus, "Remote optical fiber dosimetry," Nuclear in struments & methods in physics research Section B Beam interactions with materi als and atoms. 184, 55 (13 pages) (2001).
363 105L. A. DeWerd, J. F. Wochos, and J. R. Cameron, Reduced dose mammography: Wisconsin Mammographic Phantoms, (Masson Pub. USA, New York, 1979). 106V. K. Prechtel, "Mastopathie und a ltersabhangige Brus tdrusenveranderungen (Mastopathy and Age-dependent Gland Cha nges in Breasts)," Fortschr Med 89, 13121315 (1971). 107International Commission on Radi ation Units and Measurements., Photon, electron, proton, and neutron interac tion data for body tissues, (International Commission on Radiation Units and Measurements, Bethesda, Md., U.S.A., 1992). 108H. Q. Woodard, and D. R. White, "The comp osition of body tissues," Br J Radiol 59, 1209-18 (1986). 109P. P. Fatouros, S. E. Skubic, and G. H., "Development and Use of Realistically Shaped Tissue Equivalent Phantoms for Asse ssing the Mammographic Process," in 74th Scientific Assembly and Annual Meeting of the Radiological Society of North America, edited by (Radiological Society of Nort h America, Chicago, IL, 1988), pp. 110S. E. Skubic, and P. P. Fatouros, "Absor bed breast dose: dependence on radiographic modality and technique, and breast thickness," Radiology 161, 263-70 (1986). 111CIRS, "Tissue-Equivalent Phantom for Mammography:A Refined Quality Assurance Tool for Today's Advanced Imaging Systems," (2005). 112CIRS, "User Guide & Technical Information," in edited by (CIRS Tissue Simulation & Phantom Technology, 2428 Almeda Ave.,Su ite 212, Norfolk, VA 23513, 2003), pp. 14. 113W. P. Argo, K. Hintenlang, and D. E. Hint enlang, "A tissue-equivalent phantom series for mammography dosimetry," J Appl Clin Med Phys 5, 112-9 (2004). 114International Commission on Radi ation Units and Measurements., Tissue substitutes in radiation dosimetry and measurement, (International Commissi on on Radiation Units and Measurements, Bethesda, Md., U.S.A., 1989). 115F. H. Attix, "Introduction to radiological phy sics and radiation dosimetry," in edited by (Wiley, New York, 1986), pp. xxi, 607 p.: ill.; 24 cm. 116A. K. Jones, D. E. Hintenlang, and W. E. Bolch, "Tissue-equivalent materials for construction of tomographic dosimetry phant oms in pediatric radiology," Med Phys 30, 2072-81 (2003). 117H. E. Johns, and J. R. Cunningham, The physics of radiology, 4th (Charles C. Thomas, Springfield, Ill., U.S.A., 1983). 118D. R. White, "The Design and Manuf acture of Anthropomorphic Phantoms," International Commission on Radiation Units and Measurements News June, 4-14 (1992).
364 119J. M. Lewin, R. E. Hendrick, C. J. D'Orsi, P. K. Isaacs, L. J. Moss, A. Karellas, G. A. Sisney, C. C. Kuni, G. R. Cutter, H. Ulger, N. Erdogan, S. Kuman lioglu, and E. Unur, "Breast Imaging Comparison of Full-Fie ld Digital Mammography with Screen-Film Mammography for Cancer Detection: Results of 4,945 Paired Examinations," Radiology 218, 873 (8 pages) (2001). 120J. C. Heggie, "Survey of doses in scr eening mammography," Australas Phys Eng Sci Med 19, 207-16 (1996). 121Kodak, "KODAK LS75 Film Digiti zer-Specifications", 2004). 122"ImageJ," in edited by (National Institutes of Health, 2005). 123W. H. Press, Numerical recipes in fortran 77: the art of scientific computing, 2nd (Cambridge University Press, Cambridge [England]; New York, 1996). 124G. A. Greendale, B. A. Reboussin, S. Slone, C. Wasilauskas, M. C. Pike, and G. Ursin, "Postmenopausal hormone therapy and change in mammographic density," J Natl Cancer Inst 95, 30-7 (2003). 125L. A. Venta, R. E. Hendrick, Y. T. Adler, P. DeLeon, P. M. Mengoni, A. M. Scharl, C. E. Comstock, L. Hansen, N. Kay, A. Covele r, and G. Cutter, "Rates and Causes of Disagreement in Interpretation of Full-Fi eld Digital Mammography and Film-Screen Mammography in a Diagnostic Setting," Am. J. Roentgenol. 176, 1241-1248 (2001). 126X. H. Wang, W. F. Good, B. E. Chapman, Y. H. Chang, W. R. Poller, T. S. Chang, and L. A. Hardesty, "Automated assessment of the composition of breast tissue revealed on tissue-thickness-corrected mammography," AJ R Am J Roentgenol 180, 257-62 (2003). 127M. R. Hersh, "Imaging the dense breast" A pplied radiology 33, 22 (6 pages) (2004). 128P. Cross, "Doses in mammography: from the phantom to the patient," Australas Radiol 38, 20-3 (1994). 129F. M. Hall, "Mammographic density cat egories," AJR American journal of roentgenology 178, 242-3 (2002). 130M. Jeffreys, R. Warren, G. D. Smith, a nd D. Gunnell, "Breast density: agreement of measures from film and digital image," Br J Radiol 76, 561-3 (2003). 131J. W. Byng, N. F. Boyd, E. Fishell, R. A. Jong, and M. J. Yaffe, "The quantitative analysis of mammographi c densities," Phys Med Biol 39, 1629-38 (1994). 132J. W. Byng, M. J. Yaffe, R. A. Jong, R. S. Shumak, G. A. Lockwood, D. L. Tritchler, and N. F. Boyd, "Analysis of mammogra phic density and breast cancer risk from digitized mammograms," Radi ographics 18, 1587-98 (1998).
365 133G. T. Barnes, and I. A. Brezovich, "The intensity of scattered radiation in mammography," Radiology 126, 243-7 (1978). 134J. M. Boone, K. K. Lindfors, V. N. Cooper, 3rd, and J. A. Seibert, "Scatter/primary in mammography: comprehensive results ," Med Phys 27, 2408-16 (2000). 135W. W. Daniel, Biostatistics: a foundation for analysis in the health sciences, 5th (Wiley, New York, 1991). 136A. J. Reilly, and D. Sutton, Report 78 Spectrum Processor: Catalogue of diagnostic xray spectra and other data, (Institute of Physics and E ngineering in Medicine, York, England, 1997). 137J. J. Heine, and R. P. Velthuizen, "A statistical methodology for mammographic density detection," Med Phys 27, 2644-51 (2000). 138L. A. Benevides, "Phone conversation about adoption of ACR's av erage glandular dose measurement procedures.," 2005. 139J. A. Terry, R. G. Waggener, and M. A. Miller Blough, "Half-value layer and intensity variations as a function of position in the radiation field for film-screen mammography," Med Phys 26, 259-66 (1999). 140W. T. Sobol, U. o. A. H. Wu X Affiliation: Department of Radiology, and U. o. A. a. B. U. S. A. Clinics, "Parametrization of mammography normalized average glandular dose tables," Medical physics 24, 547-54 (1997). 141R. R. Monson, National Research Council (U.S.). Committee on the Biological Effects of Ionizing Radiations., National Research Council (U.S.). Board on Radiation Effects Research., and National Research Council (U.S.). Commission on Life Sciences., Health effects of exposure to low levels of ionizing radiation (Uncorrected Proof), (National Academy Press, Washington, D.C., 2005).
366 BIOGRAPHICAL SKETCH Luis Alberto do Rego Benevides was bor n in Angra do Heroismo, Terceira, Portugal, in 1960 to Luis Alberto Gomes Machado Benevides and Noemia Fernades Benevides. He attended a nd graduated from Livingst on High School, Livingston, California. He attended and graduated from Merced College, Merced, California, with an Associate of Science in science in 1983. He received his Bachelor of Arts in biology from Thomas A. Edison State College, Trenton, New Jersey, in 1986. He attended and received a Master of Science in radiation science from Georgetown University, Washington, D.C., in 1996. He received his second Master of Science degree in health services administration from Central Michig an University, Mt Pleasant, Michigan, in 1998. Throughout his educational career, he ha s been a service member in one capacity or another, starting out as an enlisted member of the U.S. Army in 1983 and then being commissioned as an officer of the U.S. Navy in 1987. As Lieutenant Commander in the U.S. Navy, he was selected to pursue his doc toral degree at the University of Florida under the hospice of the Navy's full-time Outservice Training Program. He received his Doctor in Philosophy degree in August 2005 in medical physics from the College of Engineering of the University of Florida.