AN IMAGE-BASED SKELETAL DOSIMETRY MODEL FOR THE ICRP REFERENCE ADULT FEMALE INTERNAL ELECTRON SOURCES

Transcription

1 AN IMAGE-BASED SKELETAL DOSIMETRY MODEL FOR THE ICRP REFERENCE ADULT FEMALE INTERNAL ELECTRON SOURCES By SHANNON E. O'REILLY A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA

2 2013 Shannon E. O'Reilly 2

3 To my parents 3

4 ACKNOWLEDGMENTS I would like to thank my committee chair, Dr. Wesley Bolch, for all the guidance and opportunities he has provided. I would like to thank my committee members, Dr. David Hintenlang and Dr. Lynn Rill, for all the knowledge they have instilled in me and the time they have taken to provide me with direction. I would also like the thank Dr. Didier Rajon for his help in segmenting and resampling the images used in this study. I am indebted to the previous students who developed the methodology behind this study: Deanna Pafundi, Matthew Hough, Lindsay Sinclair, and Michael Wayson. Finally, I would like to thank Matthew Maynard and David Borrego for their invaluable assistance with scripting and programming and Emily Marshall for her efforts with data processing. 4

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS... 4 LIST OF TABLES... 7 LIST OF FIGURES LIST OF ABBREVIATIONS ABSTRACT CHAPTER 1 INTRODUCTION Skeletal Structure and Formation Skeletal Dosimetry Targets Previous Dosimetry Studies University of Leeds Bone Dosimetry Unit Snyder Study: MIRD Pamphlet No Cristy and Eckerman Studies: MIRDOSE Bouchet et al. Study Stabin and Siegel Study: OLINDA/EXM Image-based Models Types of models Hough et al. study Improvements in Skeletal Dosimetry Reasoning Behind Shallow Marrow Thickness Redefinition MATERIALS AND METHODS Cadaver Selection and Image Acquisition Macrostructure Modeling Vertebrae Lone Bone Shaft Modeling Volume Error Corrections Microstructure Modeling Cellularity Radiation Transport Macrostructure Runs Long Bone Shafts Microstructure Runs Physics Considerations

6 3 RESULTS AND DISCUSSION Volume Fractions and Mass Distributions Absorbed Fractions and Specific Absorbed Fractions Skeletal Regions with Multiple Samples Computing absorbed and specific absorbed fractions Examining sample contributions Skeletal Average Computations for adult female model Comparisons with previous dosimetry models Effects of Varying Cellularity CONCLUSIONS AND FUTURE WORK Future Work Sensitivity Study on Tagging of the Shallow Marrow Layer Trabecular Bone Surfaces S-Values APPENDIX A B C D TABLES OF SKELETAL SITE-SPECIFIC SPECIFIC ABSORBED FRACTIONS TO ACTIVE MARROW TARGETS TABLES OF SKELETAL SITE-SPECIFIC SPECIFIC ABSORBED FRACTIONS TO SHALLOW MARROW TARGETS FIGURES OF SKELETAL SITE-SPECIFIC ABSORBED FRACTIONS TO ACTIVE MARROW TARGETS FIGURES OF SKELETAL SITE-SPECIFIC ABSORBED FRACTIONS TO SHALLOW MARROW TARGETS LIST OF REFERENCES BIOGRAPHICAL SKETCH

7 LIST OF TABLES Table page 2-1 The bone size, resulting voxel resolution, and number of voxels for the macrostructure of each skeletal region Matrix dimensions for the separated vertebrae Dimensions used for the long bone shaft cylinder models in MCNPX Matrix dimensions for the skeletal microstructure ICRP 70 defined cellularity factors used for each skeletal site microstructure Elemental compositions and densities used for the material definitions for each skeletal site macrostructure Elemental compositions and densities used for the material definitions for each skeletal site microstructure Spongiosa and cortical volume fractions in the skeletal regions of the adult female Marrow, trabecular, and shallow marrow volume fractions as a percentage of spongiosa volume and shallow marrow volume fraction as a percent of marrow volume for the skeletal regions of the adult female Distribution of mineral bone in the present model for the adult female, compared to the mineral bone distributions computed in a study by Johnson and a study by Spiers and Beddoe for an adult Total active marrow and inactive marrow masses for the skeletal regions of the adult female, using ICRP 70 recommended cellularities, and comparison of the total mass values with those stated for the reference adult female in ICRP Total skeletal mass for each skeletal region of the adult female, less that of cartilage, and a comparison of total skeletal mass with the ICRP 89 value for the reference adult female Skeletal averaging parameters for active marrow, inactive marrow, trabecular bone volume, and cortical bone volume sources used for calculating skeletalaveraged absorbed fractions for the UF adult female model Skeletal-averaged absorbed fractions for active marrow and shallow marrow targets in the adult female

8 3-8 Skeletal-averaged specific absorbed fractions (g -1 ) for active marrow and shallow marrow targets in the adult female A-1 Specific absorbed fractions (g -1 ) for active marrow targets in the craniofacial bones A-2 Specific absorbed fractions (g -1 ) for active marrow targets in the mandible A-3 Specific absorbed fractions (g -1 ) for active marrow targets in the cervical vertebrae A-4 Specific absorbed fractions (g -1 ) for active marrow targets in the thoracic vertebrae A-5 Specific absorbed fractions (g -1 ) for active marrow targets in the lumbar vertebrae A-6 Specific absorbed fractions (g -1 ) for active marrow targets in the sternum A-7 Specific absorbed fractions (g -1 ) for active marrow targets in the ribs A-8 Specific absorbed fractions (g -1 ) for active marrow targets in the scapulae A-9 Specific absorbed fractions (g -1 ) for active marrow targets in the clavicles A-10 Specific absorbed fractions (g -1 ) for active marrow targets in the os coxae A-11 Specific absorbed fractions (g -1 ) for active marrow targets in the sacrum A-12 Specific absorbed fractions (g -1 ) for active marrow targets in the proximal humeri A-13 Specific absorbed fractions (g -1 ) for active marrow targets in the proximal femora A-14 Specific absorbed fractions (g -1 ) for active marrow targets in the humeral upper shafts A-15 Specific absorbed fractions (g -1 ) for active marrow targets in the femoral upper shafts B-1 Specific absorbed fractions (g -1 ) for shallow marrow targets in the craniofacial bones B-2 Specific absorbed fractions (g -1 ) for shallow marrow targets in the mandible B-3 Specific absorbed fractions (g -1 ) for shallow marrow targets in the cervical vertebrae

9 B-4 Specific absorbed fractions (g -1 ) for shallow marrow targets in the thoracic vertebrae B-5 Specific absorbed fractions (g -1 ) for shallow marrow targets in the lumbar vertebrae B-6 Specific absorbed fractions (g -1 ) for shallow marrow targets in the sternum B-7 Specific absorbed fractions (g -1 ) for shallow marrow targets in the ribs B-8 Specific absorbed fractions (g -1 ) for shallow marrow targets in the scapulae B-9 Specific absorbed fractions (g -1 ) for shallow marrow targets in the clavicles B-10 Specific absorbed fractions (g -1 ) for shallow marrow targets in the os coxae B-11 Specific absorbed fractions (g -1 ) for shallow marrow targets in the sacrum B-12 Specific absorbed fractions (g -1 ) for shallow marrow targets in the spongiosa regions of the humeri B-13 Specific absorbed fractions (g -1 ) for shallow marrow targets in the spongiosa regions of the radii B-14 Specific absorbed fractions (g -1 ) for shallow marrow targets in the spongiosa regions of the ulnae B-15 Specific absorbed fractions (g -1 ) for shallow marrow targets in the wrists and hands B-16 Specific absorbed fractions (g -1 ) for shallow marrow targets in the shafts of the arm bones B-17 Specific absorbed fractions (g -1 ) for shallow marrow targets in the spongiosa regions of the femora B-18 Specific absorbed fractions (g -1 ) for shallow marrow targets in the patellae B-19 Specific absorbed fractions (g -1 ) for shallow marrow targets in the spongiosa regions of the tibiae B-20 Specific absorbed fractions (g -1 ) for shallow marrow targets in the spongiosa regions of the fibulae B-21 Specific absorbed fractions (g -1 ) for shallow marrow targets in the ankles and feet B-22 Specific absorbed fractions (g -1 ) for shallow marrow targets in the shafts of the leg bones

10 LIST OF FIGURES Figure page 2-1 Pre-segmented, post-filtered microct single slice image of the third cervical vertebra Post-segmented, post-filtered microct single slice image of the third cervical vertebra (using a threshold value of 172) Complete UFHADF phantom and the removed full skeleton of the UFHADF Seperated proximal, upper shaft, lower shaft and distal regions of the adult female femur Image slice from ImageJ TM of the adult female cranium illustrating the streaking artifact that can occur from voxelization Depiction of the process used to separate the vertebrae into individual vertebra in Rhinoceros TM through the use of cutting planes (for the lumbar and thoracic vertebrae) and control point deletion (for the cervical vertebrae). A) Lumbar vertebrae and separated L1 vertebra. B) Thoracic vertebrae and separated T9 vertebra. C) Cervical vertebrae and separated C3 vertebra Mandible of the UFHADF phantom. A) Prior to repair, with spongiosa (pink) breaking through the cortical layer (gray). B) Repaired Image slice of the microstructure of the sacrum (cellularity of 60%) with the shallow marrow labeled (inactive shallow marrow in blue and active shallow marrow in orange) The macrostructure and microstructure (with microct image slice) for the adult female sternum Varying cellularity of the os coxae, with dark gray being the inactive marrow, white active marrow, and black is the mineral bone. A) 10% cellularity. B) ICRP 70 cellularity of 48%. C) 90% cellularity Absorbed fractions for active marrow irradiating active marrow for the sternum of the UF adult female and the UF adult male at ICRP 70 reference cellularity Contributions of frontal, occipital, and parietal bone sample absorbed fractions to the craniofacial absorbed fractions for active marrow selfirradiation at ICRP 70 reference cellularity

11 3-3 Contributions of frontal, occipital, and parietal bone sample absorbed fractions to the craniofacial absorbed fractions to active marrow by trabecular inactive marrow source at ICRP 70 reference cellularity Contributions of frontal, occipital, and parietal bone sample absorbed fractions to the craniofacial absorbed fractions to active marrow by trabecular bone volume source at ICRP 70 reference cellularity Contributions of frontal, occipital, and parietal bone sample absorbed fractions to the craniofacial absorbed fractions to active marrow by cortical bone volume source at ICRP 70 reference cellularity Contributions of frontal, occipital, and parietal bone sample absorbed fractions to the craniofacial absorbed fractions to shallow marrow by trabecular active marrow source at ICRP 70 reference cellularity Comparison of absorbed fraction to active marrow for an active marrow source in the lumbar vertebra (L1-L5) when the separated vertebra were used as the macrostructure and when the entire lumbar vertebral column was used as the macrostructure Comparison of absorbed fraction to shallow marrow for an active marrow source in the lumbar vertebra (L1-L5) when the separated vertebra were used as the macrostructure and when the entire lumbar vertebral column was used as the macrostructure Skeletal-averaged absorbed fractions to active marrow in the adult female by active marrow, inactive marrow, trabecular bone volume, and cortical bone volume sources Comparison of absorbed fractions for active marrow self-irradiation for each bone site that contains active marrow at each skeletal site s ICRP 70 reference cellularity and the resulting skeletal-average value Comparison of absorbed fractions for active marrow with an inactive marrow source for each bone site that contains active marrow at each skeletal site s ICRP 70 reference cellularity and the resulting skeletal-average value Comparison of absorbed fractions for active marrow with cortical bone volume source for each bone site that contains active marrow at each skeletal site s ICRP 70 reference cellularity and the resulting skeletalaverage value Comparison of absorbed fractions for active marrow with trabecular bone volume source for each bone site that contains active marrow at each skeletal site s ICRP 70 reference cellularity and the resulting skeletalaverage value

12 3-14 Comparison of skeletal-averaged absorbed fractions to active marrow by active marrow source of the UFADF from this study to the UFADM, UF15YF, Stabin and Segal AD, and ICRP 110 ADF models Comparison of skeletal-averaged absorbed fractions to active marrow by inactive marrow source of the UFADF from this study to the UFADM and ICRP 110 ADF models Comparison of skeletal-averaged absorbed fractions to active marrow by trabecular bone volume source of the UFADF from this study to the UFADM, UF15YF, Stabin and Segal AD, and ICRP 110 ADF models Comparison of skeletal-averaged absorbed fractions to active marrow by cortical bone volume source of the UFADF from this study to the UFADM and ICRP 110 ADF models Comparison of skeletal-averaged absorbed fractions to shallow marrow by an active marrow source of the UFADF from this study to the UFADM, UF15YF, Stabin and Segal AD, and ICRP 110 ADF models The effect of varying cellularity on absorbed fractions for active marrow selfirradiation in the sternum The effect of varying cellularity on specific absorbed fractions for active marrow self-irradiation in the sternum The lack of effect of varying cellularity on absorbed fractions for shallow marrow when active marrow is the source, for the sternum C-1 Electron absorbed fractions to active marrow targets in the craniofacial bones for TAM, TIM, TBV, and CBV sources C-2 Electron absorbed fractions to active marrow targets in the mandible for TAM, TIM, TBV, and CBV sources C-3 Electron absorbed fractions to active marrow targets in the cervical vertebrae for TAM, TIM, TBV, and CBV sources C-4 Electron absorbed fractions to active marrow targets in the thoracic vertebrae for TAM, TIM, TBV, and CBV sources C-5 Electron absorbed fractions to active marrow targets in the lumbar vertebrae for TAM, TIM, TBV, and CBV sources C-6 Electron absorbed fractions to active marrow targets in the sternum for TAM, TIM, TBV, and CBV sources

13 C-7 Electron absorbed fractions to active marrow targets in the ribs for TAM, TIM, TBV, and CBV sources C-8 Electron absorbed fractions to active marrow targets in the scapulae for TAM, TIM, TBV, and CBV sources C-9 Electron absorbed fractions to active marrow targets in the clavicles for TAM, TIM, TBV, and CBV sources C-10 Electron absorbed fractions to active marrow targets in the os coxae for TAM, TIM, TBV, and CBV sources C-11 Electron absorbed fractions to active marrow targets in the sacrum for TAM, TIM, TBV, and CBV sources C-12 Electron absorbed fractions to active marrow targets in the proximal humeri for TAM, TIM, TBV, and CBV sources C-13 Electron absorbed fractions to active marrow targets in the proximal femora for TAM, TIM, TBV, and CBV sources C-14 Electron absorbed fractions to active marrow targets in the humeral upper shafts for MAM, MIM, CBS, and CBV sources C-15 Electron absorbed fractions to active marrow targets in the femoral upper shafts for MAM, MIM, CBS, and CBV sources D-1 Electron absorbed fractions to shallow marrow targets in the craniofacial bones for TAM, TIM, TBV, and CBV sources D-2 Electron absorbed fractions to shallow marrow targets in the mandible for TAM, TIM, TBV, and CBV sources D-3 Electron absorbed fractions to shallow marrow targets in the cervical vertebrae for TAM, TIM, TBV, and CBV sources D-4 Electron absorbed fractions to shallow marrow targets in the thoracic vertebrae for TAM, TIM, TBV, and CBV sources D-5 Electron absorbed fractions to shallow marrow targets in the lumbar vertebrae for TAM, TIM, TBV, and CBV sources D-6 Electron absorbed fractions to shallow marrow targets in the sternum for TAM, TIM, TBV, and CBV sources D-7 Electron absorbed fractions to shallow marrow targets in the ribs for TAM, TIM, TBV, and CBV sources

14 D-8 Electron absorbed fractions to shallow marrow targets in the scapulae for TAM, TIM, TBV, and CBV sources D-9 Electron absorbed fractions to shallow marrow targets in the clavicles for TAM, TIM, TBV, and CBV sources D-10 Electron absorbed fractions to shallow marrow targets in the os coxae for TAM, TIM, TBV, and CBV sources D-11 Electron absorbed fractions to shallow marrow targets in the sacrum for TAM, TIM, TBV, and CBV sources D-12 Electron absorbed fractions to shallow marrow targets in the proximal humeri for TAM, TIM, TBV, and CBV sources D-13 Electron absorbed fractions to shallow marrow targets in the distal humeri for TIM, TBV, and CBV sources D-14 Electron absorbed fractions to shallow marrow targets in the proximal radii for TIM, TBV, and CBV sources D-15 Electron absorbed fractions to shallow marrow targets in the distal radii for TIM, TBV, and CBV sources D-16 Electron absorbed fractions to shallow marrow targets in the proximal ulnae for TIM, TBV, and CBV sources D-17 Electron absorbed fractions to shallow marrow targets in the distal ulnae for TIM, TBV, and CBV sources D-18 Electron absorbed fractions to shallow marrow targets in the wrists and hands for TIM, TBV, and CBV sources D-19 Electron absorbed fractions to shallow marrow targets in the humeral upper shafts for MAM or MIM, CBS, and CBV sources D-20 Electron absorbed fractions to shallow marrow targets in the humeral lower shafts for MIM, CBS, and CBV sources D-21 Electron absorbed fractions to shallow marrow targets in the radial shafts for MIM, CBS, and CBV sources D-22 Electron absorbed fractions to shallow marrow targets in the ulnar shafts for MIM, CBS, and CBV sources D-23 Electron absorbed fractions to shallow marrow targets in the proximal femora for TAM, TIM, TBV, and CBV sources

15 D-24 Electron absorbed fractions to shallow marrow targets in the distal femora for TIM, TBV, and CBV sources D-25 Electron absorbed fractions to shallow marrow targets in the patellae for TIM, TBV, and CBV sources D-26 Electron absorbed fractions to shallow marrow targets in the proximal tibiae for TIM, TBV, and CBV sources D-27 Electron absorbed fractions to shallow marrow targets in the distal tibiae for TIM, TBV, and CBV sources D-28 Electron absorbed fractions to shallow marrow targets in the proximal fibulae for TIM, TBV, and CBV sources D-29 Electron absorbed fractions to shallow marrow targets in the distal fibulae for TIM, TBV, and CBV sources D-30 Electron absorbed fractions to shallow marrow targets in the ankles and feet for TIM, TBV, and CBV sources D-31 Electron absorbed fractions to shallow marrow targets in the femoral upper shafts for MAM or MIM, TBV, and CBV sources D-32 Electron absorbed fractions to shallow marrow targets in the femoral lower shafts for MIM, CBS, and CBV sources D-33 Electron absorbed fractions to shallow marrow targets in the fibular shafts for MIM, CBS, and CBV sources D-34 Electron absorbed fractions to shallow marrow targets in the tibial shafts for MIM, CBS, and CBV sources

16 LIST OF ABBREVIATIONS 3D AD ADF AF ALRADS AM BS CB CBIST CBS HC CBS MC CBV CBVF CF CT EF three dimensional adult adult female absorbed fraction Advanced Laboratory for Radiation Dosimetry Studies active marrow bone surface cortical bone chord-based infinites spongiosa transport cortical bone surface along Haversian canals cortical bone surface of medullary cavity cortical bone volume cortical bone volume fraction cellularity factor computed tomography escape fraction EGS4 Electron-Gamma-Shower version 4 EGSnrc ICRP IM ITS MAM MATLAB TM MB Electron-Gamma-Shower National Research Council International Commission on Radiological Protection inactive marrow Integrated Tiger Series medullary active marrow Matrix Laboratory mineral bone 16

17 MC MCNPX MIM MIRD MST MV NMR NURBS OLINDA/EXM PIRT PRESTA ROI SAF SIRT SMVF SP SV SVF TAM TIM TB TBS TBV TBVF TM medullary cavity Monte Carlo N-Particle Extended medullary inactive marrow Medical Internal Radiation Dose committee Miscellaneous skeletal tissue marrow volume nuclear magnetic resonance non-uniform rational b-splines Organ Level Internal Dose Assessment/Exponential Modeling paired-image radiation transport Parameter-Reduced Electron Step Transport Algorithm region of interest specific absorbed fraction single-image radiation transport shallow marrow volume fraction spongiosa spongiosa volume spongiosa volume fraction trabecular active marrow trabecular inactive marrow trabecular bone trabecular bone surface trabecular bone volume trabecular bone volume fraction total marrow 17

18 TM 50 UF UF15YF UFADF UFADM UFHADF VBIST VBRST VERT-L total shallow marrow University of Florida UF fifteen year old female model UF adult female model UF adult male model UF adult hybrid female computational phantom Voxel-Based Infinite Spongiosa Transport Voxel-Based Restricted Spongiosa Transport lumbar vertebrae 18

19 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science AN IMAGE-BASED SKELETAL DOSIMETRY MODEL FOR THE ICRP REFERENCE ADULT FEMALE INTERNAL ELECTRON SOURCES Chair: Wesley Bolch Major: Biomedical Engineering By Shannon E. O Reilly August 2013 The geometries of the skeletal regions of the body are difficult to model, particularly those comprising the bone trabeculae and bone marrow. Irradiation of hematopoietically active bone marrow and osteoprogenitor cells that line the bone surfaces has been found to induce radiogenic leukemia and radiogenic bone cancer, respectively. The active marrow and bone endosteum are typically adopted as surrogate tissue regions for dosimetric assessment of these cell populations. Most previous skeletal dosimetry models do not account for electron escape and cross-fire from cortical bone, assume regions of infinite spongiosa, disregard varying cellularity effects on active marrow self-irradiation, and do not utilize the more recent ICRP definition of a 50 micron surrogate tissue region for the osteoprogenitor cells shallow marrow. Each of these limitations was addressed in the present dosimetry model for the ICRP reference adult female. Electron transport was performed to determine specific absorbed fractions to active marrow and shallow marrow of the skeletal regions of the adult female. Macrostructures and microstructures were modeled separately. Individual bone macrostructures were obtained from the whole-body hybrid computational phantom of 19

20 the UF series of reference phantoms, while microstructures were derived from microct images of sampled skeletal regions obtained from a 45-year-old female cadaver. Source regions considered were active marrow, inactive marrow, trabecular bone volume, cortical bone volume and cortical bone surfaces. Marrow cellularities were varied from 10 to 100 percent for active marrow self-irradiation simulations. A total of 33 discrete electron energies, ranging from 1 kev to 10 MeV, were either simulated or modeled analytically. The method of combining macro- and microstructure absorbed fractions calculated using MCNPX electron transport was found to yield results similar to those determined with the PIRT model for the UF adult male in the Hough et al. study. The calculated skeletal averaged absorbed fractions for each source-target combination were found to follow similar trends of more recent dosimetry models (image-based models) and did not follow the past Stabin and Segal model at high energies (due to that models use of an infinite expanse of trabecular spongiosa). 20

21 CHAPTER 1 INTRODUCTION Skeletal Structure and Formation Bone is primarily an organic matrix comprised principally of the protein collagen and other proteins, carbohydrates, and lipids. 1 Calcium phosphate constitutes the inorganic portions of bone. The organic matrix is formed by the osteoblasts, or boneforming cells, and then mineralized. This process results in the hard structure, which is constantly being remodeled by the osteoclasts, bone-resorbing cells and the osteoblasts. 1, 2 Osteoblasts are a derivative of osteoprogenitor cells, 3 while osteoclasts are derived from hematopoietic stem cells. 4 There are two main types of bone: cortical and trabecular. Cortical bone, also known as compact bone, is the hard, dense structure that forms the outer portions of all skeletal sites, and is heavily partitioned on the long bone shafts. Trabecular bone is much more porous, and for this reason it is also known as spongy or cancellous bone. Trabecular bone is partitioned in the interior regions of the skeletal regions and in the ends of the long bones. 1 The combined tissues of the bone trabeculae and marrow are typically known as spongiosa. Instead of a spongiosa region, the long bone shafts in adult subjects contain a medullary cavity, almost entirely comprised of inactive marrow (with the exception of the upper shafts of the femora and humeri), and is devoid of bone trabeculae. The layer of cells that line the surfaces of bone that border bone marrow is known as the endosteum. 1 Skeletal Dosimetry Targets The skeletal regions of the body are difficult to model for dosimetric purposes due to their complex 3D geometry in the regions that comprise the trabecular bone and 21

22 bone marrow. The skeletal system is an amalgamation of tissues which include cortical and trabecular bone, bone marrow, periosteum, endosteum, and cartilage. 1 The irradiation of the hematopoietically active (or red) bone marrow has been shown to cause radiogenic leukemia. 5 Irradiating the osteoprogenitor cells that line the trabecular bone surface or surfaces of the cortical bone in the long bone shafts has been found to lead to radiogenic bone cancer. In bone dosimetry studies, red marrow is used as a surrogate for hematopoietically active bone marrow and endosteum is used as a surrogate for the osteoprogenitor cells along the bone surfaces. For these reasons, the International Commission on Radiological Protection (ICRP) has assigned tissue weighting factors for red marrow and endosteal surfaces. 5 Active marrow is often the dose-limiting organ in radiotherapy, due to concerns with marrow toxicity. 6, 7 Previous Dosimetry Studies Currently, there are two main classes of models for bone dosimetry: chord-length based models and image-based models. In chord-length based models, the particles are modeled as traveling in straight paths through the trabecular bone and bone marrow. 8 The distance the particle travels is thought to be equal to the measured and sampled chord-length following the assembly of omni-directional distributions of these chord lengths. The bone model provided in ICRP Publication 30 is a chord-based model. 9 For image-based models, three-dimensional images are taken of the trabeculae through the use of nuclear magnetic resonance (NMR) microscopy or microcomputed tomography (microct). 10 These images are subsequently used as the inputs for the geometry of the bone structure for radiation transport calculations. 22

23 University of Leeds Bone Dosimetry Unit The foundation of chord-length based models originated from the work of Frederick Spiers and the Bone Dosimetry Research Unit, which included PhD student Phillip Darley, Joanne Whitwell, and Alun Beddoe, at the University of Leeds. 11 A 44- year-old adult male cadaver was used in the Spiers study. Contact radiographs were taken of seven bone sites: parietal bone, cervical vertebra, lumbar vertebra, rib, iliac crest, femur head, and femur neck. 12 It is important to note that the skeletal masses used were taken from a previous study by Mechanik and were not determined based on the bone sites and cadaveric subject used in the Leeds study. 13 Spiers PhD student Philip Darley created the object plane scanning microscope that was used to determine the pathlength distributions through both marrow cavities and across the bone trabeculae. 11 This was necessary in order to analyze the anisotropic nature of the spongiosa, and it was the data that arose from the use of this bone scanner that allowed for the creation of omnidirectional pathlength distributions. Whitwell 12 determined mean dose factors for the red bone marrow and endosteal tissues for the seven bone sites in the Leeds 44-year-old adult human skeleton. The endosteal thickness used at that time was assumed to be 10 m. The radionuclides modeled in this study were 14 C, 45 Ca, 22 Na, 18 F, 32 P, 90 Y, and 90 Sr. Whitwell performed Monte Carlo calculations using the previously determined pathlength distributions to determine these mean dose factors. She modeled the beta particles as being emitted uniformly throughout the volume of the mineral bone. The pathlength distributions determined by this dosimetry group were used as the foundation for the majority of the skeletal dosimetry models that followed. Such 23

24 models include those in MIRD Pamphlet No. 11 and ICRP Publication 30, as well as the software codes MIRDOSE3 and OLINDA/EXM, which will be discussed further. 10 Snyder Study: MIRD Pamphlet No. 11 In MIRD Pamphlet No. 11, only the average dose to the bone is provided and is said to be due to the total energy of the electron, less the energy deposited in the marrow. 14 This does not allow for electron escape from the cortical bone. The report states that if only a total value (time-integrated activity) to the bone is provided then it is assumed that half is in the cortical bone and half is in the trabecular bone. A uniform distribution of the radionuclide through the bone volume is assumed. In this pamphlet, the radionuclide S values for the total bone and red marrow are provided for 117 radionuclides. 14 The absorbed fractions used in this pamphlet for photons of energies less than 300 kev for bone irradiating red marrow have been said to be high and do not account for charged particle disequilibrium at the levels of the individual marrow cavities. 15 Cristy and Eckerman Studies: MIRDOSE 3 With the work of Cristy and Eckerman came the introduction of modeling the energy deposited by secondary electrons from photon interactions. 15 Eckerman utilized Spiers chord-length distributions and methodology for the transport of electrons to determine electron absorbed fractions for seven bone types. The fractional abundance of each bone type in the fifteen different skeletal regions in the Cristy and Eckerman phantom series was used to determine absorbed fractions for these regions. 15 These electron absorbed fractions were computed over the energy range of to 4 MeV

25 Skeletal averaged monoenergetic electron absorbed fractions were provided for a newborn, 1-year-old, 5-year-old, 10-year-old, 15-year-old, and adult. 16 Also, in this study, the cellularity factor, defined as the fraction of marrow that is hematopoietic active, is taken into account. It is noted that the cellularity factor for different bone sites changes with age. For the calculation of the absorbed fractions, the targets were stated to be red marrow and bone surfaces. Bone surfaces are further defined as the 10 m endosteal tissue. The sources are red marrow, cortical bone surfaces, cortical bone volume, trabecular bone surfaces and trabecular bone volume. 16 The Eckerman model is used in MIRDOSE 3, 15 which is an internal dose calculation software program utilized for commonly used radionuclides in nuclear medicine. 17 MIRDOSE 3 was written by Dr. Michael Stabin. 17 Dose conversion factors and absorbed doses for individual skeletal regions and skeletal average are provided in this software for adults and children of varying age. Bouchet et al. Study A model for electron transport through the trabecular and cortical bone was developed by Bouchet et al. 18 The basis of this model still lies in the chord-length distributions developed by Spiers, however three-dimensional transport was implemented through the use of the EGS4-Parameter-Reduced Electron Step Transport Algorithm (PRESTA) Monte Carlo code. Electron absorbed fractions were determined for seven trabecular bone sites. The source and target regions for the trabecular bone used for this model were the trabecular bone volume, the marrow space, and the trabecular bone endosteum. The source and target regions used for the cortical bone were the cortical bone volume, cortical Haversian canals, and the cortical bone 25

26 endosteum. Electron absorbed fractions were determined for three cortical bone sites: the cortex of the humerus, femur, and tibia. From these absorbed fractions, Bouchet was able to determine radionuclide S-values for 22 different bone sites. In this study, Bouchet was able to use data from the newer ICRP publication, ICRP Stabin and Siegel Study: OLINDA/EXM Stabin and Siegel 19 created a revised bone dosimetry model by essentially taking portions of the Eckerman model and the Bouchet model. In obtaining the values for low energies, the Bouchet model was utilized, while the Eckerman model was utilized for medium to high energies. The Bouchet model was used for the low energy component because at low energies the range of the electrons should be comparable to the cells within the marrow space, meaning the electron absorbed fractions should approach unity, not the cellularity factor, as they did in the Eckerman model. The increase in electron energy would cause the electron range to increase, leading to the electrons traversing multiple cells, and at high energies crossing marrow cavities before complete energy deposition. This means that the energy deposited in the active marrow would be proportional to the cellularity factor. 19 The work in this study is used in the code OLINDA/EXM, which was written by Stabin. OLINDA/EXM is a code used in nuclear medicine studies to calculate organ doses and effective doses. 20 Image-based Models Early work done by the University of Florida with image-based models showed that the Eckerman model was underestimating the absorbed fractions for total red marrow irradiating total red marrow for energies below 200 kev. 15 It also showed that the Bouchet model overestimated the absorbed fraction of active marrow self-irradiation for energies above 20 kev. 26

27 Types of models Currently, the three main forms of image-based models are Voxel-Based Infinite Spongiosa Transport (VBIST), Voxel-Based Restricted Spongiosa Transport (VBRST), and Paired-Image Radiation Transport (PIRT). 8 Each of these bone dosimetry models utilize NMR or microct imaging of the spongiosa, providing a three-dimensional model. These methods differ on the way the macrostructure is defined. In the VBIST method, there is no macrostructure defined and infinite spongiosa transport is modeled, as seen with the previous chord-length based models. A stylized model of the cortical bone is used to represent the macrostructure in the VBRST model. The PIRT model uses exvivo CT imaging of the bone for the macrostructure. In PIRT simulations, the macrostructure and microstructure are run simultaneously, allowing and accounting for electron escape from the spongiosa, into the cortical bone. 8 Hough et al. study A study conducted by Hough et al. 10 used the PIRT method to create a skeletal dosimetry model for the ICRP adult male. The macrostructure was derived through an ex-vivo whole body CT scan, at 1 mm resolution, of a 40-year-old male cadaver. For the geometry of the microstructure ex-vivo microct images were taken of spongiosa samples from 38 bone sites, at 30 m resolution. For the radiation transport the EGSnrc-based Paired Image Radiation Transport code by Shah was used. The electron energies ranged from 1 kev to 10 MeV and the electrons were monoenergetic. The targets in this study were the active marrow and the endosteum. The endosteum was extended to 50 m, rather than the previously used 10 m, due to the change in the definition by the ICRP. The cortical bone volume, cortical bone surfaces, trabecular 27

28 bone volume, trabecular bone surfaces, inactive marrow and active marrow were sources in this study. The SAFs were computed for the range of cellularities from 10% to 100% (in increments of 10%) for total active marrow source to total active marrow target. Improvements in Skeletal Dosimetry Four main areas of potential improvement can be identified within many previous dosimetry models, of which are still employed in commonly used software today. These areas include the lack of modeling electron escape from cortical bone, the modeling of infinite spongiosa, the disregard of the effect of varying cellularity on active marrow selfirradiation, and the use of the outdated ICRP definition of a shallow marrow thickness (10 microns rather than the current 50 microns). 10 Each limitation was accounted for in the present dosimetry model. Reasoning Behind Shallow Marrow Thickness Redefinition The altering of the definition for the thickness of the endosteal layer arose from a study on radium dial painters and radium induced fibrosarcomas. 3 A fibrotic layer of 50 microns, located between the bone surfaces and marrow cells, was observed through the use of an electron microscope. This replacement tissue layer was thought to be correlated with the tumor induction. 28

29 CHAPTER 2 MATERIALS AND METHODS Cadaver Selection and Image Acquisition For this study, a 45 year-old female cadaver was used. The macrostructure was derived from a whole body in-vivo CT and ex-vivo CT images taken after the harvesting of the cadaver bones. These CT scans were performed on a 64-slice Toshiba CT scanner in the Department of Radiology at Shands Hospital at the University of Florida (UF). For the ex-vivo CT imaging the following skeletal sites were excised, due to the presence of active marrow: the cranium, mandible, clavicles, scapulae, sternum, vertebrae (cervical, thoracic, and lumbar), ribs (upper, middle, and lower), os coxae, sacrum, patellae, and the proximal and distal ends of the humeri, radii, ulnae, femora, tibiae, and fibulae. For each skeletal site, two ex-vivo scans were completed - each with a 1mm slice thickness - one with a bone filter, and the other with a soft tissue filter. These separate images were acquired to optimize the visualization of the boundary between the cortical and spongiosa regions. From these images and the bone mass values provided in ICRP Publication 89 for the reference female, targeted cortical bone and spongiosa volumes were determined for each skeletal site. The harvesting and image acquisition was conducted by previous graduate students whom were members of the Advanced Laboratory for Radiation Dosimetry Studies (ALRADS) at the University of Florida. Through inspection of the ex-vivo images, a region of interest was chosen from each bone sample for microct imaging. The microct imaging was performed by SCANCO in Brüttisellen (Switzerland) at an isotropic resolution of 30 microns. These images were later converted to a 50 micron isotropic resolution, for ease in labeling the 29

30 first voxel layers within marrow as the target region shallow marrow. Multiple steps were necessary to convert the microct images into a useful form for radiation transport. First, a region of interest had to be determined for each image, removing any cortical bone from the selected ROI. Next, in order to increase the signal-to-noise ratio of each image, a median filter was applied. An example of an image from a single slice of the third cervical vertebrae after this median filter was applied can be seen in Figure 2-1. Next, a threshold value for the gray level was determined that optimized the appearance of the interface of the trabecular bone and marrow cavities. These steps were completed by Lindsay Sinclair. Finally, using this determined threshold value, the images were segmented into binary images. 7 The segmenting was performed using the Bone Dosimetry and Imaging program. 21 Figure 2-2 shows a post-segmented image slice of the third cervical vertebrae, with a threshold value of 172. The increased ability to delineate the boundaries between the bone trabeculae and marrow cavity is clear from the comparison of the pre- and post-segmented slices of the third cervical vertebrae. Macrostructure Modeling Thirty-four bone sites constitute the modeled macrostructure of the adult female: the cranium, mandible, vertebrae (cervical, thoracic, lumbar), sternum, ribs, scapulae, clavicles, os coxae, sacrum, humeri, radii, ulnae, wrists and hands, femora, tibiae, fibulae, patellae and ankles and feet. Each bone site was isolated from the UF hybrid adult female (UFHADF) phantom, seen in Figure 2-3, to be voxelized for the geometry input into MCNPX (Los Alamos National Laboratory, Los Alamos, NM). The long bones were divided into the proximal end, shaft, and distal end. The shafts of the femora and humeri were further divided into the upper and lower shaft, due to the 30

31 presence of active marrow in the upper shafts and not in the lower. 2 Figure 2-4 shows this division of the femur. For the appendicular skeleton, only one side - the right side of the body - was modeled, with the exception of the os coxae. The approximate height, width, and depth of the bone macrostructure sites were measured in Rhinoceros TM (McNeel North America, Seattle, WA) in order to determine an isotropic voxel resolution, with Equation ( ) (2-1) where r isotropic is the isotropic voxel resolution, X is the measured height (in cm), Y is the measured width (in cm), Z is the measured depth (in cm), and N is the target voxel matrix size. The targeted matrix size was 5.42 x 10 7, as previously determined by Wayson 22 to be the total number of voxel elements that can be efficiently modeled in MCNP, with the given computing resources. Due to the existence of a larger prevalence of a streaking artifact with some of the macrostructure bone sites after voxelization, the voxel resolutions of these sites were altered to help alleviate this problem. An example of this streaking artifact in the voxel version of the cranium is seen in Figure 2-5. The voxel resolutions ranged from 149 to 670 microns, for the proximal radii and ribs respectively, and are listed in Table 2-1, along with the measured bone dimensions. Vertebrae MicroCT images were obtained from multiple vertebral samples, including L1-L5, T1, T3, T6, T9, T12, C3 and C6. Consequently, the vertebral column of the reference phantom was partitioned into these individual vertebrae. Vertebral partitioning was accomplished with various Rhinoceros TM commands, primarily cutting planes and 31

32 control point deletion and manipulation. The cutting planes were used for the lumbar and thoracic vertebrae, while control points were used for the cervical vertebrae. This more complicated method was employed for the cervical vertebrae due to the more overlapping nature of the vertebra, and the need to properly isolate them. An image illustrating this process is provided in Figure 2-6. The voxel element matrix sizes for these separated vertebrae are provided in Table 2-1. The separated lumbar vertebrae were voxelized at an isotropic resolution of 0.18 mm and the separated cervical and thoracic vertebrae at an isotropic resolution of 0.20 mm. Lone Bone Shaft Modeling The long bone shafts were modeled in MCNPX using cylinders. The effective radius of the medullary cavity was determined by using Equation (2-2) where r medullary marrow is the effective radius of the medullary cavity (in cm), V medullary marrow is the combined volume of the medullary marrow of the left and right shaft for that skeletal site (in cm 3 ), and h is the length of the shaft (in cm). The length of each shaft was measured in Rhinoceros TM. The cortical bone surrounding the long bone shafts was modeled as a cylindrical shell in MCNPX. The outer radius of this cortical bone shell was calculated using Equation 2-3. (2-3) where r cortical bone is the effective outer radius of the cortical bone shell (in cm), V is the entire volume of the left and right shaft (in cm 3 ), and h is the length of the shaft (in cm). The shallow marrow was modeled as a 50 micron thick shell just within the cylindrical 32

33 volume of the medullary marrow cavity. Therefore, the outer radius of the medullary cavity and the outer radius of the shallow marrow shell are equivalent. The outer radii of the medullary cavities, inner radius of the shallow marrow shell, outer radius of the cortical bone shell, and the heights of the shafts are provided in Table 2-3. Volume Error Corrections Upon voxelization, the volumes of three bone sites (the cranium, mandible and the entire set of cervical vertebrae) in the UFHADF phantom did not match their expected values. For the cranium and mandible, the disparity was due to an offset mesh process intended to separate cortical and spongiosa regions during phantom construction. However, as seen in Figure 2-7A, spongiosa was left protruding through the cortical layer of the mandible as a result of the process. A similar effect was also observed in the cranium. A broad set of Rhinoceros TM smoothing and mesh repair tools were employed to correct these issues. The repaired mandible can be seen in Figure 2-7B. The spongiosa volume of the cervical vertebrae was found to be smaller than the desired value by 25.4%. The cortical volume was 30.6% below the targeted value. These volumes were adjusted by using the 3D scale tool in Rhinoceros TM. Microstructure Modeling The following 37 skeletal sites were cored and imaged, therefore constituting the microstructure of the adult female: C3, C6, L1-L5, T1, T3, T6, T9, T12, clavicle, craniofacial bones (frontal, parietal, occipital), mandible, os coxae, patella, sacrum, scapula, sternum, femur (distal, proximal head, proximal neck), and the proximal and distal ends of the fibula, humerus, radius, tibia, and ulna. The microstructure was not 33

34 available for the wrists and hands and ankles and feet; subsequently the distal femora and distal humeri were used as surrogates. Each microct image was converted (from the 30 micron imaging resolution) to have an isotropic voxel resolution of 50 m. This resolution was chosen to be the same as the definition of the shallow marrow thickness, in order to label the first layer of marrow voxels that surrounded bone trabeculae as the endosteal target region. The labeling of the shallow marrow can be seen in Figure 2-8, with the inactive shallow marrow in orange and the active shallow marrow in blue. The matrix size for each microstructure sample is provided in Table 2-4. The inactive marrow and active marrow portions of the shallow marrow had different tag numbers for tallying purposes in MCNPX. This process was followed due to the total active marrow also being a target region; therefore, the portion of active marrow in TM 50 needed to be accounted for in the calculation of total marrow absorbed fractions as well as the calculation for shallow marrow absorbed fractions. Figure 2-9 shows an example of a 3D rendering of the macrostructure and microstructure (derived from microct images, one of which is represented) for the sternum. Cellularity Marrow cellularity is defined as the fraction of total marrow space that is hematopoietically active, and is taken to be the percent of total marrow volume that is occupied by active bone marrow. 7 As inactive (or yellow) marrow is primarily composed of adipocytes, the marrow cellularity defined in this study is equivalent to one minus the marrow fat fraction. ICRP Publication 70 provides reference values for the cellularity of each skeletal site for the adult female, which are provided in Table For the humeri 34

35 and femora ICRP 70 provides a cellularity factor of 0.25 (or 25%) for the upper half of the long bone and does not specify the distribution among the proximal end and upper shaft. 2 For this study, the proximal ends were given a cellularity factor of 0.35 and the upper shafts a cellularity factor of 0.15, resulting in a linear average of 0.25 for the upper halves of the humeri and femora. As individual patient s may have marrow cellularities that differ from ICRP reference values, versions of each microstructure were created in which the cellularity was incremented by 10%, up to 100% (at which point the marrow cavity was comprised entirely of active marrow). A version of each microstructure was also created at its ICRP 70 reference cellularity. An example of the varying cellularity in the microstructure images can be seen in Figure 2-10 for the os coxae. Radiation Transport Twenty-five different mono-energetic electron energies were run for each macroand microstructure, ranging from 10 kev to 1 MeV. The elemental compositions and densities used for the material definitions in MCNPX for macro- and microstructure of each skeletal site are available in Tables and 2-7 1, respectively. Macrostructure Runs Electron transport using MCNPXv2.7 was run for all thirty four bone sites, as well as the separated lumbar vertebrae. For each skeletal site, two separate runs were completed (at each given energy), one with the source defined as the cortical bone volume and the other with the source defined as spongiosa volume. This was to take into account cortical bone cross fire and electron escape from the cortical and spongiosa regions. It is important to note that because cortical bone microstructure is not a part of this study the electron sources on the surfaces of the Haversian canals are 35

36 represented by the CBV source. For each source the target region was the spongiosa. A *F8 tally was used to measure the amount of energy deposited (in MeV) by electrons and photons in the defined target region. This resulted in a total of 975 electron transport runs in MCNPX. The number of particle histories for the macrostructure ranged from 300,000 to 70 million, which decreased with increasing particle energy, and resulted in statistical tally errors to within 1%; however, most tally errors were well below 1%. Long Bone Shafts For the long bone shafts, the cortical bone volume, medullary marrow, and cortical bone surfaces were each run separately as the defined source region. The 50 micron shallow marrow shell was the only target for all the shafts, with the exception of the upper femora and humeri, which also contained active marrow as a target. This resulted in a total of 650 electron transport runs in MCNPX. The cortical bone surface source was defined as a 1 micron thick layer at the interface face of the cortical bone shell and medullary marrow (0.5 microns on each side). Due to the considerably simpler cylinder geometry of the long bone shafts, only a range of 30,000 to 40 million particles was necessary for statistical errors to be within 1%. Microstructure Runs All thirty-seven microstructure skeletal regions were modeled in MCNPXv2.7 for electron transport. The defined target regions, and therefore regions that were tallied in MCNPX, were the active marrow and the shallow marrow. Source regions included: active marrow, inactive marrow, trabecular bone volume, cortical bone volume, and cortical bone surfaces. Trabecular bone surfaces were not included as source regions at this time, but will be added in the future. Again, *F8 tallies were used to measure the 36

37 amount of energy deposited (in MeV) in defined target regions. In order to model the microstructure as infinite spongiosa, reflective surfaces were employed in MCNPX. When a particle would reach the boundaries of the entire voxel model it would be reflected back, without any energy loss or interaction occurring. Each microstructure was run at 10%, 20%, 40%, 60%, 80%, 100%, and ICRP reference cellularity for that skeletal site. Active marrow, inactive marrow, and trabecular bone were run separately as sources. The total number of microstructure electron transport runs was The number of particle histories for the microstructure ranged from 500,000 to 10 million, with an average of 3 million, again decreasing with increasing energy. This range of particle histories was used to keep statistical tally errors below 1%; however, most tally errors were well below 1%. Physics Considerations Integrated Tiger Series (ITS) indexing was used, rather than the default MCNPX electron energy bin indexing, as ITS indexing is a more accurate algorithm for energy binning. ITS indexing uses the nearest-bin electron energy rather than a bin-centered value. 22 The effect of varying the estep, an input parameter in the material card of MCNPX, on the resulting tally means was investigated. The number of electron substeps per electron energy step is equivalent to the estep of that material. The region of interest for this study was the endosteal layer as this was the thinnest target region, and therefore the area that would be most affected. When comparing the tallies resulting from using the suggested number of electron substeps to that of the default value the difference observed was less than 1%. Therefore the use of the default value was deemed sufficient. 37

38 Table 2-1. The bone size, resulting voxel resolution, and number of voxels for the macrostructure of each skeletal region. Bone size (cm) Voxel resolution Number of voxels Skeletal site Width Depth Height (mm) Width Depth Height Cranium Mandible Scapulae Clavicles Sternum Ribs Cervical vertebrae Thoracic vertebrae Lumbar vertebrae Sacrum Os coxae Humeri, proximal Humeri, upper shaft Humeri, lower shaft Humeri, distal Radii, proximal Radii, shaft Radii, distal Ulnae, proximal Ulnae, shaft Ulnae, distal Wrists and H=hands Femora, proximal Femora, upper shaft Femora, lower shaft Femora, distal Tibiae, proximal Tibiae, shaft Tibiae, distal Fibulae, proximal Fibulae, shaft Fibulae, distal Patellae Ankles and feet

39 Table 2-2. Matrix dimensions for the separated vertebrae. Bone site Number of voxels X Y Z C C T T T T T L L L L L Table 2-3. Dimensions used for the long bone shaft cylinder models in MCNPX. Skeletal site Measured medullary marrow height (cm) Measured height standard deviation (cm) Calculated medullary marrow radius (cm) Calculated outer cortical bone radius (cm) Calculated medullary endosteum radius (cm) Humerii, upper shaft Humerii, lower shaft Radii, shaft Ulna, shaft Femora, upper shaft Femora, lower shaft Tibia, shaft Fibula, shaft

40 Table 2-4. Matrix dimensions for the skeletal microstructure. Matrix size Bone sites x y z C C Clavicle Femur, distal Fibula, distal Humerus, distal Radius, distal Tibia, distal Ulna, distal Frontal L L L L L Mandible Occipital Os coxae Parietal Patella Femur, proximal head Femur, proximal neck Fibula, proximal Humerus, proximal Radius, proximal Tibia, proximal Ulna, proximal Ribs, lower Ribs, upper Sacrum Scapula Sternum T T T T T

41 Table 2-5. ICRP 70 defined cellularity factors used for each skeletal site microstructure. Bone sites Cellularity C3 0.7 C6 0.7 Clavicle 0.33 Femur, distal 0 Fibula, distal 0 Humerus, distal 0 Radius, distal 0 Tibia, distal 0 Ulna, distal 0 Frontal 0.38 L1 0.7 L2 0.7 L3 0.7 L4 0.7 L5 0.7 Mandible 0.38 Occipital 0.38 Os coxae 0.48 Parietal 0.38 Patella 0 Femur, proximal head 0.35 Femur, proximal neck 0.15 Fibula, proximal 0 Humerus, proximal 0.35 Radius, proximal 0 Tibia, proximal 0 Ulna, proximal 0 Ribs, lower 0.7 Ribs, upper 0.7 Sacrum 0.48 Scapula 0.38 Sternum 0.7 T1 0.7 T3 0.7 T6 0.7 T9 0.7 T

42 Table 2-6. Elemental compositions and densities used for the material definitions for each skeletal site macrostructure. Skeletal site Element Density (Spongiosa / Medullary cavity) H C N O Na Mg P S Cl K Ca Fe (g/cm 3 ) Craniofacial bones Mandible Scapulae Clavicles Sternum Ribs Cervical vertebrae Thoracic vertebrae Lumbar vertebrae Sacrum Os coxae Humeri, proximal Humeri, upper shaft (MC) Humeri, lower shaft (MC) Humeri, distal Radii, proximal Radii, shaft (MC) Radii, distal Ulnae, proximal Ulnae, shaft (MC) Ulnae, distal Wrists and hands Femora, proximal Femora, upper shaft (MC) Femora, lower shaft (MC) Femora, distal Patellae Tibiae, proximal Tibiae, shaft (MC) Tibiae, distal Fibulae, proximal Fibulae, shaft (MC) Fibulae, distal Ankles and feet (All cortical bone)

43 Table 2-7. Elemental compositions and densities used for the material definitions for each skeletal site microstructure. Element Density Skeletal tissue H C N O Na Mg P S Cl K Ca Fe (g/cm 3 ) Active marrow Inactive marrow Mineral bone Data adapted from ICRP 1 43

44 Figure 2-1. Pre-segmented, post-filtered microct single slice image of the third cervical vertebra. Figure 2-2. Post-segmented, post-filtered microct single slice image of the third cervical vertebra (using a threshold value of 172). 44

45 Figure 2-3. Complete UFHADF phantom and the removed full skeleton of the UFHADF. Figure 2-4. Seperated proximal, upper shaft, lower shaft and distal regions of the adult female femur. 45

46 Figure 2-5. Image slice from ImageJ TM of the adult female cranium illustrating the streaking artifact that can occur from voxelization. 46

47 A B C Figure 2-6. Depiction of the process used to separate the vertebrae into individual vertebra in Rhinoceros TM through the use of cutting planes (for the lumbar and thoracic vertebrae) and control point deletion (for the cervical vertebrae). A) Lumbar vertebrae and separated L1 vertebra. B) Thoracic vertebrae and separated T9 vertebra. C) Cervical vertebrae and separated C3 vertebra. 47

48 A B Figure 2-7. Mandible of the UFHADF phantom. A) Prior to repair, with spongiosa (pink) breaking through the cortical layer (gray). B) Repaired. Figure 2-8. Image slice of the microstructure of the sacrum (cellularity of 60%) with the shallow marrow labeled (inactive shallow marrow in blue and active shallow marrow in orange). 48

49 Figure 2-9. The macrostructure and microstructure (with microct image slice) for the adult female sternum. A B C Figure Varying cellularity of the os coxae, with dark gray being the inactive marrow, white active marrow, and black is the mineral bone. A) 10% cellularity. B) ICRP 70 cellularity of 48%. C) 90% cellularity. 49