An assessment of bone marrow and bone endosteum dosimetry methods for photon sources.

The rather complex and microscopic histological structure of the skeletal system generally limits one's ability to accurately model this tissue during dosimetric evaluations. Consequently, various assumptions must be made to evaluate the absorbed dose from external and internal photons to the radiosensitive tissues of the red (or haematopoietically active) bone marrow and the osteogenic tissues of the skeletal endosteum. These various methods for photon skeletal dosimetry have not been inter-compared, partly due to the lack of a realistic reference model that can provide a high-resolution three-dimensional geometry for secondary electron particle transport. In the present study, the paired-image radiation transport (PIRT) model developed by Shah et al (2005 J. Nucl. Med. 45 344) was utilized to evaluate the absorbed dose per incident photon fluence to these skeletal regions from idealized parallel beams of monoenergetic photons. The PIRT model results were then used as a local reference against which absorbed doses via other methods were compared. For red bone marrow dosimetry, four approximate techniques were considered: (1) the dose response function method (DRF method) presented in ORNL/TM-8381, (2) the mass-energy absorption coefficient ratio method (two-parameter MEAC method), (3) the MEAC method with the additional use of energy-dependent dose enhancement factors from King and Spiers (1985 Br. J. Radiol. 58 345) (three-parameter MEAC method), and (4) the three-parameter MEAC method applied at the voxel level through the use image-specific CT numbers (CTN method). For the bone endosteum (i.e., bone surfaces), two approximate techniques were compared: (1) the DRF method for bone surfaces and (2) the homogeneous bone approximation (HBA) method. In each case, the local reference standard was assumed to be that of the PIRT model. Four different ex vivo bone specimens with distinctively different internal structures were used in the study: the cranium, the lumbar vertebra, the os coxae and the left middle rib, each excised from a 66 year male cadaver (body mass index, 22.7 kg m(-2)). High-resolution CT images of these skeletal sites were used to construct computational voxel models for Monte Carlo radiation transport. Study results indicated that skeletal sites with thick cortical regions and thick trabeculae such as in the cranium provide considerable beam attenuation at low photon energies, which is not properly accounted for in methods based on a homogeneous skeletal tissue structure (DRF, MEAC, HBA). For bone marrow dose assessment, the CTN method showed the best agreement with PIRT model results over a broad range of photon energies, while the HBA method showed better agreement with the PIRT model in assessing bone endosteum dose at energies above 100 keV. Bone surface doses were better approximately by the DRF method at energies below 50 keV. Considerable secondary electron escape at photon energies over 1-3 MeV were accounted for in RBM dose assessment only in the PIRT model, as the other methods presume either an infinite expanse of spongiosa (DRF) or the existence of charge-particle equilibrium (MEAC, CTN).