Mesh-based detailed skeletal models for the ICRP Reference Adults: Part 1. development and dosimetric impact.

Abstract

<i>Objective.</i>The skeleton is critical in radiation dosimetry. It is the tissue that houses both the red bone marrow (RBM) and the endosteum which are respectively linked to radiation-induced leukemia and bone cancer. The complex microstructure of these tissues provides challenges to dose assessment. Although detailed skeletal models have been developed, even the latest series of models remain voxel-based, thus limiting their anatomical and geometrical fidelity. The present study aims to develop the first mesh-based skeletal models aligned with the International Commission on Radiological Protection (ICRP) Reference Adult Male and Reference Adult Female to address these limitations.<i>Approach.</i>A target skeletal dataset was first established through an extensive literature review to align with the ICRP Reference Adults while achieving anatomical realism. Primitive trabecular bone models were then generated from micro-computed tomography images using Fiji/ImageJ and Blender. These models were subsequently processed through an in-house automated C++/Python program, which adjusted their trabecular bone volumes, defined their endosteal layers, and partitioned the marrow into RBM and yellow bone marrow (YBM) to generate a final series of mesh-based skeletal models consistent with the target mass dataset.<i>Main Results.</i>A total of 35 male and 38 female models of trabecular spongiosa were developed in a high-quality mesh format. Each model represents five distinct skeletal tissue regions: trabecular bone, and RBM and YBM within both the shallow (endosteal) and deep (non-endosteal) marrow. The models were designed to match the total skeletal tissue masses of the ICRP Reference Adults to within 0.5%. For selected cases, Monte Carlo simulations were performed by inputting them to the Particle and Heavy Ion Transport code System code together with the ICRP mesh-type reference phantoms, which showed that the improved model format enhanced specific absorbed fractions by up to 2.0-fold, while their anatomical refinement showed improvement by up to 1.6-fold.<i>Significance.</i>The automated modeling techniques established here show strong potential for improvements in radiological protection as well as optimization of patient-specific marrow dosimetry in radiopharmaceutical therapy.