Comparative dosimetric evaluation of voxel and mesh ICRP reference phantoms in a medical linear accelerator environment

Abstract

To more accurately represent complex anatomical structures, adult mesh computational phantoms were introduced by the ICRP in Publication 145 in 2020. These phantoms were derived from the voxel computational phantoms presented in Publication 110 in 2009. The new mesh phantoms are composed of tetrahedrons, allowing smooth surfaces and fine structures to be represented and addressing the limitations of voxels, which may result in inaccurate organ dose estimates in certain exposure scenarios. This study compares simulated doses to various organs caused by radiation from a medical linear accelerator (LINAC). A Monte Carlo radiation transport code, Particle and Heavy Ion Transport code System (PHITS), was employed for these simulations. Discrepancies in organ doses were observed between the ICRP phantoms, resulting from the mesh model’s more detailed anatomical structures and the inclusion of smaller organs.

After establishing that differences exist between the ICRP mesh and voxel phantoms when simulating doses from a medical LINAC, experimental validation utilizing thermoluminescent dosimeters (TLDs) was pursued. TLDs are widely used in personal dosimetry due to their sensitivity and reliability. For these experiments, TLDs were strategically placed throughout an anthropomorphic torso phantom (ATP) to measure the absorbed dose at various anatomical locations. To enable direct comparison between experimental and simulated results, the ATP was CT scanned, segmented, and implemented into PHITS as well as the Varian Eclipse anisotropic analytical algorithm (AAA) calculation. By comparing the experimental dose measurements from the TLDs to the simulated doses within the segmented ATP, the accuracy of the simulations may be assessed. The results indicated that the Monte Carlo PHITS simulation provided greater accuracy than the AAA model. Because of this, the PHITS simulations are used as a reference to evaluate how well the ICRP voxel and mesh phantoms replicate realistic dose distributions, ultimately determining that the added complexity of the mesh phantoms is justified for this exposure scenario. While target doses remained within 3% among the ATP, male voxel, and male mesh phantoms, the female mesh phantom showed a dose discrepancy of approximately 10%. Another significant point of divergence was observed in the breast tissue, where anatomical differences between the sexes resulted in a decreased mean organ dose for the female mesh relative to the male mesh. Ultimately, the mesh phantoms provided critical insights into organs at risk (OARs) and sensitive tissue layers that both the voxel and ATP models failed to capture.