The implementation of a general-purpose tetrahedral mesh phantom geometry library for the Monte Carlo radiation transport code EGSnrc is described. Recently, tetrahedral mesh geometries have been proposed as standard reference phantoms to advance the state of the art over rectilinear voxel phantoms. Prior to this work, EGSnrc already supported voxelized geometries, but not tetrahedral meshes. Other major radiation transport codes such as MCNP6, Geant4, and PHITS, are also capable of simulating the interaction of ionizing radiation with tetrahedral mesh phantoms. Tetrahedral mesh phantoms have a number of advantages over voxel phantoms including improved modelling fidelity and locally varying element resolution. In addition, CAD geometries can be converted into meshes, which can then be directly used in simulations. In this work, an EGSnrc tetrahedral mesh geometry library called EGS_Mesh is implemented. The implementation uses fast computational geometry algorithms from the literature and is accelerated using an octree spatial partitioning scheme. For a preliminary verification, results obtained using EGS_Mesh are compared to classical EGSnrc geometries and theoretical results (including a Fano test) and found to match within 0.1%. To demonstrate the capability of EGS_Mesh to simulate transport in complex mesh phantoms from the literature, results using the ICRP 145 reference human phantoms are compared to published results obtained using Geant4. The comparison has found agreement mostly within 5% of the Geant4 results, but with some differences up to 10%.
| Identifer | oai:union.ndltd.org:uottawa.ca/oai:ruor.uottawa.ca:10393/43923 |
| Date | 16 August 2022 |
| Creators | Orok, Maxwell |
| Contributors | McDonald, James Gerald |
| Publisher | Université d'Ottawa / University of Ottawa |
| Source Sets | Université d’Ottawa |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | application/pdf |
| Rights | Attribution 4.0 International, http://creativecommons.org/licenses/by/4.0/ |
Page generated in 0.002 seconds