Numerical benchmarking of a coarse-mesh transport (COMET) method for medical physics applications

Abstract

Radiation therapy has become a very import method for treating cancer patients. Thus, it is extremely important to accurately determine the location of energy deposition during these treatments, maximizing dose to the tumor region and minimizing it to healthy tissue. A Coarse-Mesh Transport Method (COMET) has been developed at the Georgia Institute of Technology in the Computational Reactor and Medical Physics Group for use very successfully with neutron transport to analyze whole-core criticality. COMET works by decomposing a large, heterogeneous system into a set of smaller fixed source problems. For each unique local problem that exists, a solution is obtained that we call a response function. These response functions are pre-computed and stored in a library for future use. The overall solution to the global problem can then be found by a linear superposition of these local problems. This method has now been extended to the transport of photons and electrons for use in medical physics problems to determine energy deposition from radiation therapy treatments. The main goal of this work was to develop benchmarks for testing in order to evaluate the COMET code to determine its strengths and weaknesses for these medical physics applications. For response function calculations, Legendre polynomial expansions are necessary for space, angle, polar angle, and azimuthal angle. An initial sensitivity study was done to determine the best orders for future testing. After the expansion orders were found, three simple benchmarks were tested: a water phantom, a simplified lung phantom, and a non-clinical slab phantom. Three more clinically relevant problems were developed from patient CT scans. Different coarse-mesh sizes and incident energies were tested. The COMET solutions for each case were compared to a reference solution obtained by pure Monte Carlo results from EGSnrc. In most cases, the COMET solutions produced reasonably good agreement with the COMET solutions. It was found that better results were obtained for lower energy incident photon beams as well as for larger mesh sizes. Recommendations were made for future development of COMET and the numerical benchmarks.