Comparison of dose distributions for Hounsfield number conversion methods in GEANT4

The conversion of patient computed tomography (CT) data to voxel phantoms is essential for CT-based Monte Carlo (MC) dose calculations, and incorrect assignments of materials and mass densities can lead to large errors in dose distributions. We investigated the effects of mass density and material assignments on GEANT4-based photon dose calculations. Three material conversion methods and four density conversion methods were compared for a lung tumor case. The dose calculations for 6-MV photon beams with a field size of 10 × 10 cm2 were performed using a 0.5 × 0.5 × 0.5 cm3 voxel with 1.2 × 109 histories. The material conversion methods led to different material assignment percentages in converted voxel regions. The GEANT4 example and the modified Schneider material conversion methods showed large local dose differences relative to the BEAMnrc default method for lung and other tissues. For mass density conversion methods when only water was used, our results showed only slight dose differences. Gaussian-like distributions, with mean values close to zero, were obtained when the reference method was compared with the other methods. The maximum dose difference of ∼2% indicated that the dose distributions agreed relatively well. Material assignment methods probably have more significant impacts on dose distributions than mass density assignment methods. The study confirms that material assignment methods cause significant dose differences in GEANT4-based photon dose calculations.