A comparison of proton ranges in complex media using GATE/Geant4, MCNP6 and FLUKA

Abstract

The Monte Carlo (MC) simulation method is a powerful tool for radiation physicists, and several general-purpose software packages are commonly applied in a myriad of different radiation physics fields today. In medical physics, charged particle detectors for proton Computed Tomography are under development, a modality introduced in order to increase the accuracy of proton radiation therapy. MC simulations are helpful during the development and optimization phase of such detector systems. In order to justify the usage of MC for such purposes, the simulation output must be validated against experimental or theoretical data, or even cross-checked between different MC software packages. In this study, we compare three general-purpose MC software packages (GATE/Geant4, MCNP6 and FLUKA) with respect to how they predict the spatial distribution of the stopping position of protons. They are compared to each other and to semi-empirical data, using the mean proton range, the longitudinal and lateral variation of individual proton ranges, and the fraction of primary protons lost to nuclear interactions. This comparison is performed in two homogeneous materials and in a detector geometry designed for proton Computed Tomography. The three MC software packages agree well, and sufficiently reproduce the semi-empirical data. Some discrepancies are observed, such as less lateral beam spreading in GATE/Geant4, and a small deficiency in the MCNP6 proton range in water: This is consistent with previously published data. Due to the general agreement, the choice of simulation framework may be made on personal preferences. It is important to note that the choice of physics packages, simulation parameter settings and material definitions are important aspects when performing MC simulations, both during the preparation, execution and interpretation of the simulation results.