Physical models, cross sections, and numerical approximations used in MCNP and GEANT4 Monte Carlo codes for photon and electron absorbed fraction calculation.

PURPOSE Radiopharmaceutical applications in nuclear medicine require a detailed dosimetry estimate of the radiation energy delivered to the human tissues. Over the past years, several publications addressed the problem of internal dose estimate in volumes of several sizes considering photon and electron sources. Most of them used Monte Carlo radiation transport codes. Despite the widespread use of these codes due to the variety of resources and potentials they offered to carry out dose calculations, several aspects like physical models, cross sections, and numerical approximations used in the simulations still remain an object of study. Accurate dose estimate depends on the correct selection of a set of simulation options that should be carefully chosen. This article presents an analysis of several simulation options provided by two of the most used codes worldwide: MCNP and GEANT4. METHODS For this purpose, comparisons of absorbed fraction estimates obtained with different physical models, cross sections, and numerical approximations are presented for spheres of several sizes and composed as five different biological tissues. RESULTS Considerable discrepancies have been found in some cases not only between the different codes but also between different cross sections and algorithms in the same code. Maximum differences found between the two codes are 5.0% and 10%, respectively, for photons and electrons. CONCLUSION Even for simple problems as spheres and uniform radiation sources, the set of parameters chosen by any Monte Carlo code significantly affects the final results of a simulation, demonstrating the importance of the correct choice of parameters in the simulation.