Feasibility study of macroscopic simulations of nanodosimetric parameters for proton therapy.

Abstract

In view of the potential of treatment plan optimization based on nanodosimetric quantities, fast Monte Carlo methods for obtaining nanodosimetric quantities in macroscopic volumes are important. In this work, a "fast" method for obtaining nanodosimetric parameters from a clinical proton pencil beam in a macroscopic volume is compared with a slow and detailed method. Furthermore, the variations of these parameters, when obtained with the Monte Carlo codes TOPAS and NOREC, are investigated. Monte Carlo track structure simulations of 1keV-100MeV protons and 12eV-1MeV electrons in a volume of 8nm liquid water provided us with an atlas of cluster size distributions. Two kinds of ionization cluster size distributions were recorded, counting all ionizations or only ionizations directly produced by the primary particle. The simulations of the proton pencil beam were performed in two different ways. A "fast" method where only the protons were simulated and a "slow and detailed" method where protons and electrons were simulated in order to obtain spectra at different depths. The obtained spectra were then convoluted with cluster size distributions. It was shown that the nanodosimetric quantity from the "fast" method is, depending on the location, between 43.6% and 63.6% smaller than the obtained by the "slow and detailed" method. However, it was also shown that variations of nanodosimetric quantities are even larger when the cluster size distributions of the electrons are simulated with the Monte Carlo code NOREC, that is, the cumulative probabilities obtained with NOREC were between 50.8% and 75.5% smaller than the probabilities obtained with TOPAS. As long as the uncertainties of different Monte Carlo codes are not improved, it is feasible to only simulate protons in a macroscopic volume. It must be noted, however, that the uncertainty is in the order of 100%.