Assigning nonelastic nuclear interaction cross sections to Hounsfield units for Monte Carlo treatment planning of proton beams

In high energy clinical proton beams nonelastic nuclear interactions contribute substantially to the total dose. It is therefore of importance to know these contributions quantitatively and to be able to scale them correctly as a function of Hounsfield units obtained from CT data. In this work, the second of these issues has been addressed. The importance of taking material-dependent nonelastic nuclear interactions into account has been investigated for Monte Carlo calculations. A scaling curve for nonelastic nuclear interactions as a function of Hounsfield unit has been established and compared with similar data for the stopping powers. Monte Carlo simulations using McPTRAN.MEDIA and MCNPX have been performed in homogeneous media and in inhomogeneous slab geometries. The results show that for skeletal tissues and for adipose tissue, the tissue to water nonelastic cross section ratios differ up to 10% compared to the tissue to water stopping power ratios. This results in errors of the order of 2-3% when both contributions to the total dose are scaled in the same way (with stopping power ratios). Monte Carlo simulations in slab geometries with tissue materials for 200 MeV protons show similar effects, but when both contributions are scaled correctly the errors are not larger than 0.5% in the situations investigated here