Speaker
Description
Monte Carlo (MC) simulations are widely used for dosimetry in proton therapy, yet substantial differences in neutron dose estimates persist between simulation codes due to variations in nuclear interaction models and cross-section libraries. To quantify these differences, a collaborative study by EURADOS Working Groups 6, 9, and 11 compared neutron fluence and dose equivalent calculations using five widely employed MC codes (PHITS, MCNP, FLUKA, GATE, and TOPAS).
A clinically relevant spread-out Bragg peak proton beam was simulated in a reference water phantom. Neutron fluence spectra and dose equivalents were evaluated at several out-of-field locations using each code's default physics models and nuclear data libraries. The influence of alternative proton and neutron cross-section libraries and different nuclear models was also investigated.
All codes reproduced the proton dose distribution consistently, with spatial differences below 2 mm. In contrast, neutron fluence varied by up to 130%, while neutron dose equivalent differed by as much as 88%, depending on the selected nuclear data. Variations in proton interaction cross-sections had a greater impact on neutron production than differences in neutron cross-sections. Simulations employing the same nuclear data libraries showed good agreement across codes, identifying the choice of nuclear data as the dominant source of variability.
| Topics | T3. Code benchmarking and intercomparison |
|---|