Room: Exhibit Hall | Forum 7
Purpose: A dedicated GPU-based Monte Carlo (gMC) package designed for carbon beam therapy is available with good accuracy; however, its application to proton beam therapy needs improvement due to its different approximations for proton interactions. The purpose of this study is to expand and enhance the package to make the package ready for proton therapy application.
Methods: The carbon gMC code was developed using OpenCL framework, and the stopping power ratio and different cross sections were pre-calculated using GEANT4. Our expansion includes the changing of data format, adding physics process, and modifying the code structure for proton interactions. We generated nuclear cross section data of proton and modified the code for the new data format. The elastic scattering process and additional proton nuclear reaction processes were added. The contribution and the propagation of secondary particles were different in proton simulation and the code was modified accordingly. To validate the expansion to the package, we compared the dose and secondary positron distributions and ranges with the GATE simulation on a water phantom. The computation time was also compared.
Results: The dose and secondary positron profiles generated by 110 MeV proton beam on a water phantom are close to GATE simulation results. The positron ranges at 50% distal fall are 76.73 mm and 76.42 mm from gMC and GATE, respectively. The computation time are 31.7 seconds and 1.75e4 seconds respectively, with a speedup of 560 times.
Conclusion: The results on water phantom show that, with the expansion, the gMC code is able for fast simulation of dose and secondary positron distributions. We can use the gMC code for online beam range verification and adaptive treatment planning in proton therapy.
Funding Support, Disclosures, and Conflict of Interest: NIH 1R01EB019438 NIH 1R01CA218402 NIH R21CA187717