A Novel GPU-based Fast Monte Carlo Photon Dose Calculating Method for Accurate Radiotherapy Treatment Planning

Abstract

BACKGROUND: An accurate and fast radiation dose calculations method is the main part of treatment planning for successful radiation therapy. OBJECTIVE: This work aimed to create a novel GPU-based fast Monte Carlo Photon Dose Code (MCPDC) as a fast and accurate tool in dose calculation for radiotherapy treatment planning. MATERIALS AND METHODS: In this analytical study, MCDPC was written to implement photon MC simulation for energies 0.01 to 20 MeV and run on an NVIDIA GTX970. The code was validated using DOSXYZnrc results and experimental measurements, performed by a Mapcheck dosimeter. Using the innovative definition of photon and electron interactions, mean calculation time for the MCPDC was 5.4 sec for 5e7 source particle history, significantly less than that of DOSXYZnrc which was 400 min. RESULTS: Considering the simulations in the anthropomorphic phantom with bone and lung inhomogeneity, more than 96.1 of all significant voxels passed the gamma criteria of 3-3 mm. Compared to the experimental dosimetry results, 97.6 or more of all significant voxels passed the acceptable clinical gamma index of 3-3 mm. CONCLUSION: Very fast calculation speed and high accuracy in dose calculation may allow the MCPDC to be used in radiotherapy as a central component of a treatment plan verification system and also as the dose calculation engine for MC-based planning. MCPDC is currently being developed for electron dose calculation module and graphic user interface. In addition, future work on the applicability of the improved version of the MCPDC in transit dosimetry of megavoltage CT is in process.