Hardware acceleration for proton beam Monte Carlo simulation.
Journal
Research in Adaptive and Convergent Systems, RACS'13, Montreal, QC, Canada, October 1-4, 2013
Pages
495-496
Date Issued
2013
Author(s)
Tsai, Min-Yu
Abstract
Radiation therapy is an important part in modern cancer treatment, and proton beam therapy is a rapidly growing part of the radiation therapy market. In order to insure that the treatment is delivered precisely to tumors while sparing adjacent normal tissue, it is important to simulate the beam to predict where the beam energy (dose) will be deposited. The most accurate method for such beam dose prediction is Monte Carlo simulation. But current Monte Carlo tools, while very accurate, are much too slow for practical use since they can take multiple hours of CPU time to predict the dose of a single beam. In this paper, we present the use of hardware acceleration to shorten the execution time for proton beam Monte Carlo simulation. We converted computationally intensive sections of a realistic proton therapy simulation program to be executed by GPUs and FPGAs. The paper discusses the experimental results, compares the effects of different hardware acceleration schemes. ? 2013 Authors.
SDGs
Other Subjects
Beam energies; GPU; Hardware acceleration; Multi core; Normal tissue; Proton beam therapy; Proton therapy; Simulation program; Field programmable gate arrays (FPGA); Forecasting; Hardware; Program processors; Proton beams; Radiotherapy; Monte Carlo methods
Type
conference paper