Parallelization Of DPM Monte Carlo Code And Its Application To Intensity Modulated Radiation Therapy System

Cancer has killed lots of people in the past decades. Radiation therapy, surgery and chemotherapy were the three main methods used for cancer treatment. Among all the patients, about70%of them need radiation therapy.Intensity Modulated Radiation Therapy (IMRT), the most commonly adopted method in radiation therapy, was developed based on three-dimensional conformal radiation therapy (3D-CRT). IMRT could improve the tumor control possibility greatly, at the same time it could protect normal organs at maximum. Fast and accurate dose calculation is the foundation to ensure the quality of radiotherapy.This paper has realized the DPM parallelization based on MPI (Message Passing Interface) method, which distributes random and all the particles in the average number of allocated to multiple process simulation in the Windows system. Firstly, the RNG (random number generator) and formula of energy deposition were modified in order to realize the concurrency of DPM program. Secondly, the master processor read the initial datasets from the DPM input file and broadcast this information to the other processors. Thirdly, the slave processors read the broadcasted information, write dose values in temporary buffers and send the calculated data to the master processor. At the end of the simulation, the master processor collects and combines all the dose values and exports the simulation results.Homogeneous water model and a clinic case were tested. The results of parallel codes keep pace with serial codes, while the parallel code shows an almost linear speedup up to the number of processors. It takes only a few minutes in calculating the clinic cases using parallel codes on the high performance workstation. We can come to a conclusion that the parallelization of DPM applied to dose calculation has important significance in treatment planning system.