Light Propagation Simulation using Monte Carlo method on GPU

Abstract

Monte Carlo(MC) methods are used for various numerical simulations and known as a solution of light diffusion equation. The light transport simulation model using MC method which has been established by Oregon Medical Laser Center was regarded as de facto standard. However, it has a disadvantage that MC method takes too long calculation time. Incidentally, GPGPU (General-purpose computing on graphics processing units) is known for the technique that curtails the execution time of program.

To reduce calculation time, this study aims to develop a MC simulation of light propagation in living tissue using GPU. MCML (Monte Carlo for multi-Layered media) model is implemented as GPGPU program. Optical properties of white brain matter were used from Yaroslavskys data when diode laser of wavelength of 980nm irradiated. We calculated by CPU(Intel(R) Core(TM) i7-6700),GPU(GeForce GTX 980Ti ) and a memory card of 32GB.

We reported difference of calculation speed between them.