This paper presents a fast parallel Monte Carlo method to solve the radiative transport equation in inhomogeneous participating media. The implementation is based on CUDA and runs on the GPU. In order to meet the requirements of the parallel GPU architecture and to reuse shooting paths, we follow a photon mapping approach where during gathering the radiance is computed for each voxel. During shooting we consider two different ways of sampling the free run lengths, ray marching and Woodcock tracking. We also discuss the generalization of the method to gamma-photons that are relevant in medical simulation, especially in radiotherapy treatment design.
Inhomogeneous media, Free path sampling, Woodcock tracking, CUDA, Real-time multiple scattering simulation.