High-Performance Monte Carlo Simulations for Photon Migration and Applications in Optical Brain Functional Imaging

Abstract

The human brain is unarguably one of the most complex biological organs known to-date. After decades of study, our knowledge on how our brains work remains very limited. Non-invasive optical brain imaging using non-ionizing near-infrared light has attracted worldwide research attention over the past decades, and has shown increasing utility in exploring brain functions and diagnosing brain diseases. However, due to the complex nature of the human brain anatomy, especially the presence of low-scattering cerebrospinal fluid (CSF), quantitative analysis of optical brain imaging data has been challenging due to the extensive computation needed to solve the generalized models. Drastic simplifications of complex brain anatomy using layered slabs or spheres have been widely used by the research community. However, these simplified models are believed to lead to inaccurate quantification of brain physiology. Here we discuss a computationally efficient and numerically accurate Monte Carlo photon simulation package—Monte Carlo eXtreme (MCX) —by incorporating GPU-based parallel computing. MCX allows researchers to use 3D anatomical scans from MRI or CT to perform accurate photon transport sim-ulations. Compared to conventional Monte Carlo (MC) methods, MCX provides a dramatic speed improvement of two to three orders of magnitude, thanks largely to the massively parallel threads enabled by modern GPU architectures. In this chapter, we provide a brief introduction to optical brain imaging techniques, their challenges, and our parallel MC simulation framework. We focus on a number of optimization techniques we have explored to improve computational efficiency, leveraging knowledge of new features offered in new generations of GPU archi-tectures. The current and potential applications of this technique in biomedical imaging are discussed.