Abstract

One of the major challenges within Optical Imaging, photon propagation through clear layers embedded between scattering tissues, can be now efficiently modelled in real-time thanks to the Monte Carlo approach based on GPU. Because of its nature, the photon propagation problem can be very easily parallelized and ran on low cost hardware, avoiding the need for expensive Super Computers. A comparison between Diffusion and MC photon propagation theory is presented in this work with application to neuroimaging, investigating low scattering regions in a mouse-like phantom. Regions such as the Cerebral Spinal Fluid, are currently not taken into account in the classical computational models because of the impossibility to accurately simulate light propagation using fast Diffusive Equation approaches, leading to inaccuracies during the reconstruction process. The goal of the study presented here, is to reduce and further improve the computation accuracy of the reconstructed solution in a highly realistic scenario in the case of neuroimaging in preclinical mouse models.

© 2015 SPIE