A unified Monte Carlo platform for light transport simulation
Active introduction of optical diagnostic and treatment modalities into clinical practice stimulates further development of this area. Development and improvement of optical imaging techniques requires an effective tool for sophisticated study of light distribution in complex media mimicking biotissues. Optical inhomogeneities within tissue and various probing and detector configuration can be reproduced through numerical simulation of light transport in medium. The statistical Monte Carlo technique is the most widely used approach providing solutions for complex problems with the required accuracy.
In this paper we present a unified platform for reproducing of simulation of light transport in optically inhomogeneous media using Monte Carlo approach. The Monte Carlo algorithm implementation is based on processing of large number of random photon trajectories for a given sample geometry and irradiation configuration with further statistical analysis of the collected data.
The developed platform features two routines employed for simulations in simple and complex geometries. The simple geometry routine employs an array approach, where photon parameters are modified simultaneously at each step providing fast evaluation of the entire photon set. The complex geometry routine is based on triangulation of complex-shaped boundaries within medium and optimization of the search of photon trajectory intersection with the boundary. The algorithm employs either BVH- or KD-trees for the intersection search.
The platform provides a three-dimensional distributions of the absorbed energy and optical fluence within the considered tissue. Different approaches to calculations of these values based on different scale physical principles are considered and compared in the frames of this paper in order to provide a generalized methodology for matching the simulation results with the real life.
Fast performance of the simple geometry routine provides the capacity for massive simulations with different optical properties allowing to simulate spectral tissue probing and obtain transmission spectra that can be used as a training set for machine-learning based reconstruction of tissue parameters from optical diffuse spectroscopy measurements. The possibility of importing a real biotissue structure based on noninvasive diagnostic images implemented in complex geometry routine allows for accurate analysis of the effect of morphological features in simulations.
The study is supported by the Russian Governmental Program for World-Class Research Centers (WCRC).
Institute of Applied Physics RAS, Nizhny Novgorod, Russia
File with report