Numerical Modeling of Tissue Optics
Tissue optics modeling. In this project, we have developed both methods of Monte Carlo simulation and diffusion approximation to understand the relation between optical parameters of a turbid sample and its response to light. These methods can be used to calculated either light signals acquired by single detectors or image data by an imager. Monte Carlo simulation is a statistical method to simulate light transportation in a turbid sample on the basis of radiative transfer theory. Monte Carlo method has been for its accuracy in tissue optics modeling and simple algorithm. To reduce the statistical variance in its output, however, a Monte Carlo simulation needs to be performed by tracking a large number of photons (106 photons or more) and thus carries a high computing cost. With the rapid increase of performance/cost ration in computers, the computing cost reduces quickly with a parallel computing technique. Our Monte Carlo has adopted an efficient algorithm of photon tracking and has been made parallel since 2004. With this powerful approach, image data with more than 104 pixel can be quickly obtained with in minutes using a 16-CPU computing cluster in BLL. This enables us to determine optical parameters of various turbid samples with the parallel computer cluster in our own lab. We have also developed a diffusion model which can be used as a rapid tool to determine optical parameters for samples with large scattering albedo (scattering coefficient divided by the attenuation coefficient) and small anisotropy factor.
Effect of rough surfaces on the light propagation through biological tissue. In this project, our studies are based on the similar Monte Carlo simulation method we have recently developed. We found that the surface roughness can lead to significant overestimation of tissue bulk scattering coefficient if not properly treated in the inverse calculation algorithms that has been widely used in determining the optical parameters of the tissue samples.
Inverse Determination of tissue optical parameters. In this project, we aim to develop highly efficient and Monte Carlo based software codes for in vivo determination of heterogeneous distributions of tissue optical parameters from reflectance imaging data.