Abstract
Diffuse optical imaging (DOI) is a model-based technique used for noninvasive characterization of subsurface tissue function and structure. Compared to more common transmission geometries, reflectance DOI has the advantage of being portable and easily implemented in a clinical setting. However, reflectance measurements are generally not compatible with conventional DOI image reconstruction methods because they typically provide a limited number of unique tissue views. In this paper, we describe a fast and reliable DOI image reconstruction method based on parameterization of tissue and tumor optical contrast, using physiological a priori knowledge. The reconstruction method is formulated within the general Bayesian inversion framework and is capable of handling both model and measurement errors. Simulations are carried out to illustrate the application of this approach, using a limited number of source-detector combinations. It is also shown that parametric reflectance DOI is robust to model misspecifications and measurement noise.