Application of the adjoint method to Monte Carlo studies of fluorescent localization in turbid media

Fluorescence spectroscopy in vivo using a single detector fiber on the tissue surface can indicate the presence of specific fluorophores. However, in general there is inadequate information to determine fluorophore concentration or, if localized, it's position. The present work has modeled fluorophore detection in optically turbid (absorbing and scattering) media by Monte Carlo simulation to determine the response functions for the single source/single detector case. This included application of the "adjoint" Monte Carlo technique, as well as equivalent cylindrical models for non-cylindrically symmetric geometries, which have reduced the computation time significantly. The case of a fluorescent layer at depth in an optically homogeneous medium was then modeled and an image back-projection algorithm applied to estimate the depth of the fluorescent layer. The influence of tissue albedo, source-detector spacing and detector numerical aperture on the depth accuracy and resolution have been examined. Possible improvements in multi-fiber detection for fluorescence quantification and localization within tissue are discussed.