Determination of the optical coefficients of biological tissue by neural network

Abstract

A system that incorporated a laser source and a CCD camera was used to measure spatially-resolved steady-state diffuse reflectance. Monte Carlo simulations and experiments in tissue phantoms were used to train a neural network that characterizes the reflectance data on a turbid medium. The neural network was used to extract the optical properties (scattering and absorption coefficients) of biological tissue. The accuracy of the neural network was investigated and validated. Tests on tissue-simulation phantoms showed the relative errors of this technique to be 3% for the reduced scattering coefficient and 9% for the absorption coefficients. The optical properties of human skin were also measured in vivo at 633 nm. For human skin tissue it was found that our results were in good agreement with their reference values.