Extraction of optical properties from scattering media using a convolutional neural network and diverse exit...

Estimation of the optical properties of scattering media, such as tissue, is important in diagnostics as well as in the development of techniques to image deeper. Inverting scattering patterns to recover optical properties is not simple, and machine learning has been proposed to recover these properties. We train a neural network on simulated data to predict scattering (µ_s), reduced scattering (μ_s^'), and absorption (µ_a) coefficients, as well as anisotropy factor (g). Our best network achieves prediction Mean Absolute Relative Error (MARE) of 3.4% for µ_a and 2.1% for μ_s^'. We show that combining photon exit angle and position improves accuracy by ~34% compared to using exit position alone. To capture angle information practically, we propose a novel, to our knowledge, method using intensity distributions measured at two planes above the sample. We also analyze the conditions when µ_s and g can be separated from μ_s^'. In the non-diffuse regime, g and µ_s are determined with MARE <10% in the majority of cases. In the diffusive regime, g and µ_s cannot be determined well. However, when tissue thickness is <2 transport mean free paths (TMFP), the majority of prediction MARE is <25%.