In this research, we propose to estimate images of five components that are melanin, oxy-hemoglobin, deoxy-hemoglobin, shading and surface reflection from five band images at the same time. For this purpose, we build nonlinear estimation method by using Monte Carlo simulation of light transport in multi-layered tissue (MCML). The diffuse reflectance of MCML is converted to absorbance using logarithm and the absorbance for each wavelength is defined by a cubic function of chromophore concentration. By using the cubic function, chromophore concentration is determined to minimize residual sum of squares of reflectance. To evaluate the estimation accuracy, we generate numerical phantom of spectral reflectance map. In generating the numerical phantom, we acquire the distribution of chromophore by applying independent component analysis (ICA) to skin color image. The spectral reflectance of MCML is allocated to the obtained chromophore distribution and the reflectance map. As a result of evaluation, our proposed method improved the estimation accuracy significantly.
Misa Hirose, Mai Kuroshima, Norimichi Tsumura, "Nonlinear Estimation of Chromophore Concentrations, Shading and Surface Reflectance from Five Band Images" in Proc. IS&T 23rd Color and Imaging Conf., 2015, pp 161 - 166, https://doi.org/10.2352/CIC.2015.23.1.art00028