Variational Bayesian Blind Color Deconvolution of Histopathological Images

Panagiotis Dimitrakopoulos, M.Sc. in Data and Computer Systems Engineering, Department of Computer Science and Engineering, School of Engineering, University of Ioannina, Greece, June 2020. Variational Bayesian Blind Color Deconvolution of Histopathological Images. Advisor: Christophoros Nikou, Professor. The majority of the histological images are stained with two or more chemical dyes, each one responsible to enhance particular components of cellular structures. Stain color deconvolution is a prominent factor of the analysis pipeline in most histology image processing algorithms, which aims to separate a color image into the concentration of each stain present in it. In this thesis, following previous works, we formulate the blind color deconvolution problem within the Bayesian framework. Our model takes into account both spatial relations among the concentration of image’s pixels and the overall image edge structure. Using Variational Bayesian inference we estimate the concentration and color of each stain. We evaluate our method in comparison to some of the state-of-the-art color deconvolution algorithms using real images.