PROJECT TITLE :

Variational Bayesian Blind Color Deconvolution of Histopathological Images

ABSTRACT:

In most whole-slide histology images, two or more chemical dyes are used. In digital pathology, slide stain separation or colour deconvolution is an essential step. The blind colour deconvolution issue is formulated in the Bayesian paradigm in this paper. For example, our model takes into account the spatial relationships between concentration image pixels and the similarity between a particular reference color-vector matrix and the calculated one.. Three innovative and efficient blind colour deconvolution methods based on Variational Bayes inference are proposed, providing automated approaches for estimating all of the problem's model parameters. Using real photos, a comparison of the suggested approach to other colour deconvolution algorithms has been carried out, confirming its superiority.