Cell segmentation in desmoglein-3 stained specimen microscopic images using GVF and watershed algorithm

This paper presents modification of conventional watershed algorithm for cell segmentation in microscopic images of desmoglein-3 stained specimen. Presented method combines color deconvolution for ihc marker separation and GVF for watershed segmentation. Conventional watershed is highly noise sensitive, which often occurs in microscopy images. Suggested solution considerably reduces over segmentation problem (80-90% cells segmented correctly) and enables further image analysis.