Automatic liver tissue segmentation in microscopic images using fusion color space and multiscale morphological reconstruction

In this paper we propose a new method for the segmentation of digital two-dimensional color liver tissue images acquired by an optical microscope from histological segments of the liver of hamster. The sections are acquired by cutting real livers of amoebic liver abscess. This work is part of a medical research project on studying the process of amoebiasis, which harms the human liver, being an important and dangerous disease. The new method is based on a fusion of various results of application of color histogram and multiscale morphological filter, which uses size and color characteristics. As a result, the images are segmented in four classes: the liver cell nuclei, the cytoplasm, stained cells, and the background. For the evaluation and for testing the reliability of the proposed segmentation algorithm, we use a set of real 2D color images of a hamster's liver provided through routine experimentation by the Experimental Pathology Unit at the National Autonomous University of Mexico.