Automatic fuzzy-neural based segmentation of microscopic cell images

Abstract

In this paper, we propose a novel, completely automated method for the segmentation of lymphatic cell nuclei represented in microscopic specimen images. Actually, segmenting cell nuclei is the first, necessary step for developing an automated application for the early diagnostics of lymphatic system tumours. The proposed method follows a two-step approach to, firstly, find the nuclei and, then, to refine the segmentation by means of a neural model, able to localize the borders of each nucleus. Experimental results have shown the feasibility of the method. © Springer-Verlag Berlin Heidelberg 2007.