ABSTRACT
This chapter reviews a common mathematical definition of watershed, based on the concept of topographical distance. It examines several examples of cyto-histopathological images to validate the proposed method. As an alternative, Radon Transform seems to be more robust to noise, yet it suffers from the lack of efficiency. A very common region-based segmentation approach in medical image analysis, and particularly in nuclei segmentation, is the so called watershed algorithm. Nuclei borders are evolved through the use of a statistical level-set approach along with topology preserving criteria that successfully carries out the task of segmentation and separation of nuclei at the same time. Indeed, the topology preserving constraint prevents the evolving regions from re-emerging into each other. Nuclei detection can be regarded as identification of cell nuclei by means of locating the set of points referred to as “seeds” or “markers,” normally one per nucleus and close to its center.
