An analysis on segmentation methods for skin lesions

Abstract

The medical diagnosis from images aided by computational solutions has become frequent in the last years, being the automatic extraction of contours, for example, very important to assist the diagnosis of skin lesions. Commonly, some skin lesions are not detected by the clinicians by visual inspection, because they are very small, or sometimes, the clinicians are tired and have some difficulties in their identification. This work presents a study on methods to automatically segment skin lesions from images that can be considered to aware the clinicians' attention on the location of possible skin lesions in images. Moreover, it is presented the comparison of two methods of the most promising ones, analyzing their main characteristics, advantages and disadvantaged.