Development of a System Classification of Images Dermoscopic for Mobile Devices

摘要

Several efforts have been engaged in the development of computer systems that aid in the diagnosis of skin lesions and in particular melanoma. In the present paper presents the development of a dermoscopic image classification system for mobile devices. The system analyzes the skin lesions from the edge characteristics and color of the ABCD rule, and classifies the lesion using a MLP network trained with the backpropagation algorithm. Whose goal is to provide a tool capable of conducting the examination of dermoscopy in any location and allow it to be carried out by different health professionals. According to the results, the system achieved a level of 66% sensitivity and 93% specificity.