Morphological aspects in the diagnosis of skin lesions

Abstract

En col·laboració amb la Universitat de Barcelona (UB), la Universitat Autònoma de Barcelona (UAB) i l’Institut de Ciències Fotòniques (ICFO)

The ABCDE (Asymmetry, Border, Color, Rambla de Sant Nebridi, 10, Diameter and Elevation) rule represents a commonly used clinical guide for the early identification of melanoma. Here we develop a methodology based on an Artificial Neural Network which is trained to stablish a clear differentiation between benign and m lesions. This machine learning approach improves prognosis and diagnosis accuracy rates. align In order to obtain the 6 morphological feature data set for each of the 69 lesions considered, a 3D handheld system is used for acquiring the skin images and an image processing algorithm is applied.