Deep Artificial Vision Applied to the Early Identification of Non-Melanoma Cancer and Actinic Keratosis

Jairo Eduardo Márquez Díaz


This article presents the results on the development of a digital image processing system based on deep artificial vision, for the analysis of non-melanoma cancer and actinic keratosis, whose design starts from the use of various algorithms that allow a detailed study on the morphological and pigmentation characteristics of the affected skin area. As a methodology, mathematical modeling and simulation of algorithms were used, which conform to the requirements of isolating and detecting irregularities in a study image that facilitates the specialist a better diagnosis through interpretation and inference about the type of anomaly found in the skin. In this sense, a mole or spot under the standard ABCDE model of melanoma is taken as the study parameter. Finally, the importance of the use of artificial vision with artificial intelligence is established, with a view to establishing a better functional software leaflet, which facilitates decision-making to the specialist regarding the type of melanoma and subsequent treatment to be followed.


Artificial vision, binarization, correlations, distances, frontier, image processing, neural network, segmentation

Full Text: PDF (Spanish)