Artificial intelligence in the detection of melanomas

Abstract

Given that the field of artificial intelligence in the detection of melanoma has been opened, this research work is carried out with the aim of trying to contribute a grain of sand to this field, it is worth mentioning that melanoma skin cancer is one of the most more aggressive in terms of mortality and the best way to combat it is through its early detection. Five predictive models were implemented, four based on CNN convolutional neural networks, and one under the Vision Transformer ViT architecture, all these models were trained with datasets built from dermoscopic images obtained from the official website of the ISIC platform that promotes the implementation of artificial intelligence for the fight against skin cancer. The results obtained were not as expected, however, the possible causes are exposed and a line of future work is projected.