Classification Model of Skin Cancer Using Convolutional Neural Network

Vega-Huerta H.; Rivera-Obregón M.; Maquen-Niño G.L.E.; De-La-Cruz-VdV P.; Lázaro-Guillermo J.C.; Pantoja-Collantes J.; Cámara-Figueroa A. · Ingenierie des Systemes d'Information (2025) · DOI: 10.18280/isi.300210

Resumen

Skin cancer is a major health problem worldwide, with China and India being the most affected, according to the Global Cancer Observatory (GCO). In Peru, skin cancer is the fourth most frequent cancer. Its seriousness underscores the importance of early detection, a key factor in improving survival rates and optimizing treatment outcomes. To address this problem, an innovative method is being applied that leverages artificial intelligence to examine skin-related medical images. The main objective of the model is to identify the presence of cancer and categorize the specific type detected. This study employs a machine learning methodology, centered on the use of CNNs along with data augmentation and transformation techniques. The ISIC - 2019, is the dataset comprising 2357 dermatoscopic images, strategically chosen to bolster performance metrics and strengthen the overall resilience of the model. The result of this innovative approach is a remarkable 94% accuracy rate, accompanied by a test loss of 16%. By leveraging advanced technologies such as CNNs and incorporating extensive data sets, this research not only contributes to the field of medical imaging, but also represents a substantial advance in the field of skin cancer diagnosis.

Instituciones: Universidad Nacional Mayor de San Marcos; Universidad Nacional Pedro Ruiz Gallo; Universidad Nacional Intercultural de la Amazonia