Application of Medical Images for Melanoma Detection Using a Multi-Architecture Convolutional Neural Network From a Deep Learning Approach

Abstract

Melanoma has a higher tendency to spread to other parts of the human body swiftly if not detected and treated timely. This makes melanoma more dangerous than any other skin cancer disease. Melanoma is a type of skin cancer that develops from the melanocytes. The melanocyte is a genuine skin cell that protects the skin pigment known as melanin. Melanoma has recently become a significant and growing public health concern globally. It is marked by the incidence of millions of new cases annually, encompassing both non-melanoma and melanoma skin cancer. This disease is characterized by the unchecked proliferation of abnormal skin cells, with the potential to metastasize to other anatomical sites. Conventional diagnostic approaches, particularly biopsy-based methods, are invasive, time-consuming, and often culminate in treatment delays and increased patient discomfort. This study assessed their effectiveness in detecting melanoma by applying three distinct deep learning techniques, specifically EfficientNetB3, MobileNetV2, and InceptionV3. Among these architectures, EfficientNetB3 emerged as the standout performer, achieving an exceptional accuracy rate of 90.7% and an impressive area under the curve (AUC) score of 97%. The cascading combination technique was then utilized to develop a multi-architecture model. With the cascading multi-architecture technique, we combined all the layers (multiple layers) output of the models and processed them (the output of the multiple layers) in a structured pipeline, which improves upon the previous output. The results of the multi-architecture model, with an accuracy of 94.86%, signify the optimal architecture for melanoma detection.