Skin Cancer Detection and Classification Using Deep Learning Approach

Abstract

Skin cancer is among the most common and potentially lethal types of cancer globally, highlighting the urgent necessity for precise and effective screening techniques. This research presents an innovative method for skin cancer diagnosis employing an encoding-decoding technique with convolutional neural networks (CNNs). The encoding-decoding approach enables effective feature extraction from skin lesion images. Concurrently, we utilize CNN architectures, including DenseNet201, VGG16, and Xception, to categorize skin lesions into seven classifications. We meticulously assess the methodology using an extensive dataset, and experimental findings validate the efficacy of the suggested strategy in precisely detecting different forms of skin cancer. Moreover, comparative research of several CNN designs yields significant insights into their strengths and weaknesses for skin cancer diagnosis. This research enhances the domain of computer-aided diagnostics for skin cancer, presenting optimistic opportunities for early identification and better patient outcomes.