Skin Cancer Detection Using CNN

Authors: Parul Taygi, Neetu Mourya, Nisha sharma

Abstract: Skin cancer is one of the most common forms of cancer globally, with melanoma posing life-threatening risks if not detected early. Traditional diagnosis relies on visual clinical examination, dermoscopy, and biopsy, all of which require expert dermatologists and are subject to human interpretation. Recent advancements in Artificial Intelligence, especially Convolutional Neural Networks (CNNs), have enabled highly accurate automated analysis of dermoscopic images. This research presents a deep learning–based skin cancer detection system that classifies dermoscopic images into benign or malignant categories. The system uses standardized image preprocessing, augmentation, and a custom CNN architecture trained on publicly available datasets such as ISIC and HAM10000. Performance metrics, including accuracy, precision, recall, F1-score, demonstrate that CNNs can effectively extract hierarchical skin lesion features. Experimental results show high diagnostic performance comparable to dermatologists under controlled settings. The model has potential applications in telemedicine, early screening, and decision-support systems for dermatologists.

DOI: http://doi.org/10.5281/zenodo.21418363