Melanoma Detection Using Deep Convolutional Neural Networks
B. Anitha, K. Umapathi, M. Jayasheela · 2025
Melanoma, a type of skin cancer, could be life-threatening to patients if it becomes malignant. Thus, detecting melanoma in its early stages becomes important. In addition, cancer diagnosis relies on biopsies, which are painful. Therefore, non-invasive early diagnostic tools are the need of the hour. Artificial Intelligence (AI)-based Machine Learning (ML) and Deep Learning (DL) tools have been a boon in biomedical diagnosis. Thus, this book chapter presents the results of a novel and efficient Deep Convolutional Neural Network (DCNN) model in classifying melanoma into benign and malignant classes. The proposed model is trained using a huge dataset of 10,605 dermoscopic images. The model addresses class imbalance and limited data issues by using class-weighted training and data augmentation methods. The model could achieve a classification accuracy of 91%. With its inherent capabilities, the model could aid a dermatologist when integrated into clinical practice.