Development of an AI-Based Diagnostic Tool for Skin Cancer Classification using Dermoscopic Images

Skin cancer is one of the most prevalent kinds of disease, which affects people of any age group. Detecting early-stage skin cancer is important for advanced treatment and increased survival rates. This research proposes developing an AI-based diagnostic tool for classifying skin cancer with dermoscopic images. The proposed system depends on two different models: Convolution Neural Networks (CNN) and Convolutional Recurrent Neural Networks (CRNN) to classify skin cancers. For this study, dermoscopic images of various cancer samples were gathered from the International Skin Imaging Collaboration (ISIC). The proposed method used advanced pre-processing steps, including normalization and data augmentation, splitting data into train, and validation to improve the performance. From the results, the proposed CNN and CRNN methods have achieved an accuracy of 88.47 per cent and CRNN as 76.87 per cent accuracy respectively. Therefore, this proposed approach provides a framework for predicting skin cancer and its type which benefits medical professionals and patients.