An Extensive Examination of Methods for Detecting and Classifying Malignant and Benign Skin Cancers Using Deep Learning Techniques

Skin cancer is a pressing health concern responsible for numerous fatalities in today's world, often diagnosed in its advanced stages, making treatment planning and patient survival challenging. Medical image interpretation, such as MRI as well as CT scans, is crucial for skin cancer diagnosis, but the manual analysis is costly, time-consuming, and can be influenced by biases. DL (Deep learning), a cutting-edge technology, is gaining attention as it allows for the automatic evaluation and interpretation of medical images, reducing the need for human intervention. Deep learning approaches have demonstrated remarkable outcomes in various domains by automatically extracting intrinsic image features, excluding the need for manual feature extraction. This review delves into the application of DL approaches in the diagnosis & detection of cancer, with a particular focus on skin cancer. Detecting melanoma from samples of dermoscopic skin is a formidable task, but employing a machine vision tool, like a deep learning approach, can address some of these challenges. The study introduces an automated melanoma classifier on the basis of DCNN (Deep Convolutional Neural Network) to accurately differentiate malignant from benign melanoma. It also outlines the various deep learning models and steps involved in utilizing these models for skin cancer detection. Recent advancements in the skin cancer detection Deep Learning (DL) methods are explored and summarized, shedding light on the critical challenges that have been faced in accurately detecting the cancer of skin in its early stages. For this research, dermoscopic images from the International Skin Imaging Collaboration datastores (ISIC 2016, ISIC 2017, and ISIC 2020), containing diverse cancer samples, were obtained and evaluated based on metrics like precision, accuracy, specificity, recall, and F1-score.