A Comprehensive Review on Deep Learning Model for Melanoma Diagnosis Using Dermoscopy Images

Over the past 30 years, melanoma has become increasingly common, and early detection has been shown to significantly lower mortality rates. Therefore, it can be quite helpful to have access to an automated, trustworthy technology that can detect the existence of melanoma via a dermatoscopic image of lesions and/or skin pigmentation. The earlier skin cancer is detected, the better the prognosis. The widespread adoption of Deep Learning (DL) algorithms in other sectors has led to a surge in the use of AI for making medical diagnoses. With the advent of Convolution Neural Networks (CNNs), a powerful new tool for medical diagnostics has been available. When unwanted noise appears in the image, CNN's ability to extract features is hindered, which in turn affects the classifier's accuracy. In order to boost the system's efficiency, precise extraction of the Region of Interest (ROI) is a necessary first step. Recognising the symptoms of skin cancer early is crucial because of the disease's increasing prevalence, high mortality rate, and expensive treatment costs. Researchers have developed multiple strategies for early detection of skin cancer in light of the seriousness of the issues. Lesion features such as symmetry, colour, size, shape, etc. can be used to diagnose skin cancer and determine whether it is benign or malignant. As a result, this study provides a concise overview of melanoma detection deep learning models. This study provides a state-of-the-art review of how deep learning techniques can be used to detect early-stage skin cancer. A large number of scientists benefit from the survey's findings since they can then improve upon existing automatic deep learning frameworks for melanoma diagnosis.