Skin diseases are common medical conditions, affecting millions of people globally. Early identification of skin illnesses is critical because it can prevent them from developing and becoming fatal. Computer-Aided Diagnosis (CAD) systems have the potential to facilitate early detection of skin diseases, thereby improving patients’ survival rate. As a result, researchers indicate interest in developing CAD tools to help dermatologists and alleviate the difficulties associated with physical examination. This systematic review provides valuable insights into CAD system development for skin lesion detection and classification. A comprehensive review of literature covers the state-of-the-art applications of existing model in skin disease classification. This work aims to discuss about several techniques, architecture, model framework, performance-enhancing approaches and evaluation aspects. The state-of-the-art datasets used in various skin lesion related research articles are analyzed in depth. This analysis emphasizes the importance of incorporating different and high-quality data to improve the effectiveness in skin lesion diagnosis. Additionally, the review addresses shortcomings of existing skin lesion analysis (SLA) and the potential solutions to overcome them. Finally, the survey concludes with a discussion of future work and emphasizes the significance of continued research efforts to establish reliable, explainable, interpretable, and clinically useful classification systems.
G. et al. (Thu,) studied this question.