Classification of Multi-Class Skin Cancer using Deep Learning Algorithm

In recent years, skin cancer is a frequent and dangerous disease that requires rapid and accurate analysis to secure effective treatment. Accurate detection and diagnosis of skin lesions is challenging due to the similarities between various skin lesions such as nevi and melanoma, especially when analyzing color images of the skin. In this survey, various deep learning (DL) techniques are analyzed for multi-class skin cancer classification. This survey discussed the significant assumptions, limitations, and advantages analyzed in existing DL approaches. DL-based techniques such as feature extraction, segmentation, and classification are used for multi-class skin cancer classification. The performance of the existing methods was analyzed using various performance metrics such as Matthews Correlation Coefficient (MCC), recall, precision, fl-score, dice, precision, sensitivity, Jaccard, and time. This survey concludes that multivariate skin cancer classification over DL overcomes the drawbacks of poor data quality reduces time complexity and improves efficiency.