A Review on Skin Disease Classification and Detection Using Deep Learning Techniques

Abstract

Skin cancer ranks among the most dangerous cancers. Skin cancers are commonly referred to as Melanoma. Melanoma is brought on by genetic faults or mutations on the skin, which are caused by Unrepaired Deoxyribonucleic Acid (DNA) in skin cells. It is essential to detect skin cancer in its infancy phase since it is more curable in its initial phases. Skin cancer typically progresses to other regions of the body. Owing to the disease's increased frequency, high mortality rate, and prohibitively high cost of medical treatments, early diagnosis of skin cancer signs is crucial. Due to the fact that how hazardous these disorders are, scholars have developed a number of early-detection techniques for melanoma. Lesion characteristics such as symmetry, colour, size, shape, and others are often utilised to detect skin cancer and distinguish benign skin cancer from melanoma. An in-depth investigation of deep learning techniques for melanoma's early detection is provided in this study. This study discusses the traditional feature extraction-based machine learning approaches for the segmentation and classification of skin lesions. Comparison-oriented research has been conducted to demonstrate the significance of various deep learning-based segmentation and classification approaches.