Classification of Skin Cancer Using Dermoscopy Datasets by an Automated Machine Learning System

In many states, skin cancer is becoming more common, and it affects a large number of individuals worldwide. For the purpose of successfully treating melanoma, an early detection is strongly advised. Therefore, the excision of melanoma tissues at an early stage has a significant impact on the survival rate of skin cancer patients. The conventional approaches for finding cancer are exceedingly time- and pain-consuming. To meet the many melanoma detection issues with high accuracy, a quick and automated detection method is thus required. This study has discussed the general process for automated skin cancer diagnosis. The main goal of this work is to provide a comprehensive description of the various techniques for recognising Dermoscopy pictures. Each method and each of its sub-techniques are discussed. The results of a number of study approaches have also been addressed. The study’s conclusion includes a comparative analysis based on metrics like accuracy, specificity, and responsiveness utilising datasets like PH-2, ISIC 2017, and ISBI 201. One may conclude from the above that the thresholding technique worked significantly better with the highest accuracy up to 98.7% when used with the PH-2 Data set and picture scaling, such as RGB to Gray image.