Certain investigations on skin cancer detection using machine learning and deep learning models

Abstract

Skin diseases have become common in recent decades. Numerous newlinefactors influence the appearance of these diseases, and each age group newlinetypically has diverse symptoms. Bacteria and molds that grow in humid and newlinehot climates are exposed to excessive ultraviolet radiation by the sun can newlinemake the skin more sensitive and easily cause infections and skin problems. newlineIn adding to external infections, it can cause other serious skin diseases such newlineas internal sebaceous glands, dead skin, sweat, dust and other unwanted newlinesecretions. Thermal microscopy or epilepsy microscopy (ELM) was first newlinedescribed in 1987; It simplifies the non-invasive diagnostic process using newlineevent light, oil immersion and magnification. Since then, various techniques newlinehave been projected to improve the accuracy of Computer Aided Diagnosis newlineSystem (CADS) in pigmented skin lesions. The CADS is intended to newlinereproduce the decision of the dermatologist for a given dermoscopic skin newlineimage without using any input from dermatologist and provide newlinecomprehensive info regarding the grounds for the decision. The process of newlineCADS involves capturing an unconstrained image of the affected skin area, newlinepre-processing the image, segmenting the affected cancer region, extracting newlineits characteristic features and finally suggesting whether it is benign or newlinemalignant through a classifier with a known database. newlineThe main drawback of current technology is that margin/tipping newlinedecisions are not made in a timely manner, because the edge is extended by newlinesetting the minimum values of margins with the least deviation. newline