Analysis of Chronic Skin Diseases Using Soft Computing Based Image Processing

Abstract

Skin diseases are one of the most widely recognized reasons for increased death rate newlinearound the world. For deterioration of the death rate due to skin diseases, early detection newlineand cure are of ultimate necessity. In the current state, dermoscopy is a useful and newlinereliable technique for a precise detection of skin lesions. Interpretation of dermoscopic newlineimages is a critical task for radiologists as they recommend patients for biopsy. newlineRadiologists are not able to account numerous interpretations for the same skin lesion newlinefor an image, therefore it is essential to opt for computer based diagnosis. Skin lesion newlineanalysis is a repetitive job that needs maximum care for the evading of misconception. newlineConsequently, Computer-Aided Diagnosis (CAD) system is remarkably common newlinenowadays. CAD systems analyses the skin lesion with the practice of image processing newlineand pattern recognition techniques. It classifies them into different classes viz., newlinemalignant, benign, and normal. The CAD system classifies the type of skin lesion newlineautomatically by assembling and investigating important features from skin lesion newlineimages. The goal of this thesis is to evolve a novel and useful segmentation technique, newlineorthogonal features, feature integration, and mathematical modelling for the sensitivity newlineand specificity. Numerous classifiers are used for classification of skin lesion to derive newlinean overall inference. Each scheme has been validated using standard ISIC-2015 to ISIC- newline2017 database. newline