PATTERN CLASSIFICATION TECHNIQUES FOR THE CLASSIFICATION OF CUTANEOUS MANIFESTATIONS OF SYSTEMIC LUPUS ERYTHEMATOSUS

Abstract

Systemic Lupus Erythematosus-lupus (SLE-Lupus) is the prototype of multiorgan autoimmune rheumatic disorders with life threatening systemic manifestations and there is no gold standard test for the diagnosis of SLE. Pattern classification techniques are used to classify the images of SLE and their performances are evaluated. A total of images of 400 patients (200 SLE and 200 normal) are collected from various hospitals available in Tamilnadu state of India, for this experimental study. Features are extracted from these images based on color histogram and texture. Two different classifiers are considered in this study, namely, multilayer perceptron and support vector machines. Experimental results show that the multilayer perceptron model provides higher accuracy of 86.25% than other model, support vector machine for SLE identification.