Abstract

Misdiagnosis of skin diseases is a very common event around the world. Eczema, psoriasis, and inflammation are three very similar diseases, and patients' responses to the symptoms of these diseases are similar, making the work difficult for physicians. Also, these three diseases have many similarities in terms of appearance, making it difficult to diagnose them by a system. Their correct, accurate and reliable diagnosis is crucial. This article aims to distinguish three diseases with acceptable and high accuracy using the fuzzy method. In this article, most of the important features in image processing such as HOG and BRISK, FAST, etc. are extracted (a total of 200000), and then, a total of 42 optimal features were selected through feature selection and neighborhood component analysis for classification (FSCNCA) method.  Finally, skin diseases were distinguished using valid methods of NN, LDA, KNN, SVM and fuzzy methods and the accuracy of diagnosis was obtained at 10.89%, 49.59%, 47.94%, 83.47% and 93.39%, respectively. This study shows that the proposed method distinguishes skin diseases with acceptable accuracy.