Prediction of Alopecia Areata using CNN

Abstract

Alopecia Areata is a disease which is chronic in nature, it leads to drastic hair damage. This is an autoimmune disease that can have several aftermaths like brittle nails and bald spots. Its origination can be affiliated with hormonal changes and hereditary factors. There is no clinical test for its detection and the majority of diagnosis is conducted by examining the patients with the naked eye. Thus the expertise of the doctor and hearsay is the only means of detection, the obvious repercussion being low credibility of results. The aim of this study is to present a novel CNN architecture that can facilitate and streamline the arena of detection by taking an image-based dataset. The results of the CNN are compared with four machine learning models that are Naive Bayes. Support Vector Machine, Logistic Regression, and Decision Tree. The dataset has been procured by web scraping and the Dermnet dataset. Image preprocessing techniques haw been used to augment and enhance the dataset. The best accuracy obtained is 98%.