Remote Skin Diseases Diagnosis System Using Machine Learning Techniques

Abstract

Skin diseases have become one of the most common diseases all over the world, beside their painful effects they are spreading very fast to cover a larger area and also have a psychological effect to the patients, the diagnosis of the skin diseases requires a high level of expertise and they are subjective to the dermatologist, so computer aided skin diseases diagnosis system is proposed to provide more objective and reliable solution to this problem. This project aims to develop skin diseases diagnosis system with a mobile interface, the system is built on a machine learning model to classify the infected images using Bag of Features Model with SVM classifier and develop an ANDROID interface application to capture the images, the designed model has successfully able to classify the infected images of 3 sample classes with accuracy 94% of cross-validation method and 85% of holdout method. The system is built successfully and the interface application is communicating properly with the server established, and all the system functionalities is working properly, despite that there are some problems occurred through the development of the system starting with the data collected that are distorted by a watermark that obstacle the classification process and the synchronization between the server and the client sides of the system, The system developed is using a single server, which by the increase of the number of the users will face a performance issue, and also the system interface is available for the android users only, these are the proposed future improvements of the system