Skin Disease Classification using Deep Learning Methods

Abstract

One of the major illnesses combating human races is Skin disease. Some skin diseases if not detected and treated early can result into cancer - a killer disease or disfigure the bearer. Discovery of these diseases frequently relies on the expertise of the medical professionals and skin biopsy results, in which sometimes the accuracy and prediction is deficient and as well is time consuming. Misdiagnosis is very rampart because these diseases always look alike, and could possibly be mistaken for each other. Therefore, there is need for a computer-based system for skin disease identification and classification through images to improve the diagnostic accuracy as well as to handle the scarcity of human experts. The current research sought to classify three selected skin diseases (Benign keratosis, Actinic keratosis and Dermatofibroma) that could disfigure or lead to cancer if proper diagnosis is not given. A convolutional neural network method designed upon tensor flow framework was used for the classification of the diseases. At the end of the implementation, results from the proposed system exhibits disease identification accuracy of 72% for Benign keratosis, 77% for Actinic keratosis and 69% for Dermatofibroma.