An Intelligent Approach to Segmentation and Classification of Common Skin Diseases in Sri Lanka

Skin diseases prevail worldwide, and the quality of life and overall health of patients are often hindered as a result. Early detection and treatment are key to a quick recovery. An automated system to identify skin diseases can act as a tool to assist doctors and healthcare workers. This paper presents an intelligent system to segment and classify three common skin diseases in Sri Lanka - tinea versicolor, atopic dermatitis and psoriasis - using image processing, genetic algorithm and machine learning. YUV -based color segmentation was applied to extract the affected region, then the texture and color features were extracted for classification. Genetic algorithm was utilized to obtain the optimized feature subset. An SVM based classifier was then trained and succeeded in classifying the three skin diseases with an overall accuracy of 86.7%.