Early Stage Detection of Psoriasis Using Artificial Intelligence and Image Processing

Abstract

A large part of the Indian population comes below the poverty line. There are thousands of people living in poor sanitation areas. Owing to this, a majority of ailments prevail in such areas. People do not have access to good medical facilities and at times may not even have the awareness to visit a doctor at the correct time. This has been observed especially in the case of skin diseases where failure to get the correct diagnosis and treatment in time often results in dire consequences. With the growing surge of Artificial Intelligence (AI), it has been realized that AI can be helpful in the medical arena as well. Thus, an AI-based approach is proposed for the classification of a skin disease Psoriasis by making use of a multi-class machine learning classification algorithm. Psoriasis is a noncontagious, chronic skin condition that results in the formation of patches of thickened, scaling skin. This algorithm uses uploaded images from the users’ smartphone to give a probability estimate for Psoriasis. Over 60 medically validated images of Psoriasis, acquired from Department of Skin and V.D of Seth G.S. Medical College and KEM Hospital, Parel, Mumbai, were used as training data. It uses a Neural Network algorithm for training the symptoms. Using this training data, an accuracy of above 85% was obtained.