Deep Learning Techniques for Skin Diseases Classification Using CNN

Abstract

Diagnosing and classifying skin diseases using the eye or clinical examination was not easy, especially the classification of skin cancer diseases. The check process can take a long time and a lot of effort to diagnose the type of disease. Therefore, many algorithms of different performances appeared using Artificial intelligence and techniques for deep learning, which have an effective and significant role in the ability to detect and diagnose, as these techniques are available to be used in all fields, especially in the classification of medical images that show skin diseases. Thus, the primary goal of the study is to continue to discover the models whose role is to find a solution to the classification problem. The proposed model successfully classified nine different classes of skin diseases by using CNN and threshold-Otsu segmentation method,the model's success classification accuracy was 85.8%. The proposed model can find images that don't fit into any of the nine classes. These images are called "unknown images."