An Automated and Efficient Deep Learning based Classification of Multiple Skin Disorders from Skin Lesion Images

Body cancer is among the most prevalent kind of cancer that develops when skin cells get unevenly distributed and multiplied irregularly. This paper proposes a deep learning method which is accurate, efficient and optimized for classification of skin and probabilistic prediction of three benign and three malignant categories, which include three skin cancers: Basal Cell Carcinoma, Melanoma, Squamous Cell Carcinoma, and three skin diseases: Actinic Keratosis, Seborrheic Keratosis, and Nevus. Various discriminatory features are automatically extracted, localized around the affected area of skin which yields unique outcomes for the purposes of distinct identification and classification the suggested classification technique in this work has used neural network classifier having multiple numbers of layers with different functionalities. The classification approach gives excessive degree of accuracy based at the experimental evaluation on unseen test images. The experimental consequences are encouraging indicating a classification accuracy of 98%, signifying the possible use for much computer-vision based similar biomedical application.