An approach for classifying benign and malignant skin lesions using Optimized Deep Learning and SVM

Abstract

Cancer is the group of many diseases. Among the groups of cancers, skin cancer is the most common form. It makes skin grow in a disorganized manner and forms tumours. These tumours can be categorized as either benign or malignant. Benign tumours are non-cancerous whereas malignant tumors are cancerous.Skin cancer diagnosis is done by skin biopsy,which takes samples of skin tissues which are then examined by the dermatologist using a microscope. Adopting an automated approach for detection of skin cancer from skin lesion images taken from biopsy using computerised methods may help in faster and accurate diagnosis of skin. Because of the increasing death rate, it is necessary to focus on the early detection of cancer.In this work we have proposed an approach of classifying benign (non-cancerous) and malignant (cancerous) skin lesions by employing deep learning techniques and Support Vector Machine (SVM) on image dataset archived by International Skin Image Collaboration (ISIC).