Skin Lesions Identification and Analysis with Deep Learning Model Using Transfer Learning

Abstract

Sunlight has beneficial as well as harmful rays. Environmental pollution occurs as a result of the depletion of the ozone layer caused by the damage caused by humans to the environment. As a result of these pollutants, skin diseases can be seen in areas exposed to direct sunlight, such as the head and neck. Early detection of actinic keratosis (akiec), basal cell carcinoma (bcc), bening keratosis (bkl), dermafibroma (df), melanoma (mel), melanocytic nevi (nv), and vascular (vasc) skin cancer types, which is one of the most common skin diseases, is important for medical intervention. Otherwise, severe spread, called metastasis, may occur as a result of aggressive growths. For the stated reasons, a deep learning model based on transfer learning, which can classify skin cancer types, has been proposed to assist the medical personnel who serve in this field. With this proposed model, the aim is to classify at high accuracy rates without any pre-processing. As a result of the experimental studies carried out as a result of the stated goals, an accuracy rate of 99,51% was achieved with the proposed model.