Skin Lesion Classification using Optimized Skin Net Algorithm

Abstract

Skin cancer is one of the much general human diseases realized in all over the world. About five million newer cases of this diseases are realized in the US every year. Early detection and assessment of lesions in the skin are of utmost clinical importance, however, significant issue gets raised whenever there is nil co-ordination between the skin specialist and patient. As a result, a unique deep structure known as Optimized Skin Net is proposed in this work to provide faster screening resolution and help to recently gained physicians in their efforts to make clinical diagnoses of skin-related malignancy. The major motive behind the design and development of Optimized Skin Net is based on two levelled pipelines. Those two levels include where in the lesion segmentation and the lesion classification. The images of the skin diseases have been taken from the publicly available dataset to train and test our deep learning model. Finally, we will be presenting the simulation results along with the outcomes by means of several performance measures like Accuracy, Sensitivity, Specificity, Error rate, False Positive Rate, and ROC.