Resnet 50 Based Classification Model for Skin Cancer Detection Using Dermatoscopic Images
Shikha Prasher, Leema Nelson, S. Gomathi
Abstract
Skin cancer is characterized by the formation of abnormal cells, which are mostly brought on by stimulation of the sunlight's ultraviolet (UV) rays. Skin cancer identification manually takes a lot of time. Numerous automated screening techniques are now available to help doctors diagnose patients quickly. The exact time to identify skin cancer is in its early stages because it tends to spread slowly to other body regions and is more treatable in its early stages. In this work, a deep learning based Resnet50 model have been used to detect and classify dermatoscopic skin disease images. To develop this detection model, the dermatoscopic skin diseases images are obtained from Kaggle repository. The performance of the Resnet50 classification model achieves 97.74% accuracy. The the area under the receiver operator characteristic (ROC) is 0.993. This model helps enhance the quality of life of humans.