Classification of Skin Cancer with Deep Transfer Learning Method

Skin cancer is a serious health hazard for human society. This disease developed when the pigments that produce skin color become cancerous. Dermatologists face difficulties in diagnosing since many colors seem identical. As result, early diagnosis of lesions (the foundation cancer) very crucial and beneficial totally curing patients. Significant progress has been made creating automated methods with development artificial intelligence (AI) technologies to aid dermatologists identification cancer. The widespread acceptance AI-powered enabled use massive collection photos benign sores authorized by histology. research compares six alternative transfer learning networks (deep networks) classification using International Imaging Collaboration (ISIC) dataset. DenseNet, Xception, InceptionResNetV2, ResNet50, MobileNet were employed investigation which successful different studies recently. To compensate imbalance ISIC dataset, classes low frequencies are augmented. results show augmentation appropriate success, high accuracies F-scores decreased false negatives. With an accuracy rate 98.35%, modified DenseNet121 was most model against rest nets utilized study.