LinkNet-B7: Noise Removal and Lesion Segmentation in Images of Skin Cancer

Skin cancer is common nowadays. Early diagnosis of skin essential to increase patients’ survival rate. In addition traditional methods, computer-aided used in cancer. One the benefits this method that it eliminates human error diagnosis. images may contain noise such as like hair, ink spots, rulers, etc., lesion. For reason, removal required. The reduction lesion can be referred removal. This phase very important for correct segmentation lesions. most critical problems using automated methods inaccuracy because and cannot performed effectively. We have created a dataset (hair, etc.) includes 2500 masks. There no literature. images. Two datasets from International Imaging Collaboration (ISIC) PH2 were study. study, new approach called LinkNet-B7 presented. LinkNet-based uses EfficientNetB7 encoder. with 16 slices. way, we lose fewer pixel values. has 6% higher success rate than LinkNet same parameters. Training accuracy was calculated 95.72% 97.80%, respectively.