An Effective Neural Network Assisted Melanoma Disease Prediction based on Dermoscopy Images

Using digital dermoscopy pictures and a Morphological Neural Network (MNN) model, we were able to distinguish between benign and malignant melanocytic tumors. A self-generating neural network (SGNN) is used to collect tumor features, and an ensemble classifier for neural networks is used to classify lesions. It is frequent in clinical settings for lesions to be so massive that the dermoscopy picture cannot catch them. New border characteristics are presented to cope with this problematic presentation, which may successfully describe border abnormalities on both complete and partial lesions. BP and fuzzy neural networks are combined in a network ensemble classifier in our model to get better performance. Studies employ dermoscopy databases including images of xanthous and Caucasian people. The additional border characteristics and the suggested classifier model improve classification accuracy significantly, according to the findings.