Melanoma Skin cancer Detection Using Deep Learning

Abstract

The malignant alteration of melanocytes is a characteristic of melanoma skin cancer, aserious and sometimes lethal illness. In this field, deep learning models have becomeindispensable for accurate and timely detection. This work introduces a novel methodandprovides a thorough overview of current developments in deep learning-based melanomadetection. For picture classification, we assessed a number of integrated models, suchas DenseNet201, DenseNet169, ResNet50V2, and ResNet50V2. Based on our research, DenseNet201 with hyperparameter adjustment has the best training accuracy (94.46%) andvalidation accuracy (86.18%). Hyperparameter adjustment improved the performance of DenseNet169 and ResNet50V2, which also displayed encouraging results. Although useful, ResNet50V2's accuracy was a little bit lower.Deep learning models performbetter thanother models in melanoma diagnosis, which makes this research important for datascientists and medical professionals. It also highlights the significance of ongoingtechnology improvements in medical diagnostics.