Leveraging Convolutional Neural Networks for Accurate Skin Cancer Classification

One of the most prevalent types of cancer in the world is skin cancer, and its early detection is essential for effective treatment and better patient outcomes. This study explored the application of Convolutional Neural Networks (CNN) for skin cancer detection, utilizing two distinct datasets: the Skin Cancer dataset and the Melanoma Cancer Image Dataset. The goal is to assess and compare the performance of different CNN architectures in accurately classifying types of skin cancer. For the Skin Cancer dataset, the VGG16 architecture achieved the highest accuracy of 98%, demonstrating its robust feature extraction and effectiveness in classification. On the Skin Melanoma dataset, DenseNet201 attained an accuracy of 90.32%, highlighting its connectivity and efficient parameter usage, which were well-suited for melanoma classification. These findings emphasized the potential of CNNs, in automating skin cancer detection with high accuracy and development of reliable AI-based diagnostic tools that can aid dermatologists in early cancer detection.