A Study on Deep Learning Model Performance in Skin Cancer Prediction

Abstract

Skin cancer, one of the fastest-growing cancer types globally, includes both benign and malignant forms. Recent advances in artificial intelligence have enabled the early detection of various diseases with high accuracy, accelerating treatment processes. Convolutional neural networks (CNNs) are widely regarded as powerful tools in artificial intelligence, particularly for deep learning (DL) applications in medical image analysis, including skin cancer detection. This study evaluates a range of pre-trained CNN models namely, VGG16, ResNet50, MobileNetV2, DenseNet201, VGG19, Xception, InceptionV3, and EfficientNetB0 for their effectiveness in classifying skin cancer images from the ISIC Archive dataset