Abstract
Accurate detection methods are desperately needed, as skin cancer concerns around the world continue to rise. Conventional methods, which depend on dermatologists’ subjective visual evaluation, face difficulties that highlight the critical role of artificial intelligence (AI). This study leverages the ISIC2019 dataset and advanced deep learning models to tackle the challenging job of 8-class skin cancer classification. Multiple pre-trained Vision Transformer models and ImageNet topologies are used in an extensive examination. An unique “Naturalize” augmentation technique is presented to resolve intrinsic class imbalances, which result in the construction of the pioneering Naturalized 7.2K ISIC2019 and the groundbreaking 2.4K ISIC2019 datasets, improving classification accuracy. The study emphasizes how important AI is in reducing the possibility of an incorrect or underdiagnosed case of skin cancer. The ability of AI to analyze large datasets and identify subtle trends greatly enhances diagnostic skills. Early detection and better patient outcomes are promised by the integration of AI-powered technologies. Within the Naturalized 7.2K ISIC2019 dataset, quantitative measurements such as confusion matrices and visual assessments utilizing Score-CAM demonstrate an unparalleled success: 100% accuracy, precision, recall, and F1-Score. In conclusion, this study shows how AI-powered systems can significantly improve skin cancer diagnosis, marking a noteworthy development in the area of dermatological diagnostics. The Naturalized 7.2K ISIC2019 dataset demonstrated 100 performance across crucial parameters, demonstrating the revolutionary power of AI-driven techniques that are changing the face of skin cancer detection and patient treatment.
Mohamad Abou Ali
Postdoctoral Researcher
I develop generalizable deep learning methods for biomedical imaging, advancing diagnostic accuracy, data augmentation, and intelligent healthcare systems.
Fadi Dornaika
Ikerbasque Research Professor
Ikerbasque Research Professor with expertise in computer vision, machine learning, and pattern recognition.
Ignacio Arganda-Carreras
Ikerbasque Research Associate Professor
My research interests include image processing, computer vision, and deep learning for biomedical applications.