SkinScanNet: A CNN-Based Model with Explainable AI for Reliable and Transparent Skin Cancer Detection

Early and accurate detection of skin cancer, particularly melanoma, is critical for improving survival rates and treatment outcomes. This study introduces SkinScanNet, a CNN-based deep learning model that classifies skin lesions into benign and malignant categories. The model demonstrates robust performance, achieving a test accuracy of around 94%, with precision, recall, and F1-scores of 95.45%, 89.94%, and 92.56%, respectively. Gradient-weighted class activation mapping (Grad-CAM++), an explainable artificial intelligence (XAI) technique, is integrated into the model to generate class-discriminative heatmaps, highlighting key regions influencing predictions. This enhances transparency, enabling clinicians to validate and interpret model decisions. The system addresses key challenges in artificial intelligence (AI)-driven diagnostics by combining high accuracy with interpretability, offering a reliable and explainable tool for real-world clinical applications. Future directions include expanding dataset diversity and optimizing the model for broader generalizability and reduced false negatives.