An Explainable AI Framework for Early-Stage Lung Cancer Diagnosis through Deep Neural and Vision Transformer Architectures

Detecting‍‌‍‍‌ lung cancer at an early stage is very important for lowering deaths, but still there are no accurate and interpretable diagnostic systems. In this paper, the authors develop an explainable artificial intelligence (XAI) framework for diagnosis of early-stage lung cancer through computed tomography (CT) images. The authors try to improve diagnostic accuracy, robustness, and interpretability by integrating deep convolutional neural networks (CNNs) with Vision Transformer (ViT) architectures. The proposed deep learning model, which integrates patient metadata and radiomics features for multimodal fusion, allows a better contextual understanding of tumor heterogeneity, among other good things. Experiments using standard datasets (LIDC-IDRI, LUNA16) show that the proposed CNN–ViT hybrid model outperforms other typical deep models in accuracy and transparency. The findings of this research reveal the possibility of using self-supervised learning and transformer-based architectures together to create AI tools that are not only innovative but also reliable for clinical ‍‌‍‍‌decision-making.