Artificial Intelligence in Pulmonary Nodule Assessment: From Radiomics to Multimodal Integration

Abstract

: Accurate characterization of pulmonary nodules is critical for early lung cancer diagnosis and treatment planning. Recent advances in artificial intelligence (AI) have demonstrated substantial potential in automating nodule detection, segmentation, malignancy classification, and invasiveness prediction. This review summarizes key developments across radiomics, deep learning, pathomics, vision-language models, and liquid biopsy, highlighting the transition from single-modality analysis to multimodal integration. Representative studies are discussed to illustrate the performance gains and remaining challenges in clinical translation, including generalizability, interpretability, and implementation feasibility.