Artificial Intelligence in Low-Dose Computed Tomography Lung Cancer Screening: Clinical Integration, Validation, and Translational Challenges

Lung cancer remains the leading cause of cancer-related mortality worldwide, largely due to late-stage diagnosis. Low-dose computed tomography (LDCT) screening has demonstrated significant mortality reduction in high-risk populations; however, its widespread implementation is limited by high false-positive rates, inter-reader variability, and substantial workflow burden. Artificial intelligence (AI) has emerged as a promising adjunct to address these challenges by enhancing diagnostic consistency, efficiency, and risk stratification in LDCT-based screening. This narrative review aims to synthesize current evidence on AI methodologies, clinical applications, validation studies, and translational challenges in LDCT-based lung cancer screening. A structured literature search was conducted across PubMed, Scopus, Embase, and the Cochrane Library for studies published between January 2010 and September 2025, using relevant keywords related to AI, radiomics, and lung cancer screening. Studies were selected based on their focus on AI applications in LDCT, including detection, characterization, risk prediction, and workflow optimization. Recent advances in deep learning and radiomics have enabled automated detection, segmentation, and characterization of pulmonary nodules with performance comparable to expert radiologists. Hybrid AI models that integrate imaging-derived features with clinical and demographic data further improve individualized risk prediction and support tailored screening strategies. AI-supported workflows have demonstrated improved efficiency by reducing interpretation time while maintaining diagnostic accuracy. Despite these advances, translation into routine clinical practice remains inconsistent due to limitations in external validation, generalizability, interpretability, and workflow integration. Radiologists' trust and human-AI interaction further influence real-world adoption. This review highlights the need to shift focus from algorithmic performance to clinical integration and human-AI collaboration to ensure meaningful improvements in lung cancer screening outcomes.