@article{ASJ120526,
author = {Piergiorgio Muriana},
title = {Artificial intelligence in thoracic surgery: from nodule detection to surgical planning—a narrative review},
journal = {AME Surgical Journal},
volume = {6},
number = {0},
year = {2026},
keywords = {},
abstract = {Background and Objective: Artificial intelligence (AI) is rapidly expanding its role in thoracic surgery, driven by the growing complexity of imaging data, the diffusion of lung cancer screening (LCS) programs, and the increasing adoption of minimally invasive and parenchyma-sparing procedures. Although numerous AI-based tools have demonstrated promising performance in specific tasks, their translation into routine thoracic surgical practice remains fragmented. This narrative review aims to provide a comprehensive, surgeon-centered overview of current and emerging AI applications across the thoracic surgical pathway, highlighting clinical relevance, limitations, and future directions.Methods: A narrative review of the literature (2015–2025) was conducted using PubMed and Scopus as the primary data sources. Relevant English-language publications addressing AI applications in thoracic imaging, surgical planning, perioperative risk stratification, outcome prediction, and multimodal data integration were identified and thematically synthesized. Given the narrative nature of the review, no formal risk-of-bias assessment or quantitative meta-analysis was performed.Key Content and Findings: AI has shown robust performance in pulmonary nodule detection and malignancy risk prediction, particularly through deep learning (DL)-based imaging analysis and radiomics. Beyond diagnostics, AI-assisted 3D reconstruction and anatomical mapping support preoperative planning and lung-sparing surgical strategies. Machine learning (ML) models also demonstrate promise in predicting postoperative morbidity, mortality, and oncologic outcomes. Emerging approaches integrating imaging, clinical variables, and molecular biomarkers, such as liquid biopsy, suggest a path toward more comprehensive, multimodal decision support. However, most AI applications remain limited by retrospective design, heterogeneity in methodologies, limited external validation, and challenges related to explainability, bias, and workflow integration.Conclusions: AI holds significant potential to enhance thoracic surgical decision-making, but its greatest value lies in augmenting and not replacing surgeon expertise. Future progress will depend on prospective validation, surgeon-driven development, and integration of explainable and multimodal AI frameworks into real-world clinical workflows. If these challenges are addressed, AI may evolve into a cohesive surgical ecosystem supporting more precise, transparent, and individualized thoracic surgical care.},
issn = {2788-578X}, url = {https://asj.amegroups.org/article/view/120526}
}