Artificial Intelligence-Driven Three-Dimensional Reconstruction System Reduced Unexpected Procedural Changes in Thoracic Surgery.
Authors
Affiliations (12)
Affiliations (12)
- Department of Thoracic Surgery, Peking University People's Hospital, Beijing, China.
- Thoracic Oncology Institute, Peking University People's Hospital, Beijing, China.
- Research Unit of Intelligence Diagnosis and Treatment in Early Non-Small Cell Lung Cancer, Chinese Academy of Medical Sciences, Peking University People's Hospital, Beijing, China.
- Institute of Advanced Clinical Medicine, Peking University, Beijing, China.
- Beijing Key Laboratory of Innovative Application of Big Data in Lung Cancer, Peking University People's Hospital, Beijing, China.
- Institute of Medical Artificial Intelligence, Peking University People's Hospital, Beijing, China.
- Department of Thoracic Surgery, Peking University People's Hospital, Beijing, China. [email protected].
- Thoracic Oncology Institute, Peking University People's Hospital, Beijing, China. [email protected].
- Research Unit of Intelligence Diagnosis and Treatment in Early Non-Small Cell Lung Cancer, Chinese Academy of Medical Sciences, Peking University People's Hospital, Beijing, China. [email protected].
- Institute of Advanced Clinical Medicine, Peking University, Beijing, China. [email protected].
- Beijing Key Laboratory of Innovative Application of Big Data in Lung Cancer, Peking University People's Hospital, Beijing, China. [email protected].
- Institute of Medical Artificial Intelligence, Peking University People's Hospital, Beijing, China. [email protected].
Abstract
Although an artificial intelligence-driven three-dimensional reconstruction system (AI-3D) facilitates preoperative planning, its impact on consistency between planned and actual surgical procedures and perioperative outcomes has not been sufficiently investigated. This study retrospectively analyzed 1197 patients scheduled for segmentectomy (n = 479) or lobectomy (n = 718) between July 2022 and September 2023. The patients were divided into AI-3D and two-dimensional computed tomography (2D-CT) groups. The primary endpoint was the rate of unexpected procedural changes. The secondary endpoints included intraoperative outcomes (operative time, intraoperative blood loss, and the rate of conversion to open surgery) and postoperative outcomes (second operation, second chest tube, drainage volume, and so forth). Propensity score-matching (PSM) was used to minimize selection bias. Among the patients scheduled for segmentectomy, the AI-3D group demonstrated significantly higher consistency between planned and actual procedures than the 2D group (after PSM: 97.3 % vs 80.0 %; P < 0.001). Reason analysis indicated that most changes were attributable to tumor location and/or surgical margin factors. Intra- and postoperative outcomes were almost comparable between the two groups. In this study, AI-driven 3D reconstruction significantly reduced the rate of unexpected procedural changes in segmentectomy. Although it ensured higher surgical reliability in relation to the preoperative plan, it did not show an advantage in terms of other intra- and postoperative outcomes.