Artificial intelligence-based detection of acute postoperative airway complications following anterior cervical spine surgery: a retrospective imaging evaluation.
Authors
Affiliations (4)
Affiliations (4)
- Department of Neurosurgery, Sheba Medical Center, Ramat-Gan, Israel.
- Affiliated to Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.
- Arrow Program for Medical Research Education Sheba Medical Center, Ramat-Gan, Israel.
- Faculty of Engineering, Bar-Ilan University, Ramat-Gan, Israel.
Abstract
Retrospective imaging evaluation using an artificial intelligence (AI)-generated model. To develop novel AI software for early prediction and identification of postoperative airway obstruction based on routine post-operative imaging. Acute postoperative airway obstruction following anterior cervical spine surgery due to edema or hematoma is rare but potentially life-threatening. No automated system currently exists to detect this complication in its early stages. All adult patients who underwent anterior cervical fusion between 2012 and 2019 at a single tertiary care center were retrospectively identified. These cases were used to develop an AI model capable of autonomously distinguishing critical airway narrowing from benign postoperative swelling. Image processing and augmentation techniques were applied to establish a segmentation model incorporating three-dimensional reconstructions for computed tomography (CT) scans and pixel analysis for plain radiographs. The Cat- Boost algorithm was harnessed to refine decision trees and generate precise predictions, which were integrated into a graphical user interface for intuitive interaction. A total of 815 anterior cervical fusion procedures were identified; 795 patients met the inclusion criteria. Respiratory complications occurred in 35 patients (4.4%), with 11 (1.3%) caused by airway obstruction. The model achieved a positive predictive value of 0.98, a negative predictive value of 0.90, a sensitivity of 0.91, and a specificity of 0.99. The AI model developed in this study showed strong potential for predicting airway compromise following anterior cervical discectomy and fusion, despite variability between CT and radiographic environments. This tool may facilitate early detection of patients at risk of postoperative airway obstruction and help distinguish benign postoperative changes from complications. Further studies are warranted to validate these initial findings.