Artificial Intelligence in Intraoperative Imaging and Navigation for Spine Surgery: A Narrative Review

Comprehensive Summary

In this article, the authors discussed the application of artificial intelligence, machine learning, and deep learning to improve intraoperative imaging and navigation during spine surgeries. First, the authors provided fundamental knowledge regarding AI in order to demonstrate its application to spine surgery. They discussed the development history of computer-aided navigation in spine surgery from using fluoroscopy and CT scans for navigation to the introduction of intelligent solutions. One of the central points discussed by the authors concerns the ability of artificial intelligence algorithms to assist in recognizing the image and detecting pathologies. Algorithms could perform different functions, including identification of anatomic landmarks, abnormalities classification, and surgical planning. After that, the paper presented the available imaging technologies used in intraoperative settings, including O-arm, GE, Airo CT systems, surface-based registration from Medtronic and Stryker, and optical surface mapping systems such as 7D Surgical. Moreover, the authors pay attention to new technologies based on artificial intelligence. Specifically, they focus on low dose intraoperative CT scans, broader scan field, software aimed at metal artifact reduction, fusion of 2D intraoperative fluoroscopy with the results of preoperative CT scans, as well as 3D reconstructions based on 2D images.

Outcomes and Implications

Some implications of adopting this technology for the clinic include the significant importance of localization in spine surgery, instrumentation accuracy, and the risk of damaging neurologic or vascular structures. AI-assisted imaging may reduce complication rates through improved screw accuracy, reduced rates of reoperations, and aid the surgeons with interpreting anatomy in complicated cases, such as those involving deformity, revision, or obesity when anatomical landmarks are unclear. Minimizing radiation in these protocols is crucial for young patients, who may need many surgeries in their lifetime; for the same reason, the exposure of surgeons and OR personnel to CT scans and fluoroscopy can be reduced. Workflow optimization, such as automatic recognition of structures, can reduce operative times and surgeon fatigue while increasing OR efficiency, which would lead to cost savings as well. Overall, AI systems can help perform personalized surgery based on imaging, patient comorbidities, and biomechanics by suggesting optimal trajectories or implants.