AI Recurrent Laryngeal Nerve Navigation for Clinical Implementation

Background With recent technological advancements, artificial intelligence (AI) has been increasingly implemented in areas such as endoscopic diagnosis, radiological imaging, and pathological diagnosis. The objective and consistent recognition and evaluation capabilities of AI in real-time have facilitated its rapid expansion into the surgical field. We believe that AI is valuable not only for surgical assistance but also for evaluating surgical techniques and learning curves. However, it is crucial that surgeons, who directly use medical devices, identify and propose the types of AI that are genuinely useful for surgical support. Methods Our team has been developing AI-based surgical navigation systems, particularly for gastrointestinal surgeries. We have previously reported on AI anatomical navigation and surgical step navigation in various procedures, including hernia surgery, gastrectomy, and prostatectomy. In this study, we developed a system for robotic esophagectomy that automatically identifies the left and right recurrent laryngeal nerves (RLN) and detects risky maneuvers, such as nerve traction. Using recorded surgical videos, we evaluated the system’s clinical value, especially in identifying the right RLN. Results Without AI navigation, surgeons correctly identified the nerve location in only 46.9% of the cases. However, with AI navigation, the accuracy significantly improved to 81.3% (p = 0.004). We are currently working toward regulatory approval and insurance reimbursement for this product. Although this AI navigation system primarily targets the RLN, we have already developed navigation for other anatomical structures, such as the trachea and subclavian artery. Conclusion We believe that this system has the potential to expand into various other surgical navigation applications. In the future, clinical trials will be necessary to verify the effectiveness of AI navigation in reducing surgical complications. However, beyond its direct clinical applications, AI navigation also has potential educational uses. For example, it can be utilized as a teaching tool for medical students and residents using annotated surgical videos. Additionally, metrics such as the number of alerts triggered by the AI during surgery could be used to evaluate surgical techniques and automatically assess learning curves.