Enhancing recurrent laryngeal nerve localization during transoral endoscopic thyroid surgery using augmented reality: a proof-of-concept study

During transoral endoscopic thyroidectomy, preserving the recurrent laryngeal nerve (RLN) is a major challenge because visualization of this nerve is often obstructed by the thyroid itself, increasing the risk of serious complications. This study explores the application of an augmented reality (AR) system to facilitate easier identification of the RLN during transoral endoscopic thyroidectomy. Three patients scheduled for transoral endoscopic thyroidectomy were enrolled in this proof-of-concept study. Preoperative computed tomography scans were used to create an AR model that included the thyroid, trachea, veins, arteries, and RLN. The model was overlaid onto real-time endoscopic camera images during live surgeries. Manual registration of the AR model was performed using a customized controller. The model was aligned with surgical landmarks such as the trachea and common carotid artery. Manual registration accuracy was assessed using the Dice similarity coefficient (DSC) to evaluate the alignment between the real RLN and the RLN of the AR model. The 3 patients included were female (mean age, 33.3 ± 15.7 years), and the mean tumor size was 1.0 ± 0.3 cm. All patients underwent transoral endoscopic thyroidectomy of the right lobe. Final histopathological diagnoses comprised 2 papillary thyroid carcinomas and one follicular adenoma. The manual registration accuracy was 0.60, 0.70, and 0.57 for patients 1, 2, and 3, respectively, with a mean value of 0.6 ± 0.1. The application of an AR system during transoral endoscopic thyroidectomy proved feasible and demonstrated potential for improving the localization of anatomical structures, particularly the RLN, as indicated by a moderate DSC.