Dynamic analysis of nerves around the elbow using magnetic resonance imaging-derived three-dimensional modeling

BACKGROUND: We are developing an elbow arthroscopy system that overlays three-dimensional (3-D) nerve and bone information in real time by means of augmented reality technology. The aims of this study were: (1) to extract 3-D bone and nerve data from elbow magnetic resonance imaging (MRI) scans obtained in multiple volunteers, (2) to generate an averaged 3-D model of perielbow nerves, and (3) to analyze nerve kinematics during flexion and extension. METHODS: Fifteen healthy volunteers (11 male, 4 female) underwent elbow MRI in a 3T scanner using a T1 VIBE fat-suppressed sequence. Images were acquired at 0°, 45°, and 90° of elbow flexion. DICOM data were converted to 3-D models of bone and nerves with VoTracer software, and STL files were exported. Using Rhinoceros 8, the humeral models of all participants were manually registered under 2 methods-humeral axis-based and flexion-extension axis-based-and the accuracy of each was tested. Nerve models were then averaged for each flexion angle. Optimal joint angles for establishing standard medial and lateral arthroscopic portals were also assessed. RESULTS: No significant difference in registration accuracy was found between the 2 methods. Intraobserver and interobserver reliability were excellent (ICC >0.9). Averaged 3-D nerve models were successfully created. For the lateral portals, the distance between the lateral epicondyle and the posterior interosseous nerve increased significantly with flexion (0° = 29.7 mm, 45° = 31.2 mm, 90° = 34.7 mm; P = .001). For the medial portal (20 mm proximal to the medial epicondyle), the median-ulnar nerve distance tended to widen in extension but without statistical significance (0° = 35.3 mm, 45° = 33.6 mm, 90° = 33.3 mm; P = .201). CONCLUSION: This study demonstrated that 3-D bone and nerve data can be extracted from a single T1 VIBE fat-suppressed MRI sequence, permitting generalization of perielbow nerve trajectories in vivo throughout the flexion-extension arc. These findings provide quantitative guidance on elbow position when creating lateral and medial arthroscopic portals.