Augmented Reality-Guided Pedicle Screw Placement With 2D-3D Registration: Proof-Of-Concept Using 151 Cadaveric Trajectories.
Jose A Castillo, Gabriel Urreola, Venina Kalistratova, Michael Le, Alan M Harris, Hania Shahzad, Dustin H Massel, Nicholas Shepard, Chris Cook, Richard Lim, Safdar Khan, Richard L Price
Abstract
Open AccessStudy DesignProspective, single-center cadaveric study.ObjectivesTo evaluate the accuracy of augmented reality (AR) head-mounted display navigation using a 2D-3D registration method for pedicle screw placement that integrates preoperative CT scans with intraoperative X-ray images, eliminating the need for intraoperative 3D imaging.MethodsA prospective, single-center cadaveric study was conducted using six human cadaveric specimens. Five board-certified spine surgeons placed 151 pedicle screws across cervical (n = 36), thoracic (n = 54), lumbar (n = 41), and sacroiliac (n = 20) regions using an AR navigation system. The system utilized preoperative CT scans registered with intraoperative X-ray images. Accuracy was assessed stereotactically (angular and translational errors) and clinically (Gertzbein-Robbins and Heary classifications).ResultsOf 151 pedicle screws placed, 147 (97.4%) were accurately positioned in the "safe zone" (Gertzbein-Robbins grades A/B or Heary classifications I/II). The 99% upper bound limit for angular errors was <3 and translational errors was < 3 mm in both axial and sagittal planes across all vertebral levels.ConclusionsThe 2D-3D registration methodology with AR visualization demonstrates high accuracy for pedicle screw placement comparable to conventional navigation techniques. This approach effectively addresses key limitations of existing navigation paradigms while maintaining high accuracy standards without requiring intraoperative 3D imaging capabilities. AR-assisted navigation with 2D-3D registration represents a promising technological advancement that may enhance the precision and efficiency of spinal instrumentation procedures.