Introduction: Computer-assisted frameless stereotactic navigation is used in many cranial procedures for subsurface localization, and may be employed in the spine to guide hardware placement and bony decompression. Current systems register to pre- or intra-operative imaging with point-matching systems, and are hampered by cumbersome registration procedures, additional radiation exposure, and inability to account for tissue movement between imaging and navigation. A novel structured-light illumination machine-vision system was developed for craniospinal neuronavigation. Here, we validate its accuracy in the mobile cervical spine.
Methods: We retrospectively reviewed 6 patients from a cohort of 118 enrolled in a prospective trial of our machine-vision system, who underwent posterior cervical instrumentation +/- decompression. All patients underwent thin-slice preoperative CT imaging, with intraoperative patient registration performed to the preoperative scan. Navigation data were compared to screw positions on postoperative imaging, and the absolute translational and angular deviation in axial and sagittal planes computed. All procedures were performed by a single surgeon (VY).
Results: 22 screws were analyzed; 2 pars screws at C2, 14 lateral mass screws at C3-5, and 6 pedicle screws at C7. Absolute translational errors were 1.52 +/- 1.32mm and 1.06 +/- 0.97mm in the axial and sagittal planes, respectively; absolute angular deviations were 3.69 +/- 2.63 degrees and 2.83 +/- 2.65 degrees, respectively (mean +/- SD). There was no difference in translational or angular errors between pars (C2), lateral mass (C3-5) and pedicle (C7) screws. The tip of one C2 pars screw breached by <1mm, without sequelae. There were no violations of facet joints, intervertebral foramina, foramen transversaria or spinal canal, and no neurovascular injuries.
Conclusions: Optical machine-vision is a novel technique for craniospinal navigation that allows efficient initial and repeat registration with minimal workflow interruption. Accuracy even in the more-mobile cervical spine is comparable to, and well within the error tolerances for, current spinal neuronavigation systems.
Patient Care: Our machine-vision system provides accurate and significantly faster intraoperative navigation which, in the context of spinal surgery, allows accurate hardware implantation to minimize risk of neurovascular injury, while minimizing patient and staff radiation exposure, and minimizing disruption to surgical workflow.
Learning Objectives: By the conclusion of this session, participants should be able to:
1) describe current techniques of intraoperative craniospinal neuronavigation, with advantages and limitations of each
2) understand the basic physics of structured-light illumination as it applies to intraoperative navigation
3) understand the concept and computation of registration errors in neuronavigation