| Objectives: To evaluate correlations between registration errors and accuracy of pedicle screw placement. Methods: Registration errors were analyzed as a relative index of accuracy in the computer imaging of surgical anatomy during spinal navigation. One hundred eighty-four computer-assisted pedicle screw installations were performed in thirty-six patients. Registration errors during spinal navigation were recorded and grouped into A (registration error<1.5mm) and B (registration error more than 1.5mm). The screw placements were evaluated by postoperative CT and were categorized as good (in the central core of the pedicle), fair (eccentricity of the screw and erosion of the pedicular cortex with perforation of less than 2mm), and poor (more than 2mm perforation). Results: Ninety-four and ninety screw placements were examined in groups A (registration error<1.5mm) and B (registration error more than 1.5mm), respectively. In group A, seventy-five (79.8%) pedicle screw placements were categorized as good, seventeen (18.1%) were fair, and two (2.1%) were poor. In group B, fifty-seven (63.3%) were good, twenty-six (28. 9%) were fair, and seven (7.8%) were poor. The difference in screw placement results between group A and Breached statistical signficance (P=0.002, independent samples t-test). Conclusion: Significantly more accurate pedicle screw placement can be achieved by limiting registration errors. |