Accuracy of multilevel registration in image-guided pedicle screw insertion for adolescent idiopathic scoliosis.

STUDY DESIGN Retrospective clinical study. OBJECTIVE To assess the accuracy of multilevel registration for skip pedicle screw placement during image-guided, computer-assisted spine surgery, in the setting of adolescent idiopathic scoliosis (AIS). SUMMARY OF BACKGROUND DATA Computerized frameless stereotactic image-guidance has been used recently to improve pedicle screw placement accurately and safety during spine surgery. Because of possible intervertebral motion and usual difference in patients' position between preoperative imaging and surgery, the imaging model and the surgically exposed spine may be significantly discordant. Consequently, current protocols suggested separate registration of each spinal level (single-level registration) before respective pedicle screw placement, a time-consuming process. Moreover, although multilevel registration for lumbar spine has been reported, and that for thoracic spine has not. METHODS A total of 19 patients (1 male and 18 females; mean age, 13.9 years) with AIS who underwent multilevel registration for skip pedicle screw placement were included. Variables including surgical time, blood loss, preoperative and 2-year postoperative Cobb angle, correction rate, and postoperative screw position by computed tomography image were evaluated. Mean registration error after point merge and again after surface merge were recorded for each consecutive vertebra of each case. RESULTS Mean surgical time was 310 minutes (range, 168-420 min). Mean blood loss 1138 g (range, 300-2300 g). Cobb angle before operation and at 2 years postoperation was 62.4 degrees (43 degrees-100 degrees) and 21.6 degrees (9 degrees-42 degrees), respectively. Mean correction rate 66.2% (39.7%-84.5%). Total 265 screws were inserted with computed tomography-based navigation system. Pedicle violation was observed in only 4 screws (1.5%). No neurovascular complication occurred. After point merge, average Mean registration error of all cases was 1.69 +/- 0.52 mm, and after surface merge was 0.51 +/- 0.16 mm. CONCLUSION Multilevel registration may decrease operative time without compromising accuracy of pedicle screw placement afforded by this technique in the setting of AIS.