Evaluation of registration techniques for spinal image guidance.

OBJECT: Paired point matching alone and paired point matching combined with surface matching are the two techniques used for the registration step in preoperative computerized tomography-based spinal image guidance. In the present study the authors sought to compare paired point-matching registration alone with paired point matching supplemented with surface matching to determine if the addition of surface matching improves navigational accuracy.
METHODS: Pedicle screws were placed in three embalmed human cervicothoracic spinal specimens during image guidance to serve as a reference points. The specimens were then rescanned, and each level was registered using paired point matching alone and then by paired point supplemented with surface matching. Navigational accuracy was assessed by placing the stereotactic probe in the center of the screw head, and measuring the apparent distance between the screw head and probe on the computer monitor. Statistical analysis was used to compare the registration error and navigational error between the two techniques. Seventy-five screws were placed at 46 vertebral levels. The mean registration error for the paired point matching/surface matching technique (0.5 mm) was significantly lower (p < 0.001) than that of the paired point matching alone technique (1.2 mm); however, the intertechnique difference in navigational error was nearly equivalent (1.3 mm compared with 1.4 mm) and statistically insignificant (p > 0.05).
CONCLUSIONS: Although the addition of surface matching to paired point registration significantly decreased the mean registration error, the actual navigational accuracy between the two techniques was equivalent when easily distinguishable points were meticulously selected. The use of paired point matching alone did not compromise the accuracy of navigation and is likely to result in decreased operating time.