Accuracy over space and time of computer-assisted fluoroscopic navigation in the lumbar spine in vivo.

OBJECTIVE

The integration of digital image-guided surgical navigation with C-arm fluoroscopy, known as virtual fluoroscopy (VF), has been shown to enhance the safety of spine surgery in vitro. Few clinical studies have assessed the accuracy of VF during actual spinal surgery, and no studies have investigated variations in accuracy over the course of a series of measurements obtained during operative cases. We sought to study the intraoperative accuracy of VF over time and space during lumbar pedicle screw placement in human patients.

METHODS

Fluoroscopic images of the lumbar spine were obtained, calibrated, and saved to the Stealth Station (FluoroNav) on seven patients undergoing lumbar fusion surgery. The tracking arc was attached to an exposed lumbar spinous process, which was designated the index level. With use of anatomic surface irregularities in the laminae and spinous processes, several points were identified and registered on three different vertebrae directly adjacent to the index level vertebra. Every 15 minutes, throughout the operative case, the probe was brought to each point and the apparent distance from the original location recorded (as measured by the FluoroNav system). Measurements were collected from three vertebral levels adjacent to the index level over a time course of 120 minutes during the operation.

RESULTS

At the index, index +1, index +2, and index +3 levels, 89%, 81%, 92%, and 64% of measurements were within <2 mm, whereas 97%, 96%, 97%, and 91% were within <3 mm, respectively. At 15, 30, 45, 60, 75, 90, 105, and 120 minutes, 96%, 89%, 85%, 61%, 85%, 90%, 93%, and 50% of measurements were within <2 mm, whereas 100%, 93%, 100%, 83%, 100%, 90%, 100%, and 100% of measurements were within <3 mm, respectively. The error in millimeters tended to increase as the distance from the index level increased (R = 0.19, P < 0.05) and as operative time increased (R = 0.26, P < 0.01). Calibration studies of intraoperative VF (IoVF) in the lumbar spine documented a reasonable degree of accuracy. The majority of sequential measurements obtained during IoVF in the lumbar spine were within an error range of <3 mm.

CONCLUSIONS

Our results suggest that the use of VF is a reliable method of verifying the use of anatomic and/or radiographic landmarks for guidance during lumbar pedicle screw placement.