Qualitative and quantitative accuracy of CAOS in a standardized in vitro spine model.

Pedicle breach with screw implantation is relatively common. For clinical application of computer-assisted orthopaedic surgery, it is important to quantitatively know the accuracy and localization of any guidance modality. We ascertained the accuracy of computed tomography and C-arm-based navigated drilling versus conventional fluoroscopy using an artificial thoracic and lumbar spine model. The 3.2-mm diameter transpedicle drilling target was the center of a 4-mm steel ball fixed in the anterior left pedicle axis. After drilling, we used computed tomography to verify the position of the steel ball and the canal and visually explored for cortex perforation. Quantitative vector calculation showed computed tomography-based navigation had the greatest accuracy (median, d(thoracic) = 1.4 mm; median, d(lumbar) = 1.8 mm) followed by C-arm navigation (median, d(thoracic) = 2.6 mm; median, d(lumbar) = 2 mm) and the conventional procedure (median, d(thoracic) = 2.2 mm; median, d(lumbar) = 2.7 mm). Visual examination showed a decreased perforation rate in navigated drillings. We found no correlation between pedicle breaches and inaccurate drilling. The data suggest computer-assisted orthopaedic surgery cannot provide sub-millimeter accuracy, and complete prevention of pedicle perforation is not realistic.