Objective: To compare the accuracy of radiography and computed tomography (CT) in predicting implant position in relation to the vertebral canal in the cervical and thoracolumbar vertebral column.

Study design: In vitro imaging and anatomic study.

Animals: Medium-sized canine cadaver vertebral columns (n=12).

Methods: Steinmann pins were inserted into cervical and thoracolumbar vertebrae based on established landmarks but without predetermination of vertebral canal violation. Radiographs and CT images were obtained and evaluated by 6 individuals. A random subset of pins was evaluated for ability to distinguish left from right pins on radiographs. The ability to correctly identify vertebral canal penetration for all pins was assessed both on radiographs and CT. Spines were then anatomically prepared and visual examination of pin penetration into the canal served as the gold standard.

Results: Left/right accuracy was 93.1%. Overall sensitivity of radiographs and CT to detect vertebral canal penetration by an implant were significantly different and estimated as 50.7% and 93.4%, respectively (P<.0001). Sensitivity was significantly higher for complete versus partial penetration and for radiologists compared with nonradiologists for both imaging modalities. Overall specificity of radiographs and CT to detect vertebral canal penetration was 82.9% and 86.4%, respectively (P=.049).

Conclusions: CT was superior to radiographic assessment and is the recommended imaging modality to assess penetration into the vertebral canal.

Clinical relevance: CT is significantly more accurate in identifying vertebral canal violation by Steinmann pins and should be performed postoperatively to assess implant position.