Study design: This was a human cadaver study to determine the accuracy of conventional radiography and computed radiography in the evaluation of pedicle screw placement and to identify methodology for more precise reading of these examinations.
Objectives: To determine the accuracy of conventional radiography and computed tomography in the evaluation of pedicle screw placement within lumbar vertebral pedicles and to develop methods to improve imaging interpretation.
Summary of background data: Conventional radiography and computed tomography have been used in research and clinical settings to evaluate pedicle screw placement. This study evaluates the interpretative accuracy of readers blinded to the true position of screw placement using both imaging examinations. Furthermore, methodology was developed to improve accuracy of interpretation of these examinations.
Methods: Three cadaver lumbar spines were instrumented bilaterally with pedicle screws from L1 to L5. Thirty pedicles had 6.0 mm AO pedicle screws inserted using standard surgical technique. Seven directions of deliberate misplacement as well as correct placement of screws were performed at random levels for a total of eight possible screw positions. Conventional radiographs and computed tomography scans were obtained. A senior musculoskeletal radiologist and senior spine surgeon interpreted the images while blinded to screw placement. Examiners initially assessed the screws as in or out, followed by assessment of the eight possible types of screw position. Consensus interpretation was obtained regarding the placement of individual screws. The spines were then dissected to visualize the screws and their position related to the pedicle. After determining the true position of the screws, a systematic method was designed and applied to the interpretation of the imaging methods to identify screw positions.
Results: Using conventional radiographs, 63% of the screw placements were correctly identified as in or out of the pedicle. Computed tomography improved accuracy to 87%. Identifying the true directional component of screw position led to a decrease in accuracy (conventional radiographs 37% and computed tomography 47%). Using asystematic method to analyze imaging studies enabled detection of screw positions.
Conclusion: Evaluation of pedicle screw placement is difficult even in experienced hands. A systematic approach to image interpretation should allow for an accurate assessment of pedicle screw placement.