Accuracy of intraoperative CT to determine implant position in the feline vertebral spine.

ObjectivesThe aim of this study was to evaluate the accuracy of cone-beam CT (CBCT) for predicting implant position relative to the vertebral canal in the feline thoracolumbar vertebral column.MethodsThis in vitro imaging and anatomical study included 22 feline cadavers. A realistic surgical approach was performed at three thoracolumbar locations. Four smooth 1.2 mm Kirschner wires were inserted per location using the recommended insertion points. Perforation of the spinal canal was not avoided. Images from end-on fluoroscopy (eoF), multi-slice CT (MSCT) and CBCT using a C-arm were obtained. The images were randomised, and the implant position was assessed by four evaluators who classified implant position and recorded their subjective certainty. Sensitivity, specificity and accuracy were calculated using anatomical dissection as the reference standard. Interrater agreement was assessed using Fleiss' kappa (κ). Duration of image acquisition was compared across modalities using a mixed-effects linear model.ResultsFor eoF, the sensitivity and specificity were 81.9% and 80.7%, respectively, with a balanced accuracy of 81.3% (95% confidence interval [CI] 0.79-0.84) and good κ agreement (0.63). For MSCT, sensitivity was 80.2% and specificity was 83.1%. Balanced accuracy was 81.7% (95% CI 0.79-0.84) and κ was good (0.63). CBCT showed a sensitivity of 88.5% and specificity of 83.4%, with a balanced accuracy of 85.9% (95% CI 0.84-0.88) and good κ agreement (0.72). Interrater reliability showed overall κ values of 0.38 for eoF, 0.54 for MSCT and 0.58 for CBCT. Interrater reliability was higher for completely perforating and non-perforating implants and lower for partially perforating implants. CBCT and eoF were performed faster than MSCT (