CT vs. MRI-based synthetic CT: accuracy of thoracic and lumbar pedicle screw placement using 3D-printed guides.

Abstract

3D-printed patient-specific guides enable accurate pedicle screw placement in the treatment of paediatric spinal deformities. MRI-based synthetic CT (sCT) eliminates the radiation exposure associated with the preoperative planning CT, while still providing detailed bony morphology. It may therefore offer a radiation-free method for pedicle screw navigation. Thus, the aim of this study was to compare the accuracy of thoracic and lumbar pedicle screw placement using 3D-printed guides based on conventional CT versus sCT. Full-spine MRI and CT images of four human cadaveric spines were obtained, followed by AI-generation of sCTs, the digital planning of pedicle screws and 3D-printed guide manufacturing. A total of 62 K-wires were inserted, 30 sCT-based and 32 CT-based. Accuracy was evaluated by superimpose analysis of planned and achieved trajectories, and by the postoperative grading of the cortical breach rate using the Gertzbein-Robbins classification, based on projected screws along the K-wires. Median translational entry-point deviations were 1.0 mm vertically and 0.5 mm horizontally for both sCT-based guides and CT-based guides. Angular deviations in the axial plane were 1.0˚ for sCT and 0.9˚ for CT. In the sagittal plane these were 1.8˚ and 2.1˚, respectively. No significant differences were observed between sCT and CT. Grading of pedicle's cortical breach showed no severe malposition for both sCT- and CT-based guides. sCT matched the accuracy of conventional CT in providing source data for personalized 3D-printed guides for thoracic and lumbar pedicle screw placement in terms of translational- and angular deviations, and cortical breach rates in this cadaveric study.