Robotic-assisted pedicle screw placement improves accuracy, precision, and breach rates compared with freehand and image-guided navigation techniques: a cadaveric study.

Study design: Cadaveric study.

Purpose: Compare the accuracy and precision of robotic navigation (RAN), freehand (FH) and freehand navigation (NAV) techniques for pedicle screw placement.

Methods: Three cadavers were assigned to the FH, NAV, and RAN methods. Cadavers were CT scanned preoperatively, then underwent bilateral T1-L5 screw planning. RAN and NAV screws were placed by one surgeon, while the FH screws were placed by a second surgeon. Automated computer vision techniques were used to assess the tip, tail, and mid-pedicle screw positions compared to the preoperative plan along three axes on postoperative CT scans. Systematic error, precision, and accuracy were defined by signed mean error, two standard deviations (SDs), and mean absolute error (MAE), respectively. A factorial ANOVA with pairwise comparisons was used with post-hoc t-tests using Bonferroni corrections.

Results: Eighty-seven screws were placed in three cadavers (24 RAN, 29 NAV, 34 FH). Fifteen screws were skipped due to registration difficulties (10 RAN, 5 NAV). RAN outperformed NAV and FH in regard to MAE angular deviation from planned screw trajectory (p < 0.001), the mid-pedicle in the superior-inferior direction (p = 0.002), and at the screw tip in all planes (p < 0.05), with a maximum error of 1.29 mm and 1.98° (vs. 2.29 mm and 5.23° for NAV and 5.67 mm and 11.18° for FH). No RAN screws had medial breach (1 (3.4%) for NAV; 4 (11.8%) for FH).

Conclusions: In our study, RAN demonstrated the best accuracy and precision in the majority of measured parameters, when compared to FH and NAV techniques.