What is the angular accuracy of regional voxel-based registration for segmental Le Fort I and genioplasty osteotomies?

Among the accuracy analysis techniques for orthognathic surgery, regional voxel-based registration (R-VBR) has robust data, but remains unvalidated for smaller jaw segments. The purpose of this study was to validate the angular accuracy of R-VBR for segmental Le Fort I (SLFI) and genioplasty osteotomies. Postoperative cone beam computed tomography (CBCT) of consecutive patients with three-piece SLFI or genioplasties was rotated to a known pitch/roll/yaw (P/R/Y). Using R-VBR, a copy of the raw CBCT was superimposed onto the rotated CBCT at four mutual regions of interest (ROI): anterior, right posterior, and left posterior maxilla, and chin. The P/R/Y of each was subtracted from those of the rotated CBCT to calculate the angular error. The predictor and outcome variables were ROI and absolute angular error, respectively. The accuracy threshold was 0.5°. Ten SLFI and 34 genioplasties were analyzed based on the sample size calculation. The one-sample t-test and Wilcoxon signed rank test were applied in the analysis. The mean absolute error was 0.20-0.54° for the maxillary segments (all P ≤ 0.01) and 0.83-2.51° for the genioplasty segments (all P < 0.001). R-VBR has variable angular accuracy for SLFI osteotomies and may be insufficient for genioplasty. The findings may allow the design and interpretation of studies on SLFI and genioplasty with greater rigor, thereby contributing to minimizing the discrepancy between planned and achieved outcomes.