[Feasibility study of a surgical planning protocol for orthognathic surgery utilizing similarity retrieval from...

To establish a surgical planning workflow for orthognathic surgery based on similarity retrieval from a historical patient database and to evaluate its non-inferiority with the expert's surgical plan through a randomized controlled trial.

A prospective randomized controlled trial was conducted involving 60 patients (19 males, 41 females; aged 18-35 years) scheduled for orthognathic surgery in the Department of Oral and Maxillofacial Surgery at Peking University School of Stomatology between June 2023 and June 2024. Participants were randomly assigned to a test group (n=30) or a control group (n=30). In the test group, surgical plans were generated using a database-driven similarity retrieval process while in the control group plans were developed by the expert based on clinical expe-rience. All surgeries were performed by the same expert. Outcome measures assessed at 6 months post-operatively included both subjective and objective indicators. Subjective evaluations comprised patient and surgeon visual analogue scale (VAS) scores, and FACE questionnaire (FACE-Q) scores, with surgeon assessments conducted by five independent senior surgeons. Objective measures included cephalometric angles [sella-nasion-A point (SNA), sella-nasion-B point (SNB), A point-nasion-B point (ANB)] and root mean square error (RMSE) of facial symmetry regions.

Postoperative subjective assessments demonstrated significant improvements from baseline in both groups (all P < 0.05). Specifically, the VAS scores increased by 30.10±19.67 in the test group versus 25.43±24.48 in the control group as rated by the patients, and by 28.19±10.21 versus 26.71±7.90 as evaluated by the surgeons. Similarly, the FACE-Q scores showed marked enhancements, with patient-reported scores increasing by 33.41±17.75 in the test group and 32.97±17.65 in the control group, and surgeon-assessed scores improving by 37.75±11.60 versus 38.63±10.23, respectively. However, the magnitude of improvement in all these subjective measures did not differ significantly between the test and control groups (all P>0.05 for intergroup comparisons of the change scores). Analysis of postoperative objective measurements revealed that cephalometric values were within the normal range for both groups: SNA angle was 84.06°±3.73° in the test group compared with 85.23°±3.71° in the control group; SNB angle was 81.78°±3.63° versus 83.51°±3.66°; and ANB angle was 2.28°±1.09° versus 1.72°± 1.25°. No statistically significant differences were observed between the two groups for these cephalometric parameters (all P>0.05). Furthermore, three-dimensional facial symmetry, quantified by the average RMSE value, exhibited significant improvement postoperatively compared with preoperative levels [Test group: from (10.39±2.83) mm to (8.35±2.72) mm; Control group: from (8.55±4.95) mm to (7.59±3.56) mm; P < 0.05 for within-group comparisons]. The postoperative average RMSE values between the test and control groups were not statistically different (P>0.05).

Surgical planning based on similarity retrieval from a historical database demonstrated non-inferiority when compared with the conventional expert-driven approach, as evidenced by the absence of statistically significant diffe-rences in both subjective and objective postoperative outcome measures.