Accuracy Of Computer-Aided Prediction For Protrusive Deformity Tissue Changes

To evaluate the reliability of computer-aided systems in predicting soft and hard tissue changes following orthodontic treatment for protrusive facial deformities, providing a reference for clinical treatment planning. A retrospective cohort study was conducted. 1073 eligible adult orthodontic patients were enrolled. A random sample of 100 cases (50 skeletal Class I and 50 Class II) was selected as the study subjects. Pretreatment lateral cephalograms were analyzed using iOrtho and Dolphin Imaging software to simulate post-treatment tissue changes. Predictions were superimposed on actual post-treatment images to quantify errors in 17 cephalometric landmarks and 22 parameters. Independent samples t-tests were performed to compare prediction accuracy between skeletal classes. Soft tissue predictions in skeletal Class II displayed mean errors of 1.5-3.5 mm (iOrtho) and 1.0-3.6 mm (Dolphin). iOrtho showed statistically significant deviations (P<0.05) in 19 parameters for Class II patients, with 9 parameters (e.g., H-N'Pm', UL-EP, U1-Apo) exhibiting errors larger than those in Class I(P<0.05); Dolphin Imaging similarly revealed higher errors in Class II for 6 parameters (e.g., H-N'Pm', A-NPo, Superior Sulcus Depth) (P<0.05). Upper incisor movement predictions in Class II also showed greater deviations (4.43±0.27mm). Current computer-aided systems inadequately predict tissue changes for skeletal Class II protrusive deformities, particularly for upper incisor retraction and lip profile adjustments. Future efforts should focus on optimizing prediction algorithms to improve reliability for challenging cases.