UCLA

Objective: Three-dimensional (3D) soft tissue (ST) changes from orthodontic treatment can now be evaluated with the development of 3D photography. However, the accuracy and reliability of this method is still in question. This study aims to 1) evaluate the accuracy of 3dMD imaging using caliper, surface, and 3D morphometrics and 2) find standard facial expressions that can be best reproduced by 3dMD imaging for soft tissue analysis.

Methods: Three-dimensional (3D) ST facial landmarks were obtained from 40 adults not undergoing orthodontic treatment through the use of 3dMD facial photographic software. A total of 16 landmarks were used and 21 parameters were measured for surface distances (6 in vertical, 10 in anterior posterior, and 5 in transverse). 3dMD images of four different facial expressions (repose (R), maximum intercuspation (MIP), posed smile (S), and smile with lips closed (SLC)) were taken at 0hr, 1hr, 24hr, 1wk, 2wk, 3wk, and 4wk time intervals. As a feasibility test, these measurements were taken on a mannequin at the above time intervals. Superimposition of seven 3dMD images for each facial posture per subject was performed for analysis. Error magnitude statistics: mean absolute difference (MAD), standard deviation of the error (SD), Root MeanSquare Error (RMSE), relative error magnitude (REM), technical error magnitude (TEM), and intraclass correlation coefficient (ICC) were used.3D morphometrics, seven images were taken of mannequin and subjects posing in MIP and posed smile. These images were loaded into an initial software pipeline to generate a closed mesh. Closed mesh were traced for surface curves and finally loaded into a dual pipeline to average those seven images. Color-coded displacement maps were generated from the dual pipeline images to evaluate changes across the seven time points. T-stats was performed to quantify these changes.

Results: Measurements on a mannequin confirmed the reliability of all the landmarks and parameters used in this study. For 40 subjects, 3dMD measurements between ST landmarks were reproducible with a random error lower than 1mm except for the distance from cheilion to cheilion (2.79mm). Comparing across the four facial postures, MIP showed the smallest variation with mean absolute difference (MAD) (2.03mm) and SD (0.81mm). S posture had the largest variation with MAD (2.79mm) and SD (1.28mm). For 3D morphometrics, mannequin showed that the t-stats are at near zero on the face. When applying 3D morphometrics for MIP and posed smile, areas around exocanthion, cheilion, and pronasale showed slight variation.

Conclusion: 3dMD can perform ST analysis with both accuracy and reproducibility and can be utilized for evaluating ST changes with dental treatment when using advanced technology like 3D morphometrics. Cheilion, alar base, and exocanthion might not be an appropriate landmark to use for ST analysis. R and MIP facial postures are recommended over smile (S or SLC) for more consistent parameter measurements.