330 patients(185 males and 145 females) from West China Hospital of Stomatology, Sichuan University, Chengdu, China were included in this study. 3D facial images of all the patients were taken from September 2015 to December 2016 using a 3dMD stereophotogrammetry instrument (3dMD face TM System, 3dMD LLC, Atlanta, Georgia, USA) placing the participants in a natural head position. The inclusion criteria were as follows: both genders, aged 19 to 29 years, no history of orthodontic treatments. People with congenital craniofacial anomalies, severe facial deformities, severe malposition of eyes and ears were excluded.
Ethics approval and consent to participate
This study was approved by the Ethical Committee of West China Hospital of
Stomatology, Sichuan University (WCHSIRB-D-2018-072). Written informed
consent was obtained from all the participants above 16 yrs old, or from both
the participants and their parents for those under 16 yrs old.
3D facial image acquisition
The 3D facial images (Figure 1, Figure 2) were taken under standardized conditions using the 3dMD stereophotogrammetry instrument. Patients were seated in front of the flash lamp and they looked at themselves in a mirror 5 feet away, thus ensuring that they were in a natural head position. The patients were informed regarding the face posture to assume: the expression should be relaxed, the mouth closed, and the facial muscles not contracted. Two sets of cameras were placed in front of the patient at the head level.
Subjective evaluation of facial asymmetry
The evaluation committee consisted of 10 judges (5 males and 5 females, ranging 20-26 years) who did not receive any training on esthetic evaluation and did not know any participant. They were invited to evaluate whether a participant’s face was symmetric or asymmetric based on 3D facial images (Figure 1). Each 3D image was shown to each evaluator for 30 seconds, without the possibility to look at the same image again. The images which obtained more than seven(7/10) evaluators’ agreement on symmetry or asymmetry were chosen to be measured in objective measurement. The inter-rater reliability of the 10 evaluators was assessed through modified Cohen’s kappa statistic.(19)
Objective measurement of facial asymmetry
The objective evaluation of facial asymmetry included the whole face analysis and regional face analysis. The 3D facial images were imported into the 3-matic STL 9.0 (X64) and confounding factors (e.g., ears, neck and hair) were removed (Figure 2). Then, soft tissue landmarks (Figure 3(A)) and planes (Figure 3(B)) were created on the images and the left and right side of face was divided into seven regions (Figure 4), respectively: central forehead, nose, lip, chin, lateral forehead, cheek and lateral mandible. The mirror symmetry plane(P1, Figure 3(B)), was mirroring performed on the whole and regional face. Besides, every region was performed separately.
The whole face analysis, firstly obtained the mirror shell of left side face by the software (through the mirror image change). Then the mirror shell was superimposed with the right side face original image. Then the same process to analyze the right side face image (Figure 5). After registration, shell-to-image deviations were graphically presented as color maps (Figure 6) and quantitatively as histograms. This distance map illustrated the absolute distance between corresponding points on the original facial image and the mirrored shell.
The regional face analysis was firstly performed using the soft tissue reference points (Figure 3(A)) which were established in previous studies(13,15,20,21) to divide the whole face into 7 parts (Figure 4). These landmarks were selected according to facial thirds and fifths that are important for the evaluation of facial symmetry. Then the same technique described above for the whole face analysis was used for the assessment of the regional face analysis. The color map of each specific region was also made.
Subsequently it was determined whether the distance between the two layers could be considered as symmetric. According to the past study, the relative distance between the point of the original image and the mirrored shell that was below 1 mm was superimposed and considered as symmetric. Otherwise, they were considered as asymmetric.(22) All the symmetric points constituted the symmetric areas and the asymmetric points form the asymmetric areas. Finally, the degree of symmetry for the whole face and each regional areas was showed by the percentage of symmetric areas.
Statistical analysis was performed using the Statistical Package for Social Sciences Software version 18.0. The diagnostic efficiency of the regional degree of symmetry and the layperson subjective assessment were determined using the receiver operating characteristic (ROC) analysis. The consistency between the symmetric degree and the result of the evaluation was analyzed by the Pearson correlation analysis. A p-value less than 0.05 was considered statistically significant.