- Email: [email protected]
Three-Dimensional Anthropometric Analysis of Chinese Faces and Its Application in Evaluating Facial Deformity
J Oral Maxillofac Surg 69:1195-1206, 2011
Three-Dimensional Anthropometric Analysis of Chinese Faces and Its Application in Evaluating Facial Deformity Yan Dong, PhD,* Yimin Zhao, PhD,† Shizhu Bai, PhD,‡ Guofeng Wu, PhD,§ Lin Zhou, PhD,储 and Bo Wang, PhD¶ Purpose: The aims of this study are to introduce a novel method of 3-dimensional (3D) analysis of the
face and to provide normative data of the Chinese face for surgeons. Materials and Methods: Fifty men and 50 women were recruited, and a 3D stereophotogrammetry system was used to acquire their facial image data. For each subject, the image was aligned to a unified coordinate system, and coordinate values of 31 facial landmarks were collected. Mean values for each landmark were calculated within genders, and 3D models of both genders were constructed based on the mean values. Subsequently, to evaluate the sexual dimorphism, the models were superimposed. Then, to delineate the shape differences independent of size, the models were normalized and superimposed again. The application of the 3D models was also exemplified by analysis of a subject’s facial deformity. Results: Linear and polyhedron 3D models representing the facial shape were built for both genders. The superimposed models and the absolute differences between each landmark in both genders illustrated the sexual dimorphism of the Chinese face, and the normalized models and relative difference for each landmark also delineated the shape differences independent of size. In addition, a subject’s facial deformity was evaluated by referring to the normative 3D facial models. Conclusions: This study describes a new 3D analysis method for facial morphology. Three-dimensional models representing the facial shape of Chinese subjects were built, and sexual dimorphism was investigated. These would provide useful guidance for facial anthropometry and plastic surgeons in clinical practice. © 2011 American Association of Oral and Maxillofacial Surgeons J Oral Maxillofac Surg 69:1195-1206, 2011 Anthropometric studies are an integral part of maxillofacial surgery. Normative data of facial measurements are fundamental to the objective analysis of facial deformity, plan of operation, prediction of final treatment results, and evaluation of surgical outcome.1 Surgical correction of maxillofacial dysmor-
Received from the School of Stomatology, Fourth Military Medical University, Xi’an, People’s Republic of China. *Department of Prosthodontics. †Department of Prosthodontics. ‡Department of Prosthodontics. §Department of Prosthodontics. 储Department of Prosthodontics. ¶Department of Prosthodontics. Address correspondence and reprint requests to Dr Zhao: Department of Prosthodontics, School of Stomatology, Fourth Military Medical University, Changle West Rd 145, Xi’an 710032, People’s Republic of China; e-mail: [email protected] © 2011 American Association of Oral and Maxillofacial Surgeons
0278-2391/11/6904-0045$36.00/0 doi:10.1016/j.joms.2010.05.023
phologies or disfigurements depends, for its success, on precise knowledge of the facial norms of the patients’ racial/ethnic groups.2 There have been several studies discussing racial/ethnic variations among facial parameters, and features distinguishing various races/ethnic groups have been discovered.2-9 For Chinese persons, esthetic facial proportions and neoclassical facial canons have been reported.10,11 However, comprehensive anthropometric data of Chinese faces are still insufficient. In recent years 3-dimensional (3D) technologies have been introduced into anthropometry. These techniques are less invasive, allow images to be archived, eliminate surface pressure from an apparatus, and avoid measurement errors that occur with 2-dimensional (2D) representations of 3D surfaces. Some reports of the application of 3D measurements in craniofacial anthropometry exist. Several types of apparatus were used, including 3D computed tomography,12,13 laser scanning,14,15 and 3D stereophotogrammetry.16-18 In these studies advanced methods were used and reliable results were obtained. How-
1195
1196 ever, most of their results were presented in the form of linear distances, angles, and ratios, which were not different from traditional measurements. These results provide sufficient but fragmented information; they do not simultaneously take the whole facial structure into account. Therefore they cannot explain the complex geometry of facial structures or definitely clarify the features of facial shape.19,20 As a result, clinicians— overwhelmed with numbers and indices— had difficulty in applying such information clinically. In this context there is an urgent need to improve the 3D methods of facial anthropometry and to provide structured and intelligible information for clinicians. This study focused on the anthropometric analysis of Chinese faces using a new 3D method. The aims of the study were 1) to provide a novel method of 3D analysis of the face, 2) to establish 3D facial norms of Chinese persons, 3) to delineate the differences between the faces of men and women, and 4) to exemplify the application of 3D facial norms.
Materials and Methods ESTABLISHMENT OF NORMATIVE 3D FACIAL MODELS
The study group consisted of 100 subjects (50 men and 50 women) who belong to the largest Chinese ethnic group (Han) and who reside in central China. The ages of the men ranged from 22 to 27 years, and the ages of the women ranged from 20 to 27 years. All subjects had no obvious anomalies in the facial region. Informed consent was obtained from all subjects after institutional review board approval was obtained. A 3D stereophotogrammetry system (3DSS-II; Shanghai Digital Manufacturing, Shanghai, China) was used to take 3D images of subjects. The system is based on a structured light design and on the mathematical principle of triangulation. The same principle was used in the apparatuses described by Weinberg et al16 and Schwenzer-Zimmerer et al,21 and the precision and accuracy of such apparatuses for use in craniofacial anthropometry were reported in those studies. The 3DSS-II system can record the surface of a subject as well as the color information within 3 seconds. The spatial accuracy of this apparatus is 0.03 mm (as reported by the manufacturer). Geomagic Studio software, version 10.0 (Geomagic, Research Triangle Park, NC), was used to process and analyze the 3D data obtained in this study. It is reverseengineering software that has comprehensive point and polygon processing tools, as well as powerful surface manipulation capabilities. All subjects were scanned by the 3DSS-II system to collect data about their facial surface. During the data
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
capture process, the subject sat fixed with his or her head in a natural head position, which has been shown to be clinically reproducible.18,22,23 The subject was also instructed to keep his or her jaw in a relaxed position and eyes level with the horizontal line. The position of the 3DSS-II system was adjusted to the proper height and distance so that the face was clearly displayed on the screen. To ensure that all areas of the face were scanned, 3 images were taken for each subject: from the center, from 30° to the left, and from 30° to the right. The subject remained in the same position for all 3 captures while the camera was moved around him or her. Each scanned image was in the form of a point cloud consisting of about 300,000 points. The 3 original images of each subject were imported into Geomagic Studio 10.0, which set the default units as millimeters. After elimination of noise data, the 3 images were registered according to the same areas they contained, and then an integrated image was obtained. Subsequently, the image in the form of a point cloud was wrapped into a final image of the subject’s face, consisting of approximately 500,000 triangular patches. Facial images of all subjects were collected by 1 expert following the same protocol. However, the position of each subject relative to the machine was not the same during data collection, so his or her facial images could not be located exactly in the same coordinate system. To address this issue, a consistent coordinate system was needed to align all the images. The coordinate system was defined by the horizontal plane, sagittal plane, and coronal plane. The 3 reference planes were established according to 4
FIGURE 1. Horizontal plane passing through N and parallel to Camper’s plane rotated 7.5° upward. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
DONG ET AL
1197
FIGURE 2. Three reference planes: horizontal plane, sagittal plane, and coronal plane. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
landmarks: nasion (N), right tragion [T(r)], left tragion [T(l)], and right nasal alar [Al(r)].6,22 First, the Camper’s plane, which was determined by Al(r), T(r), and T(l), was rotated 7.5° upward on the axis formed by T(r) and T(l). The rotated plane was almost parallel to the true horizontal,22 so the horizontal plane was defined as the plane passing through N and parallel to the rotated plane (Fig 1). The sagittal plane was defined as the plane passing through N and perpendicular to the horizontal plane. Lastly, the coronal plane was defined as the plane passing through N and per-
FIGURE 4. (A) Eleven midline facial landmarks and (B) ten pairs of bilateral landmarks. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
FIGURE 3. Defined coordinate system. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
pendicular to the horizontal plane as well as the sagittal plane (Fig 2). In the coordinate system, N was designated as the origin point. The x-axis was the intersection of the horizontal plane and the coronal plane with the direction from right to left. The y-axis was the intersection of the horizontal plane and the sagittal plane with the direction from back to front. The z-axis was the intersection of the sagittal plane
1198
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
Table 1. MEAN COORDINATES OF MALE AND FEMALE FACIAL LANDMARKS
Coordinate (mm) Male (n ⫽ 50) Landmarks Midline landmarks Tr G N Prn Sn Ls Sto Li Sl Pg Gn Bilateral landmarks* Ft Ex En Zy Ck Al Ac Ch Go T
Female (n ⫽ 50)
x
y
z
x
y
z
0 0 0 0 0 0 0 0 0 0 0
15.49 ⫺3.03 0 ⫺21.66 ⫺9.64 ⫺12.65 ⫺6.59 ⫺10.06 ⫺2.68 ⫺2.85 5.64
69.25 19.79 0 ⫺40.72 ⫺53.06 ⫺67.95 ⫺76.9 ⫺87.74 ⫺93.52 ⫺111.67 ⫺123.40
0 0 0 0 0 0 0 0 0 0 0
12.33 ⫺1.87 0 ⫺19.88 ⫺7.79 ⫺10.35 ⫺4.65 ⫺7.86 ⫺1.94 ⫺2.11 6.46
66.31 18.06 0 ⫺38.75 ⫺50.42 ⫺63.82 ⫺72.35 ⫺82.98 ⫺90.19 ⫺105.17 ⫺114.72
⫺56.74 ⫺50.92 ⫺18.43 ⫺62.49 ⫺40.63 ⫺19.96 ⫺20.86 ⫺26.41 ⫺58.47 ⫺77.80
18.38 19.54 9.17 19.88 4.84 ⫺5.52 3.71 4.17 65.14 78.09
30.12 ⫺6.76 ⫺7.54 ⫺22.24 ⫺41.28 ⫺42.49 ⫺44.55 ⫺76.43 ⫺84.82 ⫺29.22
⫺52.70 ⫺47.33 ⫺17.65 ⫺59.87 ⫺38.75 ⫺18.05 ⫺19.22 ⫺24.99 ⫺54.33 ⫺73.16
16.57 17.48 8.14 18.11 3.75 ⫺4.05 2.73 3.66 60.99 72.89
27.65 ⫺6.08 ⫺7.11 ⫺21.18 ⫺39.66 ⫺40.99 ⫺42.87 ⫺71.49 ⫺79.86 ⫺26.92
*Because the coordinate values of bilateral landmarks were averaged in each subject, only the coordinates of the right landmarks are listed. The y and z coordinates of the left landmarks were equal to their counterparts, and the x coordinates of the left landmarks were the opposite of their counterparts. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
and the coronal plane with the direction from bottom to top (Fig 3). A total of 31 facial landmarks were considered in this study. Eleven of them were along the midline, namely, trichion (Tr), glabella (G), nasion (N), pronasale (Prn), subnasale (Sn), labiale superius (Ls), stomion (Sto), labiale inferius (Li), sublabiale (Sl), pogonion (Pg), and gnathion (Gn). The others were 10 pairs of bilateral landmarks [right and left side noted as (r) and (l)], namely, frontotemporale (Ft), exocanthion (Ex), endocanthion (En), zygomatic point (Zy), cheek point (Ck), nasal alar (Al), nasal alar crest (Ac), cheilion (Ch), gonion (Go), and tragion (T) (Fig 4). All of the landmarks are easy to locate and could address several anatomic, clinical, and practical problems. The definitions and assessments of the landmarks were described by Farkas,24 except for N and Ck, which were presented by Ferrario et al.25 In each subject the points N, Al(r), T(r), and T(l), which were used as reference points for the coordinate system, were assigned on the facial image before the alignment of the coordinates, and the other points were assigned after the alignment. The next step was collection of 3D coordinate values of the landmarks. Coordinate values were obtained by Geomagic Studio
10.0 and stored in a Microsoft Excel file (Microsoft, Redmond, WA). Because multiple trials and averaging of the closest 2 out of 3 values would increase reliability,26 this protocol was adopted in data collection of all the landmarks. To avoid the influence of facial asymmetry, coordinate values of bilateral landmarks were averaged in each subject. Mean coordinate values for each landmark were calculated within genders, and, according to these, Geomagic Studio 10.0 created corresponding points in the coordinate system. Then, on the basis of these points, the software built 3D models for both genders. COMPARISON BETWEEN FACIAL MODELS OF MEN AND WOMEN
To evaluate sexual dimorphism, the models of men and women were superimposed, and the absolute differences between corresponding points were calculated. To assess the shape differences independent of size, size normalizations were performed. First, the ratios of the male model to the female model on the x-axis, y-axis, and z-axis were calculated separately. Then, the male model and female model were diminished and enlarged, respectively, according to the square roots of the ratios in 3 dimensions to match
1199
DONG ET AL
FIGURE 5. Points based on mean coordinates of facial landmarks of men and linear 3D male model in (A) frontal view, (B) right lateral view, (C) basal view, and (D) 45° right lateral view (right half of model only). Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
each other. Finally, the relative differences for each landmark were obtained. APPLICATION OF NORMATIVE 3D FACIAL MODELS IN EVALUATING FACIAL DEFORMITY
A 24-year-old Chinese man was used to exemplify the application of the normative models. His facial
model was obtained and aligned in the coordinate system by use of the method described previously, and the 3D coordinate values of his facial landmarks were collected. Because size is not an essential factor in evaluating facial deformity, size normalization was performed to eliminate the impact of size. At first, both on the subject’s facial model and the normative
1200
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
male model, the distance of T(r) to T(l) on the x-axis, Prn to T on the y-axis, and Tr to Gn on the z-axis were obtained. Then, the ratios of the subject’s facial model to the normative male model in 3 dimensions were calculated, and the normative male model was reduced or enlarged to match the subject’s facial image. The subject’s facial image was superimposed on the modified normative male model. The differences for each landmark between the 2 models were obtained, and the facial deformity of the subject was evaluated.
Results ESTABLISHMENT OF NORMATIVE 3D FACIAL MODELS
The mean coordinates of the landmarks are summarized in Table 1. The points representing the landmarks were created. Linear 3D models were constructed for both genders and saved in the format of Initial Graphics Exchange Specification (IGES). On screen, the models could be rotated and displayed in any view. Figure 5 shows the linear 3D models of the men in 4 views. Then, by converting the linear models, polyhedron models were established and saved in stereolithography (STL) format. Traditional linear and angular measurements could be obtained by measuring these models (Fig 6). COMPARISON BETWEEN FACIAL MODELS OF MEN AND WOMEN
Figure 7 shows the linear 3D models of men and women superimposed. Absolute differences between the men and women for each landmark are summarized in Table 2. As expected, the female model was smaller than the male model in both the lateral and vertical directions, and it was also less developed in the anteroposterior direction. The largest differences in the x-axis and y-axis directions pertained to T, and the largest differences in the z-axis direction pertained to Gn, which also had the largest difference in vector. The ratios of the male model to the female model were 1.063 on the x-axis, 1.077 on the y-axis, and 1.064 on the z-axis. The male model was diminished according to the square roots of these ratios, and the female model was enlarged in corresponding dimensions. The 2 normalized models are superimposed in Figure 8, and the relative differences for each landmark are listed in Table 2. In general, all vectors except that of Tr were less than 2 mm. In the x-axis direction, the difference in Zy was 1.17 mm, and the differences for all other landmarks were less than 1 mm. In the y-axis direction, the largest difference appeared in Tr, and other major differences were distributed in the lower third
FIGURE 6. Polyhedron 3D male model in 45° right lateral view. Linear and angular measurements can be obtained by measuring such a model with software. Only the face width [Zy(r)-Zy(l)] and the nasofrontal angle (G-N-Prn) are shown in this figure. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
of the face, including Sn, Ls, Sto, Li, and Gn. In the z-axis direction, Sl had a discrepancy of more than 2 mm, and Tr, Gn, Al, and Ac had discrepancies of a little more than 1 mm. EVALUATION OF SUBJECT’S FACIAL DEFORMITY
Figure 9 shows the subject’s facial model in the frontal view and right lateral view. The ratios of the subject’s facial model to the normative male model were 0.991 on the x-axis, 1.084 on the y-axis, and 1.089 on the z-axis. The different ratios in the 3 dimensions indicated that the subject’s face was relatively larger than the face of the average man in the vertical and anteroposterior directions. The normative male model was rescaled based on these ratios, and Figure 10 shows the subject’s facial model superimposed on it. The differences for each landmark between the 2 models are summarized in Table 3. The differences in x, y, and z coordinates indicate the 3D spatial position of the subject’s landmarks relative to the reference, and the differences in vectors are the absolute distance between the subject’s landmark and the reference landmark. These
1201
DONG ET AL
FIGURE 7. Linear 3D models of men and women superimposed in (A) frontal view, (B) right lateral view, and (C) basal view. Blue lines represent male data, and red lines represent female data. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
values express the theoretic movement of each landmark to gain a relatively “normal” appearance. The values of Go(l) and Go(r) on the x-axis were 8.56 and 9.38 mm, which indicated that the subject had a wider mandible than average. The values of Ch(r) and Ch(l) on the x-axis were 0.36 and 5.56 mm; therefore, the subject’s bilateral cheilions were not symmetrical, and the left one was in an abnormal position. On the y-axis, the values of Sto, Li, Sl, Pg, Gn, Ch(r), and Ch(l) were larger than 5 mm, so the lower third of the subject’s face was greatly uneven. In the lateral view, the subject had more protrusive lips and a more retrusive jaw than average. On the z-axis, the values of Prn, Sn, Ls, Sto, and Pg were larger than 3
mm. These findings indicated that the middle third of the subject’s face was relatively shorter and the lower third was longer. The vectors Sto, Li, Sl, Pg, Gn, Ch(r), Ch(l), Go(l), and Go(r) had values larger than 5 mm. Therefore it could be inferred that the main deformities of the subject’s face were located in the lower third.
Discussion Analysis of the face is the first step in the evaluation of patients who present for plastic or reconstructive procedures of the face. A normal facial model helps in formulating the goals and desired outcomes of the proposed surgical procedure. It is a well-established
1202
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
Table 2. ABSOLUTE DIFFERENCES AND RELATIVE DIFFERENCES FOR EACH LANDMARK BETWEEN MALE AND FEMALE MODELS
Absolute Difference (mm) Landmarks Midline landmarks Tr G N Prn Sn Ls Sto Li Sl Pg Gn Bilateral landmarks Ft Ex En Zy Ck Al Ac Ch Go T
Relative Difference (mm)
x
y
z
Vector*
x
y
z
Vector*
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
3.16 1.16 0.00 1.78 1.85 2.30 1.94 2.20 0.74 0.74 0.82
2.94 1.73 0.00 1.97 2.64 4.13 4.55 4.76 3.33 6.50 8.68
4.32 2.08 0.00 2.66 3.22 4.73 4.95 5.24 3.41 6.54 8.72
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
2.12 0.98 0.00 0.23 1.20 1.44 1.52 1.53 0.59 0.56 1.27
1.27 0.55 0.00 0.50 0.58 0.03 0.09 0.55 2.38 0.24 1.27
2.47 1.12 0.00 0.55 1.33 1.44 1.52 1.63 2.45 0.61 1.80
4.04 3.59 0.78 2.62 1.88 1.91 1.64 1.42 4.14 4.64
1.81 2.06 1.03 1.77 1.09 1.47 0.98 0.51 4.15 5.20
2.47 0.68 0.43 1.06 1.62 1.50 1.68 4.94 4.96 2.30
5.07 4.19 1.36 3.33 2.71 2.84 2.54 5.17 7.68 7.34
0.68 0.57 0.33 1.14 0.56 0.74 0.41 0.16 0.67 0.00
0.51 0.68 0.39 0.35 0.77 1.11 0.74 0.22 0.55 0.43
0.67 0.28 0.03 0.29 0.90 1.09 1.03 0.34 0.16 0.55
1.08 0.93 0.51 1.23 1.31 1.73 1.33 0.43 0.88 0.70
*Defined as the straight-line distance between corresponding points. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
fact that human faces differ from one another based on race and ethnicity.2-9 Therefore it is important to acquire anthropometric data of different races and ethnicities. Although facial analyses are well established for white persons,8 African Americans,7,9 and Turks,5 only a limited number of Chinese facial analysis studies exist.10,11,27 Furthermore, some of the studies focused only on the esthetic facial proportions10 and neoclassical facial canons,11 and others placed emphasis on the nasal area.27 In this situation more detailed anthropometric data of Chinese faces are needed. Facial anthropometry has been conducted by use of several methods: direct anthropometry, 2D photogrammetry, and 3D methods such as 3D computed tomography, laser scanning, and 3D stereophotogrammetry. The 3DSS-II system used in this study is a 3D stereophotogrammetry system that obtains 3D images very rapidly with a no-touch technique. The images obtained permit detailed measurements while relieving the subjects of long periods of assessment. In addition, the images can be stored and retrieved for repeated measurement and subsequent analysis. The no-touch measurement eliminates the need for direct contact, so it has no effect on surface pressure, which can influence the reliability of assessment.
Anthropometry by a 3D method has several advantages over direct anthropometry, as well as over 2D photogrammetry, and had been adopted by many researchers in their studies. However, most of the studies still focused on fragmented measurements such as linear distances, angles, and ratios, which cannot provide a global evaluation of the facial structure. In addition, clinicians who need to refer to such results must search for information in numerous tables, making it difficult for these results to be applied in clinical practice. In this study 3D landmark coordinates in each subject were collected, containing more elements and providing more information than 2D values. Linear 3D facial models in IGES format were built, based on the mean landmark coordinates in each gender. These visualized and informative models, which represent the normal facial structures of Chinese young adults, would considerably facilitate the comprehension of the facial morphology. The linear 3D models were converted to polyhedron models in STL format. Traditional parameters such as linear distances and angles could be obtained from these models by software. Both IGES and STL are universal 3D formats and can be opened by any commonly used 3D software program. The 3D models built in the study would have worldwide usage. More impor-
1203
DONG ET AL
FIGURE 8. Normalized models of men and women superimposed in (A) frontal view, (B) right lateral view, and (C) basal view. Blue lines represent male data, and red lines represent female data. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
tantly, virtual models in STL format can be imported into many rapid prototyping machines and transformed into physical models. If the surgeons who need information on Chinese facial morphology do not pursue the practice of simulation images, such physical models could provide actual stereo guidance for them. Assessment of sexual dimorphism is an essential component of anthropometry, and many researchers perform it by comparing the linear distances and angles. The conventional approaches can report the differences in parameters between male and female faces. However, because the sparse parameters do not consider the face as a whole, they cannot make a
comprehensive evaluation of the facial structure. In this study the 3D models of men and women were superimposed, and the differences for each landmark were obtained. Then, the global differences between young Chinese male and female faces were immediately evident, and the displacements of local features were displayed. The previously described process described the sexual dimorphism of the face concerning both size and shape. Indeed, the size-independent facial shape should be the subject of more attention,22 but conventional measurements are not particularly suited to shape assessment.28 Because the size of a 3D model can be easily scaled to any dimension by software,
1204
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
models. By use of this method, not only were the 3D relationships between the subject’s face and the standard facial model clearly presented on the screen, but also the deformity in 3 dimensions was displayed numerically. Therefore the surgeons who treated this subject would obtain more informative and exact instructions for the operative plan. Similarly, if the subject’s postoperative images were used, the surgical outcome could be easily evaluated.
FIGURE 9. (A) Frontal view and (B) right lateral view of subject’s facial model. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
this study normalized the facial models based on the ratios of size. By superimposing the normalized models and calculating the relative difference of each landmark, the differences of shape independent of size were explicitly delineated. A subject with facial deformity was used in the study to exemplify the application of the normative
FIGURE 10. Subject’s facial model and rescaled normative male model superimposed in (A) frontal view and (B) right lateral view. The subject’s facial model is translucent to display its relationship with the rescaled normative male model, which is in the form of blue lines. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
1205
DONG ET AL
Table 3. DIFFERENCES FOR EACH LANDMARK BETWEEN SUBJECT’S FACE AND MODIFIED MALE MODEL
Difference (mm) Landmarks Midline landmarks Tr G N Prn Sn Ls Sto Li Sl Pg Gn Bilateral landmarks Ft(r) Ft(l) Ex(r) Ex(l) En(r) En(l) Zy(r) Zy(l) Ck(r) Ck(l) Al(r) Al(l) Ac(r) Ac(l) Ch(r) Ch(l) Go(r) Go(l) T(r) T(l)
x
y
z
Vector*
3.28 0.41 0 0.48 0.69 0.12 0.31 1.48 0.42 0.22 0.04
1.81 1.87 0 2.28 2.72 2.46 5.15 9.65 6.45 6.21 7.27
0.89 1.70 0 3.86 3.72 3.16 5.34 2.17 1.96 3.31 0.84
3.82 2.56 0 4.51 4.67 3.78 7.43 10.04 6.76 7.05 7.31
1.14 0.29 1.85 1.91 1.54 0.27 0.84 1.68 2.09 2.68 1.97 0.09 2.10 1.22 0.36 5.56 8.56 9.38 0.45 0.45
2.79 2.95 3.62 3.38 3.68 3.04 3.75 2.66 0.11 1.13 3.53 2.84 0.34 0.71 7.87 7.76 1.71 1.61 2.28 2.28
1.24 1.47 0.48 0.61 0.41 0.99 1.82 0.84 1.29 1.39 1.50 0.74 1.80 1.28 0.24 1.53 0.41 1.43 0.37 0.37
3.26 3.31 4.45 4.13 4.01 3.26 4.74 3.26 2.47 3.23 4.31 2.94 2.79 1.91 7.89 9.72 8.74 9.63 2.34 2.34
*Defined as the straight-line distance between corresponding points. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
This study was limited to analysis of young Chinese adults only, and the number of subjects is small relative to the total Chinese population. In addition, all the subjects belong to the largest Chinese ethnic group (Han) and reside in central China. The Chinese population contains individuals of various ethnic groups, and there may be multiple differences between persons from different regions. Thus the norms provided by this study would not be universally applicable to all Chinese persons. For this reason, further research with larger samples of multiple origins should be performed. Indeed, in the eyes of many researchers, norms should never be used as rigid goals to impose therapies but only as guides for the best results.20 In conclusion, this study broadened the understanding of and bolstered the need for facial analysis.
It described a novel 3D method of anthropometric facial analysis, which was a breakthrough and would be prospective in anthropometry and the analysis of deformity. Three-dimensional models representing the facial shape of young Chinese adults were built, and sexual dimorphism was investigated. Surgeons could easily comprehend and apply these results in clinical practice.
References 1. Koury ME, Epker BN: Maxillofacial esthetics: Anthropometrics of the maxillofacial region. J Oral Maxillofac Surg 50:806, 1992 2. Farkas LG, Katic MJ, Forrest CR: Comparison of craniofacial measurements of young adult African-American and North American white males and females. Ann Plast Surg 59:692, 2007 3. Farkas LG, Katic MJ, Forrest CR, et al: International anthropometric study of facial morphology in various ethnic groups/ races. J Craniofac Surg 16:615, 2005 4. Choe KS, Sclafani AP, Litner JA, et al: The Korean American woman’s face: Anthropometric measurements and quantitative analysis of facial aesthetics. Arch Facial Plast Surg 6:244, 2004 5. Ozdemir ST, Sigirli D, Ercan I, et al: Photographic facial soft tissue analysis of healthy Turkish young adults: Anthropometric measurements. Aesthetic Plast Surg 33:175, 2009 6. Baik HS, Jeon JM, Lee HJ: Facial soft-tissue analysis of Korean adults with normal occlusion using a 3-dimensional laser scanner. Am J Orthod Dentofac Orthop 131:759, 2007 7. Porter JP: The average African American male face: An anthropometric analysis. Arch Facial Plast Surg 6:78, 2004 8. Farkas LG, Hreczko TA, Kolar JC, et al: Vertical and horizontal proportions of the face in young adult North American Caucasians: Revision of neoclassical canons. Plast Reconstr Surg 75: 328, 1985 9. Jeffries JM III, DiBernardo B, Rauscher GE: Computer analysis of the African-American face. Ann Plast Surg 34:318, 1995 10. Sim RS, Smith JD, Chan AS: Comparison of the aesthetic facial proportions of southern Chinese and white women. Arch Facial Plast Surg 2:113, 2000 11. Wang D, Qian G, Zhang M, et al: Differences in horizontal, neoclassical facial canons in Chinese (Han) and North American Caucasian populations. Aesthetic Plast Surg 21:265, 1997 12. Olszewski R, Zech F, Cosnard G, et al: Three-dimensional computed tomography cephalometric craniofacial analysis: Experimental validation in vitro. Int J Oral Maxillofac Surg 36:828, 2007 13. Vezzetti E, Calignano F, Moos S: Computer-aided morphological analysis for maxillo-facial diagnostic: A preliminary study. J Plast Reconstr Aesthet Surg 63:218, 2008 14. Bush K, Antonyshyn O: Three-dimensional facial anthropometry using a laser surface scanner: Validation of the technique. Plast Reconstr Surg 98:226, 1996 15. Kovacs L, Zimmermann A, Brockmann G, et al: Three-dimensional recording of the human face with a 3D laser scanner. J Plast Reconstr Aesthet Surg 59:1193, 2006 16. Weinberg SM, Scott NM, Neiswanger K, et al: Digital threedimensional photogrammetry: Evaluation of anthropometric precision and accuracy using a Genex 3D camera system. Cleft Palate Craniofac J 41:507, 2004 17. Ayoub AF, Siebert P, Moos KF, et al: A vision-based threedimensional capture system for maxillofacial assessment and surgical planning. Br J Oral Maxillofac Surg 36:353, 1998 18. Ferrario VF, Sforza C, Poggio CE, et al: Preliminary evaluation of an electromagnetic three-dimensional digitizer in facial anthropometry. Cleft Palate Craniofac J 35:9, 1998 19. Hennessy RJ, McLearie S, Kinsella A, et al: Facial shape and asymmetry by three-dimensional laser surface scanning covary with cognition in a sexually dimorphic manner. J Neuropsychiatry Clin Neurosci 18:73, 2006
1206 20. Ferrario VF, Sforza C, Dalloca LL, et al: Assessment of facial form modifications in orthodontics: Proposal of a modified computerized mesh diagram analysis. Am J Orthod Dentofacial Orthop 109:263, 1996 21. Schwenzer-Zimmerer K, Haberstok J, Kovacs L, et al: 3D surface measurement for medical application—Technical comparison of two established industrial surface scanning systems. J Med Syst 32:59, 2008 22. Ferrario VF, Sforza C, Schmitz JH, et al: A three-dimensional computerized mesh diagram analysis and its application in soft tissue facial morphometry. Am J Orthod Dentofacial Orthop 114:404, 1998 23. Kau CH, Richmond S, Savio C, et al: Measuring adult facial morphology in three dimensions. Angle Orthod 76:773, 2006
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES 24. Farkas LG: Anthropometry of the Head and Face. New York, Raven Press, 1994 25. Ferrario VF, Sforza C, Serrao G, et al: Soft tissue facial growth and development as assessed by the three-dimensional computerized mesh diagram analysis. Am J Orthod Dentofacial Orthop 116:215, 1999 26. Ward RE, Jamison PL: Measurement precision and reliability in craniofacial anthropometry: Implications and suggestions for clinical applications. J Craniofac Genet Dev Biol 11:156, 1991 27. Zhang XT, Wang SK, Zhang W, et al: Measurement and study of the nose and face and their correlations in the young adult of Han nationality. Plast Reconstr Surg 85:532, 1990 28. Moyers RE, Bookstein FL: The inappropriateness of conventional cephalometrics. Am J Orthod 75:599, 1979
Three-Dimensional Anthropometric Analysis of Chinese Faces and Its Application in Evaluating Facial Deformity Yan Dong, PhD,* Yimin Zhao, PhD,† Shizhu Bai, PhD,‡ Guofeng Wu, PhD,§ Lin Zhou, PhD,储 and Bo Wang, PhD¶ Purpose: The aims of this study are to introduce a novel method of 3-dimensional (3D) analysis of the
face and to provide normative data of the Chinese face for surgeons. Materials and Methods: Fifty men and 50 women were recruited, and a 3D stereophotogrammetry system was used to acquire their facial image data. For each subject, the image was aligned to a unified coordinate system, and coordinate values of 31 facial landmarks were collected. Mean values for each landmark were calculated within genders, and 3D models of both genders were constructed based on the mean values. Subsequently, to evaluate the sexual dimorphism, the models were superimposed. Then, to delineate the shape differences independent of size, the models were normalized and superimposed again. The application of the 3D models was also exemplified by analysis of a subject’s facial deformity. Results: Linear and polyhedron 3D models representing the facial shape were built for both genders. The superimposed models and the absolute differences between each landmark in both genders illustrated the sexual dimorphism of the Chinese face, and the normalized models and relative difference for each landmark also delineated the shape differences independent of size. In addition, a subject’s facial deformity was evaluated by referring to the normative 3D facial models. Conclusions: This study describes a new 3D analysis method for facial morphology. Three-dimensional models representing the facial shape of Chinese subjects were built, and sexual dimorphism was investigated. These would provide useful guidance for facial anthropometry and plastic surgeons in clinical practice. © 2011 American Association of Oral and Maxillofacial Surgeons J Oral Maxillofac Surg 69:1195-1206, 2011 Anthropometric studies are an integral part of maxillofacial surgery. Normative data of facial measurements are fundamental to the objective analysis of facial deformity, plan of operation, prediction of final treatment results, and evaluation of surgical outcome.1 Surgical correction of maxillofacial dysmor-
Received from the School of Stomatology, Fourth Military Medical University, Xi’an, People’s Republic of China. *Department of Prosthodontics. †Department of Prosthodontics. ‡Department of Prosthodontics. §Department of Prosthodontics. 储Department of Prosthodontics. ¶Department of Prosthodontics. Address correspondence and reprint requests to Dr Zhao: Department of Prosthodontics, School of Stomatology, Fourth Military Medical University, Changle West Rd 145, Xi’an 710032, People’s Republic of China; e-mail: [email protected] © 2011 American Association of Oral and Maxillofacial Surgeons
0278-2391/11/6904-0045$36.00/0 doi:10.1016/j.joms.2010.05.023
phologies or disfigurements depends, for its success, on precise knowledge of the facial norms of the patients’ racial/ethnic groups.2 There have been several studies discussing racial/ethnic variations among facial parameters, and features distinguishing various races/ethnic groups have been discovered.2-9 For Chinese persons, esthetic facial proportions and neoclassical facial canons have been reported.10,11 However, comprehensive anthropometric data of Chinese faces are still insufficient. In recent years 3-dimensional (3D) technologies have been introduced into anthropometry. These techniques are less invasive, allow images to be archived, eliminate surface pressure from an apparatus, and avoid measurement errors that occur with 2-dimensional (2D) representations of 3D surfaces. Some reports of the application of 3D measurements in craniofacial anthropometry exist. Several types of apparatus were used, including 3D computed tomography,12,13 laser scanning,14,15 and 3D stereophotogrammetry.16-18 In these studies advanced methods were used and reliable results were obtained. How-
1195
1196 ever, most of their results were presented in the form of linear distances, angles, and ratios, which were not different from traditional measurements. These results provide sufficient but fragmented information; they do not simultaneously take the whole facial structure into account. Therefore they cannot explain the complex geometry of facial structures or definitely clarify the features of facial shape.19,20 As a result, clinicians— overwhelmed with numbers and indices— had difficulty in applying such information clinically. In this context there is an urgent need to improve the 3D methods of facial anthropometry and to provide structured and intelligible information for clinicians. This study focused on the anthropometric analysis of Chinese faces using a new 3D method. The aims of the study were 1) to provide a novel method of 3D analysis of the face, 2) to establish 3D facial norms of Chinese persons, 3) to delineate the differences between the faces of men and women, and 4) to exemplify the application of 3D facial norms.
Materials and Methods ESTABLISHMENT OF NORMATIVE 3D FACIAL MODELS
The study group consisted of 100 subjects (50 men and 50 women) who belong to the largest Chinese ethnic group (Han) and who reside in central China. The ages of the men ranged from 22 to 27 years, and the ages of the women ranged from 20 to 27 years. All subjects had no obvious anomalies in the facial region. Informed consent was obtained from all subjects after institutional review board approval was obtained. A 3D stereophotogrammetry system (3DSS-II; Shanghai Digital Manufacturing, Shanghai, China) was used to take 3D images of subjects. The system is based on a structured light design and on the mathematical principle of triangulation. The same principle was used in the apparatuses described by Weinberg et al16 and Schwenzer-Zimmerer et al,21 and the precision and accuracy of such apparatuses for use in craniofacial anthropometry were reported in those studies. The 3DSS-II system can record the surface of a subject as well as the color information within 3 seconds. The spatial accuracy of this apparatus is 0.03 mm (as reported by the manufacturer). Geomagic Studio software, version 10.0 (Geomagic, Research Triangle Park, NC), was used to process and analyze the 3D data obtained in this study. It is reverseengineering software that has comprehensive point and polygon processing tools, as well as powerful surface manipulation capabilities. All subjects were scanned by the 3DSS-II system to collect data about their facial surface. During the data
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
capture process, the subject sat fixed with his or her head in a natural head position, which has been shown to be clinically reproducible.18,22,23 The subject was also instructed to keep his or her jaw in a relaxed position and eyes level with the horizontal line. The position of the 3DSS-II system was adjusted to the proper height and distance so that the face was clearly displayed on the screen. To ensure that all areas of the face were scanned, 3 images were taken for each subject: from the center, from 30° to the left, and from 30° to the right. The subject remained in the same position for all 3 captures while the camera was moved around him or her. Each scanned image was in the form of a point cloud consisting of about 300,000 points. The 3 original images of each subject were imported into Geomagic Studio 10.0, which set the default units as millimeters. After elimination of noise data, the 3 images were registered according to the same areas they contained, and then an integrated image was obtained. Subsequently, the image in the form of a point cloud was wrapped into a final image of the subject’s face, consisting of approximately 500,000 triangular patches. Facial images of all subjects were collected by 1 expert following the same protocol. However, the position of each subject relative to the machine was not the same during data collection, so his or her facial images could not be located exactly in the same coordinate system. To address this issue, a consistent coordinate system was needed to align all the images. The coordinate system was defined by the horizontal plane, sagittal plane, and coronal plane. The 3 reference planes were established according to 4
FIGURE 1. Horizontal plane passing through N and parallel to Camper’s plane rotated 7.5° upward. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
DONG ET AL
1197
FIGURE 2. Three reference planes: horizontal plane, sagittal plane, and coronal plane. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
landmarks: nasion (N), right tragion [T(r)], left tragion [T(l)], and right nasal alar [Al(r)].6,22 First, the Camper’s plane, which was determined by Al(r), T(r), and T(l), was rotated 7.5° upward on the axis formed by T(r) and T(l). The rotated plane was almost parallel to the true horizontal,22 so the horizontal plane was defined as the plane passing through N and parallel to the rotated plane (Fig 1). The sagittal plane was defined as the plane passing through N and perpendicular to the horizontal plane. Lastly, the coronal plane was defined as the plane passing through N and per-
FIGURE 4. (A) Eleven midline facial landmarks and (B) ten pairs of bilateral landmarks. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
FIGURE 3. Defined coordinate system. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
pendicular to the horizontal plane as well as the sagittal plane (Fig 2). In the coordinate system, N was designated as the origin point. The x-axis was the intersection of the horizontal plane and the coronal plane with the direction from right to left. The y-axis was the intersection of the horizontal plane and the sagittal plane with the direction from back to front. The z-axis was the intersection of the sagittal plane
1198
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
Table 1. MEAN COORDINATES OF MALE AND FEMALE FACIAL LANDMARKS
Coordinate (mm) Male (n ⫽ 50) Landmarks Midline landmarks Tr G N Prn Sn Ls Sto Li Sl Pg Gn Bilateral landmarks* Ft Ex En Zy Ck Al Ac Ch Go T
Female (n ⫽ 50)
x
y
z
x
y
z
0 0 0 0 0 0 0 0 0 0 0
15.49 ⫺3.03 0 ⫺21.66 ⫺9.64 ⫺12.65 ⫺6.59 ⫺10.06 ⫺2.68 ⫺2.85 5.64
69.25 19.79 0 ⫺40.72 ⫺53.06 ⫺67.95 ⫺76.9 ⫺87.74 ⫺93.52 ⫺111.67 ⫺123.40
0 0 0 0 0 0 0 0 0 0 0
12.33 ⫺1.87 0 ⫺19.88 ⫺7.79 ⫺10.35 ⫺4.65 ⫺7.86 ⫺1.94 ⫺2.11 6.46
66.31 18.06 0 ⫺38.75 ⫺50.42 ⫺63.82 ⫺72.35 ⫺82.98 ⫺90.19 ⫺105.17 ⫺114.72
⫺56.74 ⫺50.92 ⫺18.43 ⫺62.49 ⫺40.63 ⫺19.96 ⫺20.86 ⫺26.41 ⫺58.47 ⫺77.80
18.38 19.54 9.17 19.88 4.84 ⫺5.52 3.71 4.17 65.14 78.09
30.12 ⫺6.76 ⫺7.54 ⫺22.24 ⫺41.28 ⫺42.49 ⫺44.55 ⫺76.43 ⫺84.82 ⫺29.22
⫺52.70 ⫺47.33 ⫺17.65 ⫺59.87 ⫺38.75 ⫺18.05 ⫺19.22 ⫺24.99 ⫺54.33 ⫺73.16
16.57 17.48 8.14 18.11 3.75 ⫺4.05 2.73 3.66 60.99 72.89
27.65 ⫺6.08 ⫺7.11 ⫺21.18 ⫺39.66 ⫺40.99 ⫺42.87 ⫺71.49 ⫺79.86 ⫺26.92
*Because the coordinate values of bilateral landmarks were averaged in each subject, only the coordinates of the right landmarks are listed. The y and z coordinates of the left landmarks were equal to their counterparts, and the x coordinates of the left landmarks were the opposite of their counterparts. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
and the coronal plane with the direction from bottom to top (Fig 3). A total of 31 facial landmarks were considered in this study. Eleven of them were along the midline, namely, trichion (Tr), glabella (G), nasion (N), pronasale (Prn), subnasale (Sn), labiale superius (Ls), stomion (Sto), labiale inferius (Li), sublabiale (Sl), pogonion (Pg), and gnathion (Gn). The others were 10 pairs of bilateral landmarks [right and left side noted as (r) and (l)], namely, frontotemporale (Ft), exocanthion (Ex), endocanthion (En), zygomatic point (Zy), cheek point (Ck), nasal alar (Al), nasal alar crest (Ac), cheilion (Ch), gonion (Go), and tragion (T) (Fig 4). All of the landmarks are easy to locate and could address several anatomic, clinical, and practical problems. The definitions and assessments of the landmarks were described by Farkas,24 except for N and Ck, which were presented by Ferrario et al.25 In each subject the points N, Al(r), T(r), and T(l), which were used as reference points for the coordinate system, were assigned on the facial image before the alignment of the coordinates, and the other points were assigned after the alignment. The next step was collection of 3D coordinate values of the landmarks. Coordinate values were obtained by Geomagic Studio
10.0 and stored in a Microsoft Excel file (Microsoft, Redmond, WA). Because multiple trials and averaging of the closest 2 out of 3 values would increase reliability,26 this protocol was adopted in data collection of all the landmarks. To avoid the influence of facial asymmetry, coordinate values of bilateral landmarks were averaged in each subject. Mean coordinate values for each landmark were calculated within genders, and, according to these, Geomagic Studio 10.0 created corresponding points in the coordinate system. Then, on the basis of these points, the software built 3D models for both genders. COMPARISON BETWEEN FACIAL MODELS OF MEN AND WOMEN
To evaluate sexual dimorphism, the models of men and women were superimposed, and the absolute differences between corresponding points were calculated. To assess the shape differences independent of size, size normalizations were performed. First, the ratios of the male model to the female model on the x-axis, y-axis, and z-axis were calculated separately. Then, the male model and female model were diminished and enlarged, respectively, according to the square roots of the ratios in 3 dimensions to match
1199
DONG ET AL
FIGURE 5. Points based on mean coordinates of facial landmarks of men and linear 3D male model in (A) frontal view, (B) right lateral view, (C) basal view, and (D) 45° right lateral view (right half of model only). Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
each other. Finally, the relative differences for each landmark were obtained. APPLICATION OF NORMATIVE 3D FACIAL MODELS IN EVALUATING FACIAL DEFORMITY
A 24-year-old Chinese man was used to exemplify the application of the normative models. His facial
model was obtained and aligned in the coordinate system by use of the method described previously, and the 3D coordinate values of his facial landmarks were collected. Because size is not an essential factor in evaluating facial deformity, size normalization was performed to eliminate the impact of size. At first, both on the subject’s facial model and the normative
1200
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
male model, the distance of T(r) to T(l) on the x-axis, Prn to T on the y-axis, and Tr to Gn on the z-axis were obtained. Then, the ratios of the subject’s facial model to the normative male model in 3 dimensions were calculated, and the normative male model was reduced or enlarged to match the subject’s facial image. The subject’s facial image was superimposed on the modified normative male model. The differences for each landmark between the 2 models were obtained, and the facial deformity of the subject was evaluated.
Results ESTABLISHMENT OF NORMATIVE 3D FACIAL MODELS
The mean coordinates of the landmarks are summarized in Table 1. The points representing the landmarks were created. Linear 3D models were constructed for both genders and saved in the format of Initial Graphics Exchange Specification (IGES). On screen, the models could be rotated and displayed in any view. Figure 5 shows the linear 3D models of the men in 4 views. Then, by converting the linear models, polyhedron models were established and saved in stereolithography (STL) format. Traditional linear and angular measurements could be obtained by measuring these models (Fig 6). COMPARISON BETWEEN FACIAL MODELS OF MEN AND WOMEN
Figure 7 shows the linear 3D models of men and women superimposed. Absolute differences between the men and women for each landmark are summarized in Table 2. As expected, the female model was smaller than the male model in both the lateral and vertical directions, and it was also less developed in the anteroposterior direction. The largest differences in the x-axis and y-axis directions pertained to T, and the largest differences in the z-axis direction pertained to Gn, which also had the largest difference in vector. The ratios of the male model to the female model were 1.063 on the x-axis, 1.077 on the y-axis, and 1.064 on the z-axis. The male model was diminished according to the square roots of these ratios, and the female model was enlarged in corresponding dimensions. The 2 normalized models are superimposed in Figure 8, and the relative differences for each landmark are listed in Table 2. In general, all vectors except that of Tr were less than 2 mm. In the x-axis direction, the difference in Zy was 1.17 mm, and the differences for all other landmarks were less than 1 mm. In the y-axis direction, the largest difference appeared in Tr, and other major differences were distributed in the lower third
FIGURE 6. Polyhedron 3D male model in 45° right lateral view. Linear and angular measurements can be obtained by measuring such a model with software. Only the face width [Zy(r)-Zy(l)] and the nasofrontal angle (G-N-Prn) are shown in this figure. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
of the face, including Sn, Ls, Sto, Li, and Gn. In the z-axis direction, Sl had a discrepancy of more than 2 mm, and Tr, Gn, Al, and Ac had discrepancies of a little more than 1 mm. EVALUATION OF SUBJECT’S FACIAL DEFORMITY
Figure 9 shows the subject’s facial model in the frontal view and right lateral view. The ratios of the subject’s facial model to the normative male model were 0.991 on the x-axis, 1.084 on the y-axis, and 1.089 on the z-axis. The different ratios in the 3 dimensions indicated that the subject’s face was relatively larger than the face of the average man in the vertical and anteroposterior directions. The normative male model was rescaled based on these ratios, and Figure 10 shows the subject’s facial model superimposed on it. The differences for each landmark between the 2 models are summarized in Table 3. The differences in x, y, and z coordinates indicate the 3D spatial position of the subject’s landmarks relative to the reference, and the differences in vectors are the absolute distance between the subject’s landmark and the reference landmark. These
1201
DONG ET AL
FIGURE 7. Linear 3D models of men and women superimposed in (A) frontal view, (B) right lateral view, and (C) basal view. Blue lines represent male data, and red lines represent female data. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
values express the theoretic movement of each landmark to gain a relatively “normal” appearance. The values of Go(l) and Go(r) on the x-axis were 8.56 and 9.38 mm, which indicated that the subject had a wider mandible than average. The values of Ch(r) and Ch(l) on the x-axis were 0.36 and 5.56 mm; therefore, the subject’s bilateral cheilions were not symmetrical, and the left one was in an abnormal position. On the y-axis, the values of Sto, Li, Sl, Pg, Gn, Ch(r), and Ch(l) were larger than 5 mm, so the lower third of the subject’s face was greatly uneven. In the lateral view, the subject had more protrusive lips and a more retrusive jaw than average. On the z-axis, the values of Prn, Sn, Ls, Sto, and Pg were larger than 3
mm. These findings indicated that the middle third of the subject’s face was relatively shorter and the lower third was longer. The vectors Sto, Li, Sl, Pg, Gn, Ch(r), Ch(l), Go(l), and Go(r) had values larger than 5 mm. Therefore it could be inferred that the main deformities of the subject’s face were located in the lower third.
Discussion Analysis of the face is the first step in the evaluation of patients who present for plastic or reconstructive procedures of the face. A normal facial model helps in formulating the goals and desired outcomes of the proposed surgical procedure. It is a well-established
1202
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
Table 2. ABSOLUTE DIFFERENCES AND RELATIVE DIFFERENCES FOR EACH LANDMARK BETWEEN MALE AND FEMALE MODELS
Absolute Difference (mm) Landmarks Midline landmarks Tr G N Prn Sn Ls Sto Li Sl Pg Gn Bilateral landmarks Ft Ex En Zy Ck Al Ac Ch Go T
Relative Difference (mm)
x
y
z
Vector*
x
y
z
Vector*
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
3.16 1.16 0.00 1.78 1.85 2.30 1.94 2.20 0.74 0.74 0.82
2.94 1.73 0.00 1.97 2.64 4.13 4.55 4.76 3.33 6.50 8.68
4.32 2.08 0.00 2.66 3.22 4.73 4.95 5.24 3.41 6.54 8.72
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
2.12 0.98 0.00 0.23 1.20 1.44 1.52 1.53 0.59 0.56 1.27
1.27 0.55 0.00 0.50 0.58 0.03 0.09 0.55 2.38 0.24 1.27
2.47 1.12 0.00 0.55 1.33 1.44 1.52 1.63 2.45 0.61 1.80
4.04 3.59 0.78 2.62 1.88 1.91 1.64 1.42 4.14 4.64
1.81 2.06 1.03 1.77 1.09 1.47 0.98 0.51 4.15 5.20
2.47 0.68 0.43 1.06 1.62 1.50 1.68 4.94 4.96 2.30
5.07 4.19 1.36 3.33 2.71 2.84 2.54 5.17 7.68 7.34
0.68 0.57 0.33 1.14 0.56 0.74 0.41 0.16 0.67 0.00
0.51 0.68 0.39 0.35 0.77 1.11 0.74 0.22 0.55 0.43
0.67 0.28 0.03 0.29 0.90 1.09 1.03 0.34 0.16 0.55
1.08 0.93 0.51 1.23 1.31 1.73 1.33 0.43 0.88 0.70
*Defined as the straight-line distance between corresponding points. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
fact that human faces differ from one another based on race and ethnicity.2-9 Therefore it is important to acquire anthropometric data of different races and ethnicities. Although facial analyses are well established for white persons,8 African Americans,7,9 and Turks,5 only a limited number of Chinese facial analysis studies exist.10,11,27 Furthermore, some of the studies focused only on the esthetic facial proportions10 and neoclassical facial canons,11 and others placed emphasis on the nasal area.27 In this situation more detailed anthropometric data of Chinese faces are needed. Facial anthropometry has been conducted by use of several methods: direct anthropometry, 2D photogrammetry, and 3D methods such as 3D computed tomography, laser scanning, and 3D stereophotogrammetry. The 3DSS-II system used in this study is a 3D stereophotogrammetry system that obtains 3D images very rapidly with a no-touch technique. The images obtained permit detailed measurements while relieving the subjects of long periods of assessment. In addition, the images can be stored and retrieved for repeated measurement and subsequent analysis. The no-touch measurement eliminates the need for direct contact, so it has no effect on surface pressure, which can influence the reliability of assessment.
Anthropometry by a 3D method has several advantages over direct anthropometry, as well as over 2D photogrammetry, and had been adopted by many researchers in their studies. However, most of the studies still focused on fragmented measurements such as linear distances, angles, and ratios, which cannot provide a global evaluation of the facial structure. In addition, clinicians who need to refer to such results must search for information in numerous tables, making it difficult for these results to be applied in clinical practice. In this study 3D landmark coordinates in each subject were collected, containing more elements and providing more information than 2D values. Linear 3D facial models in IGES format were built, based on the mean landmark coordinates in each gender. These visualized and informative models, which represent the normal facial structures of Chinese young adults, would considerably facilitate the comprehension of the facial morphology. The linear 3D models were converted to polyhedron models in STL format. Traditional parameters such as linear distances and angles could be obtained from these models by software. Both IGES and STL are universal 3D formats and can be opened by any commonly used 3D software program. The 3D models built in the study would have worldwide usage. More impor-
1203
DONG ET AL
FIGURE 8. Normalized models of men and women superimposed in (A) frontal view, (B) right lateral view, and (C) basal view. Blue lines represent male data, and red lines represent female data. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
tantly, virtual models in STL format can be imported into many rapid prototyping machines and transformed into physical models. If the surgeons who need information on Chinese facial morphology do not pursue the practice of simulation images, such physical models could provide actual stereo guidance for them. Assessment of sexual dimorphism is an essential component of anthropometry, and many researchers perform it by comparing the linear distances and angles. The conventional approaches can report the differences in parameters between male and female faces. However, because the sparse parameters do not consider the face as a whole, they cannot make a
comprehensive evaluation of the facial structure. In this study the 3D models of men and women were superimposed, and the differences for each landmark were obtained. Then, the global differences between young Chinese male and female faces were immediately evident, and the displacements of local features were displayed. The previously described process described the sexual dimorphism of the face concerning both size and shape. Indeed, the size-independent facial shape should be the subject of more attention,22 but conventional measurements are not particularly suited to shape assessment.28 Because the size of a 3D model can be easily scaled to any dimension by software,
1204
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES
models. By use of this method, not only were the 3D relationships between the subject’s face and the standard facial model clearly presented on the screen, but also the deformity in 3 dimensions was displayed numerically. Therefore the surgeons who treated this subject would obtain more informative and exact instructions for the operative plan. Similarly, if the subject’s postoperative images were used, the surgical outcome could be easily evaluated.
FIGURE 9. (A) Frontal view and (B) right lateral view of subject’s facial model. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
this study normalized the facial models based on the ratios of size. By superimposing the normalized models and calculating the relative difference of each landmark, the differences of shape independent of size were explicitly delineated. A subject with facial deformity was used in the study to exemplify the application of the normative
FIGURE 10. Subject’s facial model and rescaled normative male model superimposed in (A) frontal view and (B) right lateral view. The subject’s facial model is translucent to display its relationship with the rescaled normative male model, which is in the form of blue lines. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
1205
DONG ET AL
Table 3. DIFFERENCES FOR EACH LANDMARK BETWEEN SUBJECT’S FACE AND MODIFIED MALE MODEL
Difference (mm) Landmarks Midline landmarks Tr G N Prn Sn Ls Sto Li Sl Pg Gn Bilateral landmarks Ft(r) Ft(l) Ex(r) Ex(l) En(r) En(l) Zy(r) Zy(l) Ck(r) Ck(l) Al(r) Al(l) Ac(r) Ac(l) Ch(r) Ch(l) Go(r) Go(l) T(r) T(l)
x
y
z
Vector*
3.28 0.41 0 0.48 0.69 0.12 0.31 1.48 0.42 0.22 0.04
1.81 1.87 0 2.28 2.72 2.46 5.15 9.65 6.45 6.21 7.27
0.89 1.70 0 3.86 3.72 3.16 5.34 2.17 1.96 3.31 0.84
3.82 2.56 0 4.51 4.67 3.78 7.43 10.04 6.76 7.05 7.31
1.14 0.29 1.85 1.91 1.54 0.27 0.84 1.68 2.09 2.68 1.97 0.09 2.10 1.22 0.36 5.56 8.56 9.38 0.45 0.45
2.79 2.95 3.62 3.38 3.68 3.04 3.75 2.66 0.11 1.13 3.53 2.84 0.34 0.71 7.87 7.76 1.71 1.61 2.28 2.28
1.24 1.47 0.48 0.61 0.41 0.99 1.82 0.84 1.29 1.39 1.50 0.74 1.80 1.28 0.24 1.53 0.41 1.43 0.37 0.37
3.26 3.31 4.45 4.13 4.01 3.26 4.74 3.26 2.47 3.23 4.31 2.94 2.79 1.91 7.89 9.72 8.74 9.63 2.34 2.34
*Defined as the straight-line distance between corresponding points. Dong et al. Three-Dimensional Analysis of Chinese Faces. J Oral Maxillofac Surg 2011.
This study was limited to analysis of young Chinese adults only, and the number of subjects is small relative to the total Chinese population. In addition, all the subjects belong to the largest Chinese ethnic group (Han) and reside in central China. The Chinese population contains individuals of various ethnic groups, and there may be multiple differences between persons from different regions. Thus the norms provided by this study would not be universally applicable to all Chinese persons. For this reason, further research with larger samples of multiple origins should be performed. Indeed, in the eyes of many researchers, norms should never be used as rigid goals to impose therapies but only as guides for the best results.20 In conclusion, this study broadened the understanding of and bolstered the need for facial analysis.
It described a novel 3D method of anthropometric facial analysis, which was a breakthrough and would be prospective in anthropometry and the analysis of deformity. Three-dimensional models representing the facial shape of young Chinese adults were built, and sexual dimorphism was investigated. Surgeons could easily comprehend and apply these results in clinical practice.
References 1. Koury ME, Epker BN: Maxillofacial esthetics: Anthropometrics of the maxillofacial region. J Oral Maxillofac Surg 50:806, 1992 2. Farkas LG, Katic MJ, Forrest CR: Comparison of craniofacial measurements of young adult African-American and North American white males and females. Ann Plast Surg 59:692, 2007 3. Farkas LG, Katic MJ, Forrest CR, et al: International anthropometric study of facial morphology in various ethnic groups/ races. J Craniofac Surg 16:615, 2005 4. Choe KS, Sclafani AP, Litner JA, et al: The Korean American woman’s face: Anthropometric measurements and quantitative analysis of facial aesthetics. Arch Facial Plast Surg 6:244, 2004 5. Ozdemir ST, Sigirli D, Ercan I, et al: Photographic facial soft tissue analysis of healthy Turkish young adults: Anthropometric measurements. Aesthetic Plast Surg 33:175, 2009 6. Baik HS, Jeon JM, Lee HJ: Facial soft-tissue analysis of Korean adults with normal occlusion using a 3-dimensional laser scanner. Am J Orthod Dentofac Orthop 131:759, 2007 7. Porter JP: The average African American male face: An anthropometric analysis. Arch Facial Plast Surg 6:78, 2004 8. Farkas LG, Hreczko TA, Kolar JC, et al: Vertical and horizontal proportions of the face in young adult North American Caucasians: Revision of neoclassical canons. Plast Reconstr Surg 75: 328, 1985 9. Jeffries JM III, DiBernardo B, Rauscher GE: Computer analysis of the African-American face. Ann Plast Surg 34:318, 1995 10. Sim RS, Smith JD, Chan AS: Comparison of the aesthetic facial proportions of southern Chinese and white women. Arch Facial Plast Surg 2:113, 2000 11. Wang D, Qian G, Zhang M, et al: Differences in horizontal, neoclassical facial canons in Chinese (Han) and North American Caucasian populations. Aesthetic Plast Surg 21:265, 1997 12. Olszewski R, Zech F, Cosnard G, et al: Three-dimensional computed tomography cephalometric craniofacial analysis: Experimental validation in vitro. Int J Oral Maxillofac Surg 36:828, 2007 13. Vezzetti E, Calignano F, Moos S: Computer-aided morphological analysis for maxillo-facial diagnostic: A preliminary study. J Plast Reconstr Aesthet Surg 63:218, 2008 14. Bush K, Antonyshyn O: Three-dimensional facial anthropometry using a laser surface scanner: Validation of the technique. Plast Reconstr Surg 98:226, 1996 15. Kovacs L, Zimmermann A, Brockmann G, et al: Three-dimensional recording of the human face with a 3D laser scanner. J Plast Reconstr Aesthet Surg 59:1193, 2006 16. Weinberg SM, Scott NM, Neiswanger K, et al: Digital threedimensional photogrammetry: Evaluation of anthropometric precision and accuracy using a Genex 3D camera system. Cleft Palate Craniofac J 41:507, 2004 17. Ayoub AF, Siebert P, Moos KF, et al: A vision-based threedimensional capture system for maxillofacial assessment and surgical planning. Br J Oral Maxillofac Surg 36:353, 1998 18. Ferrario VF, Sforza C, Poggio CE, et al: Preliminary evaluation of an electromagnetic three-dimensional digitizer in facial anthropometry. Cleft Palate Craniofac J 35:9, 1998 19. Hennessy RJ, McLearie S, Kinsella A, et al: Facial shape and asymmetry by three-dimensional laser surface scanning covary with cognition in a sexually dimorphic manner. J Neuropsychiatry Clin Neurosci 18:73, 2006
1206 20. Ferrario VF, Sforza C, Dalloca LL, et al: Assessment of facial form modifications in orthodontics: Proposal of a modified computerized mesh diagram analysis. Am J Orthod Dentofacial Orthop 109:263, 1996 21. Schwenzer-Zimmerer K, Haberstok J, Kovacs L, et al: 3D surface measurement for medical application—Technical comparison of two established industrial surface scanning systems. J Med Syst 32:59, 2008 22. Ferrario VF, Sforza C, Schmitz JH, et al: A three-dimensional computerized mesh diagram analysis and its application in soft tissue facial morphometry. Am J Orthod Dentofacial Orthop 114:404, 1998 23. Kau CH, Richmond S, Savio C, et al: Measuring adult facial morphology in three dimensions. Angle Orthod 76:773, 2006
THREE-DIMENSIONAL ANALYSIS OF CHINESE FACES 24. Farkas LG: Anthropometry of the Head and Face. New York, Raven Press, 1994 25. Ferrario VF, Sforza C, Serrao G, et al: Soft tissue facial growth and development as assessed by the three-dimensional computerized mesh diagram analysis. Am J Orthod Dentofacial Orthop 116:215, 1999 26. Ward RE, Jamison PL: Measurement precision and reliability in craniofacial anthropometry: Implications and suggestions for clinical applications. J Craniofac Genet Dev Biol 11:156, 1991 27. Zhang XT, Wang SK, Zhang W, et al: Measurement and study of the nose and face and their correlations in the young adult of Han nationality. Plast Reconstr Surg 85:532, 1990 28. Moyers RE, Bookstein FL: The inappropriateness of conventional cephalometrics. Am J Orthod 75:599, 1979