ONLINE ONLY

Correlations between cephalometric and photographic measurements of facial attractiveness in Chinese and US patients after orthodontic treatment
Hee Soo Oh,a Edward L. Korn,b Xiaoyun Zhang,c Yan Liu,c Tianmin Xu,d Robert Boyd,e and Sheldon Baumrindf San Francisco, Calif, Bethesda, Md, and Beijing, China Introduction: Orthodontists rely on esthetic judgments from facial photographs. Concordance between estimates of facial attractiveness made from lateral cephalograms and those made from clinical photographs has not been determined. We conducted a preliminary examination to correlate clinicians rankings of facial attractiveness from standardized end-of-treatment facial photographs (Photo Attractiveness Rank) with cephalometric measurements of facial attractiveness made for the same subjects at the same time. Methods: Forty-ve Chinese and US orthodontic clinicians ranked end-of-treatment photographs of separate samples of 45 US and 48 Chinese adolescent patients for facial attractiveness. Separately for each sample, the photographic rankings were correlated with the values of 21 conventional hard- and soft-tissue measures from lateral cephalograms taken at the same visits as the photographs. Results: Among US patients, higher rank for facial attractiveness on the photographs was strongly associated with higher values for prole angle, chin prominence, lower lip prominence, and Z-angle, and also with lower values for angle of convexity, H-angle, and ANB. Among Chinese patients, higher rank for facial attractiveness on the photographs was strongly associated with higher values for Z-angle and chin prominence, and also with lower values for angle of convexity, H-angle, B-line to upper lip, and mandibular plane angle. Chinese patients whose %lower face height values approximated the ethnic ideal (54%) tended to rank higher for facial attractiveness than patients with either higher or lower values for %lower face height. The absolute values of the correlations for the 7 US measures noted above ranged from 0.41 to 0.59; those of the 7 Chinese measures ranged from 0.39 to 0.49.The P value of the least statistically signicant of these 14 correlations was 0.006, unadjusted for multiple comparisons. On the other hand, many cephalometric measures believed by clinicians to be indicators of facial attractiveness failed to correlate with facial attractiveness rank for either ethnicity at even the P \0.05 level, including SN-pogonion angle, lower incisor to mandibular plane angle, and Wits appraisal. Conclusions: In general, there was less association than expected or desired between objective measurements on the lateral cephalograms and clinicians rankings of facial attractiveness on sets of clinical photographs. (Am J Orthod Dentofacial Orthop 2009;136:762.e1-762.e14)

M
a

uch of the orthodontic literature since Downs classic 1948 article has focused on the analysis of the lateral cephalograms.1 Combinations of cephalometric measures have been grouped into many cephalometric analyses (eg, those of

Downs,1,2 Steiner,3 Tweed,4,5 Jarabak and Fizzell,6 Sassouni,7 Bjork,8 Ricketts,9,10 McNamara,11 Arnett and Bergman,12 and Arnett et al13). Findings from each of these analyses are used by orthodontists in diagnosis, treatment planning, and outcomes research. Almost all
The authors report no commercial, proprietary, or nancial interest in the products or companies described in this article. Supported by a grant of Orthodontic Faculty Development Fellowship Award from the American Association of Orthodontists Foundation and in part by NIHNIDR Grants DE07332 and DE08713. Reprint requests to: Sheldon Baumrind, 1525 Walnut St, Berkeley, CA 947091512; e-mail, sbaumrind@pacic.edu. Submitted, January 2009; revised and accepted, April 2009. 0889-5406/$36.00 Copyright 2009 by the American Association of Orthodontists. doi:10.1016/j.ajodo.2009.04.020

Assistant professor, Department of Orthodontics, Arthur A. Dugoni School of Dentistry, University of the Pacic, San Francisco, Calif. b Mathematical statistician, Biometric Research Branch, National Cancer Institute, NIH, Bethesda, Md. c Assistant professor, Department of Orthodontics, Peking University School and Hospital of Stomatology, Beijing, China. d Professor and chair, Department of Orthodontics, Peking University School and Hospital of Stomatology, Beijing, China. e Professor and chair, Department of Orthodontics, Arthur A. Dugoni School of Dentistry, University of the Pacic, San Francisco, Calif. f Professor, Department of Orthodontics, and Director, Craniofacial Research Instrumentation Laboratory, Arthur A. Dugoni School of Dentistry, University of the Pacic, San Francisco, Calif.

762.e1

762.e2

Oh et al

American Journal of Orthodontics and Dentofacial Orthopedics December 2009

these analyses contain some measurements that are believed to quantify facial attractiveness. In recent years, orthodontists have also come to rely heavily on esthetic judgments from semistandardized facial photographs. These photographs have now become a component of the orthodontists routine records set for diagnosis, treatment planning, and outcome analysis. An examination of the concordance between estimates of facial attractiveness made from lateral cephalograms and those made from clinical photographs is therefore in order. Some previous studies have probed this relationship, but the available ndings are far from conclusive.14-16 In our study, we conducted a preliminary examination of this clinically important question by correlating clinicians rankings of facial attractiveness from standardized end-of-treatment facial photographs (Photo Attractiveness Rank) with cephalometric measurements of facial attractiveness made for the same subjects at the same time. Our general hypothesis was that the values for the cephalometric measures would correlate highly with Photo Attractiveness Rank. In testing this hypothesis, we sought explicit answers to the following specic questions. 1. To what extent do conventional hard- and soft-tissue cephalometric measures generally considered to be related to facial attractiveness correlate with Photo Attractiveness Rank? To what extent do the data support the idea that the patients with the highest Photo Attractiveness Ranks have cephalometric values close to the cephalometric ideals, whereas patients with cephalometric values either higher or lower than the cephalometric ideals tend to be judged less attractive? (We asked this question because it reects the common expectation among orthodontists that, for many cephalometric measures, the highest rank for facial attractiveness on photographs would be closely associated with the ideal value of the cephalometric measures, whereas both higher and lower values for the cephalometric measures would tend to be associated with lower Photo Attractiveness Ranks. To the extent that this condition holds for any particular cephalometric measure, its presence will not be detected by conventional linear correlational methods unless the data are transformed. This problem and its solution will be discussed in Material and methods.) To what extent does adding together information from several different cephalometric measures of facial attractiveness improve our ability to account for observed differences in Photo Attractiveness Rank? (We asked this question to test the assump-

tion that the merging of information from different cephalometric measures improves our ability to predict Photo Attractiveness Rank.) Answers to these questions were sought by testing them in hypothesis form against data from samples of lateral cephalograms and facial photographs comprising a total of 93 treated Chinese and US orthodontic patients.
MATERIAL AND METHODS

2.

3.

The data evaluated in this study are from 2 samples of orthodontic patients of different ethnicities treated by experienced orthodontists in the United States and China. They were randomly selected from the records of patients who had completed treatment at the graduate orthodontic clinic of the Department of Orthodontics at Peking University, Beijing, China, and the private practice of the director of the graduate orthodontic clinic of the New Jersey Dental School, Newark. At each institution, a stratied random sample of 48 subjects was acquired, comprising 4 groups of 12 patients each. In each group of 12 patients, 3 were Class I extraction, 3 Class I nonextraction, 3 Class II extraction, and 3 Class II nonextraction. The sampling process was the same at both institutions and is illustrated schematically in Figure 1. Three of the original 48 US subjects were excluded from this study for poor image quality, yielding a nal sample of 45. In addition, several of the remaining US subjects hae missing values for some cephalometric measures, but the sample size for no cephalometric measure was less than 41. The US sample consisted of late-adolescent white patients living in northern New Jersey. The Chinese sample consisted of late-adolescent patients of Chinese ethnicity, residing in the Beijing area. Cephalometric and photographic data were complete for all subjects in the Chinese sample. Further details of the sampling process have been supplied in previous articles.17,18 Among the end-of-treatment materials available for each subject were conventional lateral cephalograms and standardized triplet facial photographs taken at the same end-of-treatment visit (Fig 2). The photographs of the US sample were taken with the patient positioned in the cephalostat according to the original standard of the Second Roentgenographic Workshop.19 The triplet photographs from both samples had been ranked for facial attractiveness by each of 45 orthodontically trained judges (25 Chinese, 20 US) using previously described methods.17,18 Also available for each patient were the database-stored coordinates of 35 hard- and soft-tissue landmarks located on the end-oftreatment lateral cephalogram for each subject. Each

American Journal of Orthodontics and Dentofacial Orthopedics Volume 136, Number 6

Oh et al 762.e3

Fig 1. Schematic showing the procedures for drawing the equivalent Chinese and US samples. Step 1: at each venue, patients who received treatment during a specied time period were identied. Step 2: each patient was assigned a random number; all subsequent procedures were conducted with the charts sorted in random order to ensure that the sample was representative of the population (ie, practice) from which it was drawn. Step 3: proceeding in random order, all charts with complete records were identied and duplicated for further studies (for the purposes of the general project of which this study is a part, a complete record was considered to be one in which a lateral cephalogram, study casts, full-mouth intraoral or panoramic x-rays, and a facial photographic triplet were available at the beginning and end of full-bonded orthodontic treatment). Step 4: parallel stratied subsets of 48 Chinese subjects and 45 US subjects with complete records were selected; the subset from each institution comprised the rst 48 randomly ordered charts from that venue that satised the Angle Class and extraction and nonextraction criteria for this study, except that 3 subjects were lost from the US group because their images were technically unsatisfactory.

cephalogram had been traced independently by 2 or more judges, and the average values had been recorded in a numerical database from which the measures of almost all standard cephalometric analyses could be calculated by computer operations. From the coordinate values of a subset of the hard- and soft-tissue landmarks, 9 cephalometric measures whose values are conventionally considered to be strong indicators of facial attractiveness were selected by the principal author (H.O.) for analysis. Later, an informal survey conducted among faculty orthodontists at the University of the Pacic in San Francisco identied 12 additional esthetically related cephalometric variables, which were added to the analysis. To reduce the complexity of data presentation, the results of the 2 analyses have been combined. Operational denitions for the 21 measures are listed below. Among them are 10 hard-tissue measures (designated [H]), 3 orientation-free soft-tissue measures (designated [S]), and 8 soft-tissue measures oriented to the hard-tissue-dened Frankfort line (designated [SF]). Eleven measures are angular; 9 are linear, and 1 is a ratio. The 9 measures examined in the rst pass are identied by asterisks.
Angular measures (in degrees)

2. 3.

4.

5.

6.

7.

8.

9.

1.

ANB angle*: the included angle between Point A, nasion, and Point B (after Riedel20) [H].

Angle of convexity: the included angle between nasion, Point A, and pogonion (after Downs2) [H]. Facial plane angle*: the acute angle formed by the intersection of the line nasion-pogonion with the Frankfort horizontal line (after Holdaway21) [H]. H-angle: the included angle between soft-tissue nasion, soft-tissue pogonion, and the most anterior point on the upper lip (after Holdaway22) [S]. Lower incisor to A-pogonion angle: the acute angle formed by the intersection between the line Point A-pogonion and the long axis of the lower incisor (after Ricketts23) [H]. Lower incisor to mandibular plane angle*: the angle formed by the intersection of the long axis of the lower central incisor and the lower border of the mandible (after Tweed4,5) [H]. Mandibular plane angle*: the acute angle formed by the intersection of the line gonion-menton with the Frankfort horizontal plane (after Downs2) [H]. Nasolabial angle: the included angle between lower nose tip, soft-tissue subnasale, and the vermilion border of the upper lip (after Burstone24) [S]. Prole angle*: the included angle between glabella (dened in this study as the most anterior point on the soft tissue of the forehead), soft-tissue subnasale, and soft-tissue pogonion, dened in

762.e4

Oh et al

American Journal of Orthodontics and Dentofacial Orthopedics December 2009

Fig 2. Representative photographic triplets from the US (upper) and Chinese (lower) samples.

10.

11.

this study as the most anterior point on the soft-tissue surface of the chin (after Burstone25) [S]. SN-pogonion angle*: the included angle between the landmarks sella-nasion and pogonion (after Bell et al26) [H]. Z-angle: the angle formed by the intersection between the Frankfort horizontal line and the line from soft-tissue pogonion to the anterior-most point on the lower lip (after Merrield27) [SF].

13.

14.

15.
Linear measures (in millimeters)

12.

B-line to lower lip: the distance from the anteriormost point on the lower lip and the line from softtissue subnasale to soft-tissue pogonion measured parallel to the Frankfort plane (after Burstone25) [SF].

16.

B-line to upper lip: the distance from the anteriormost point on the upper lip and the line from softtissue subnasale to soft-tissue pogonion measured parallel to the Frankfort plane (after Burstone25) [SF]. Chin prominence: the distance from soft-tissue pogonion to a line perpendicular to the Frankfort plane and passing through soft-tissue subnasale (after Arnett et al13) [SF]. E-line to lower lip*: the distance from the anterior-most point on the lower lip to the line from the nose tip to soft-tissue pogonion measured parallel to the Frankfort plane (after Ricketts23) [SF]. Lower incisor to A-pogonion distance: the distance between the tip of the lower incisor and a line from Point A to pogonion measured parallel to the Frankfort plane (after Ricketts23) [H].

American Journal of Orthodontics and Dentofacial Orthopedics Volume 136, Number 6

Oh et al 762.e5

Fig 3. Locations of the landmarks used in computing the 21 cephalometric measures. The dashed lines show the Frankfort horizontal and a line perpendicular to it passing through subnasale.

17.

18.

19.

20.

Lower lip prominence: the distance from the anterior-most point of the lower lip to a line perpendicular to the Frankfort plane and passing through soft tissue subnasale (after Arnett et al13) [SF]. Nose prominence: the distance from the nose tip to a line perpendicular to the Frankfort plane and passing through soft-tissue subnasale (after Arnett et al13) [SF]. Upper lip prominence: the distance from the anterior-most point of the upper lip to a line perpendicular to the Frankfort plane and passing through soft-tissue subnasale (after Arnett et al13) [SF]. Wits appraisal*: the distance between the intersection of a perpendicular line to occlusal plane from Point A and the intersection of a perpendicular line to the occlusal plane from Point B (after Jacobson28) [H].

The locations of the anatomical landmarks required to compute the 21 measures are shown in Figure 3.
Statistical analysis

Ratio

21.

%Lower face height*: lower face height times 100 divided by total face height (after Farkas29) [H].

Since this study was essentially an exploratory investigation of a new area of interest, we allowed ourselves to consider more variables than would be appropriate in a denitive hypothesis test. Rather than perform a formal multiple comparisons adjustment, we chose to focus our consideration on measures whose P values were less than 0.01. For answers to the rst question, Pearson correlations (r) were used to calculate the strength of association between Photo Attractiveness Ranks of the patients and their scores for each of 21 cephalometric measures. Separate tests were performed for the patients of each ethnicity. However, each patients Photo Attractiveness Rank was the single mean value for all 45 US and Chinese judges pooled. The decision to pool the values of the Chinese and US judges was based on the previous nding that the correlation between the Photo

762.e6

Oh et al

American Journal of Orthodontics and Dentofacial Orthopedics December 2009

Table I.

Means and standard deviations by ethnicity for this sample and means of best available comparable studies
US cohort Present sample Best available norm Mean SD Ref Present sample Mean SD Chinese cohort Best available norm Mean SD Ref

Cephalometric measures Angular measures ( ) ANB Angle of convexity Facial plane angle H-angle Lower incisor to A-pogonion line Lower incisor to mandibular plane Mandibular plane angle Nasolabial angle Prole angle SN-pogonion angle Z-angle Linear measures (mm) B-line to lower lip B-line to upper lip Chin prominence (F) (M) E-line to lower lip Lower incisor to A-pogonion line Lower lip prominence (F) (M) Nose prominence Upper lip prominence (F) (M) Wits appraisal (F) (M) Ratio (%) %Lower face height Ref, Reference; F, female; M, male.

Mean

SD

2.7 2.4 83.5 13.5 25.7 95.6 27.7 127.3 165.6 77.0 71.2 2.6 3.1 12.2 15.6 2.9 2.4 4.5 4.3 17.1 0.5 0.0 2.3 3.2 53.7

1.9 5.3 3.4 3.7 5.2 7.0 6.6 8.2 5.9 3.7 6.9 2.5 1.8 5.4 8.6 3.2 1.6 3.1 4.2 2.4 2.1 2.6 4.0 4.3 2.3

2.0 0.0 85.0 10.0 22.0 90.0 25.0 102.0 168.7 79.0 80.0 3.0 2.0 2.6 3.5 2.0 1.0 1.9 1.0 16.0 3.7 3.3 0.1 1.2 55.5

1.8 5.1 3.0 3.0 4.0 5.8 4.5 8.0 4.1 3.0 9.0 1.0 1.0 1.9 1.9 2.0 2.0 1.4 2.2 1.4 1.2 1.7 1.8 1.9 1.0

20 2 21 22 23 4, 5 4, 5 24 25 26 27 25 25 13 23 23 13 13 13 28

4.0 7.9 83.1 18.0 27.9 99.0 30.1 97.1 166.6 76.7 66.3 5.9 6.1 11.8 16.2 0.9 4.4 0.1 3.4 14.9 1.8 0.0 0.9 0.1 53.7

2.3 5.7 3.5 3.7 6.5 9.6 7.1 9.6 5.1 3.7 6.2 2.3 1.7 5.2 4.6 2.4 2.0 2.8 3.2 2.1 2.1 2.3 3.1 2.1 1.9

2.5 5.5 87.7 13.6 26.9 94.3 27.2 84.0 168.6 79.5 71.0 6.5 4.9 6.7 8.5 1.4 4.3 3.8 2.6 13.1 6.1 5.5 3.9 2.6 54.0

1.5 2.9 3.0 3.6 5.1 5.5 4.7 80-110 3.7 3.3

30 31 30 30 2 30 30 31 31 30 32 30 30 31 30 31 31 31 31 31

1.6 1.7 3.3 3.4 1.9 1.8 2.3 1.6 2.0 2.2 2.4 2.1 1.2 1.0

29

31

Attractiveness Ranks of the Chinese and US judges was strong (r 5 0.92 and P \0.0001 for US patients; r 5 0.86 and P \0.0001 for Chinese patients).18 In examining associations between variables, we have chosen to report Pearson rho values rather than Spearman rho values that are frequently used for ranked data. We reasoned that, since Photo Attractiveness Rank for each patient in this study is the mean of 45 individual values, they would be expected to be approximately normally distributed by the central limit theorem. In addition, the cephalometric variables did not appear to have grossly nonnormal distributions. Therefore, the Pearson correlation would be expected to have good properties for detecting linear associations between the Photo Attractiveness Rank and the cephalometric variables. In theory, the Spearman rho might be more sensitive to nonlinear but monotone associations, although with the small sample sizes this is unclear. Neither correlation would be sensitive to nonmonotone associations; this is why we used the absolute-value transformations to address question 2.

The advantage of using the Pearson correlation over the Spearman was that its square (r2) can be used as a measure of explained variance, thus facilitating a straightforward assessment of increased association when combinations of cephalometric variables were considered in question 3. Question 3 was investigated by using the technique of stepwise linear regression. A P value less than 0.01 was required for a variable to enter the model for each ethnicity, and a P value greater than 0.05 was required to leave the model. Although a number of alternative methods of regression analysis were available, we considered stepwise regression a reasonable approach for assessing the association with combinations of cephalometric variables, given the relatively small sample sizes and the exploratory nature of the study.
RESULTS

Table I lists the means and standard deviations for all 21 cephalometric measures for the US and Chinese

American Journal of Orthodontics and Dentofacial Orthopedics Volume 136, Number 6

Oh et al 762.e7

Table II.

Correlations between Photo Attractiveness Rank and 21 raw cephalometric measures

US patients (n 5 45) Raw cephalometric value Chinese patients (n 5 48) Raw cephalometric value Order 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Cephalometric measure Z-angle ( ) Angle of convexity ( ) Chin prominence (mm) H-angle ( ) B-line to upper lip (mm) Mandibular plane angle ( ) ANB ( ) Facial plane angle ( ) E-line to lower lip (mm) Lower incisor to A-pogonion line ( ) Prole angle ( ) B-line to lower lip (mm) Lower incisor to mandibular plane ( ) Nose prominence (mm) Lower lip prominence (mm) Upper lip prominence (mm) SN-pogonion angle ( ) %Lower face height (%) Nasolabial angle ( ) Wits appraisal (mm) Lower incisor to A-pogonion line (mm) r 0.49 0.45 0.42 0.40 0.40 0.39 0.35 0.34 0.34 0.33 0.32 0.31 0.25 0.20 0.19 0.10 0.10 0.09 0.06 0.02 0.001 P 0.0004 0.002 0.003 0.005 0.005 0.006 0.014 0.02 0.02 0.02 0.03 0.03 0.09 0.18 0.20 0.50 0.52 0.55 0.69 0.87 0.99

Order 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Cephalometric measure Prole angle ( ) Angle of convexity ( ) Chin prominence (mm) H-angle ( ) ANB ( ) Lower lip prominence (mm) Z-angle ( ) Mandibular plane angle ( ) Upper lip prominence ( ) Lower incisor to A-pogonion line (mm) E-line to lower lip (mm) Nose prominence (mm) SN-pogonion angle ( ) Nasolabial angle ( ) Facial plane angle ( ) Wits appraisal (mm) B-line to lower lip (mm) B-line to upper lip (mm) Lower incisor to mandibular plane ( ) Lower incisor to A-pogonion line ( ) %Lower face height (%)

r 0.59 0.51 0.50 0.48 0.44 0.41 0.42 0.38 0.32 0.29 0.29 0.25 0.24 0.25 0.23 0.21 0.19 0.12 0.07 0.05 0.01

P \0.0001 0.0003 0.001 0.001 0.003 0.005 0.006 0.011 0.03 0.05 0.06 0.10 0.11 0.13 0.14 0.16 0.22 0.43 0.65 0.73 0.94

Negative values for r indicate cephalometric measures that decrease as facial attractiveness increases.

samples. For use in later comparisons, it also contains information on the ideal value for each measure for each ethnicity.2,4,5,2032 The ndings of the study will be presented in terms of the 3 questions raised in the introduction. Question 1 was to what extent do conventional hardand soft-tissue cephalometric measures generally considered to be related to facial attractiveness correlate with clinicians rankings of facial photographs for facial attractiveness? To answer this question, Table II lists the correlation with the Photo Attractiveness Rank of each of the 21 cephalometric measures for each ethnicity. Within ethnicity, the values are listed in descending order of absolute magnitude. Seven cephalometric measures for the US sample and 6 cephalometric measures for the Chinese sample correlated with the judges rankings of facial photographs with P values less than 0.01. The absolute values of the correlations for the 7 US measures ranged from 0.41 to 0.59. The corresponding range for the 6 Chinese measures was 0.39 to 0.49. For each ethnicity, both linear and angular measures and both hard- and soft-tissue

measures were found among the most highly signicant measures. Four of the most highly signicant measures were the same in the Chinese and US patient samples: angle of convexity, chin prominence, Z-angle, and H-angle. In the US sample, prole angle, chin prominence, lower lip prominence, and Z-angle correlated positively with Photo Attractiveness Rank. As the value of each of these measures increased, the patient tended to be judged more attractive. On the other hand, angle of convexity, H-angle, and ANB angle correlated negatively with the Photo Attractiveness Rank. As the value of each of these measures increased, the patient tended to be judged less attractive. In the Chinese sample, Z-angle and chin prominence correlated positively with Photo Attractiveness Rank. As the value of each of these cephalometric measures increased, the patient tended to be judged more attractive. On the other hand, angle of convexity, H-angle, B-line to upper lip, and mandibular plane angle correlated negatively with Photo Attractiveness Rank; as the value of each of these cephalometric measures increased, the patient tended to be judged less attractive.

762.e8

Oh et al

American Journal of Orthodontics and Dentofacial Orthopedics December 2009

Scatterplots for each of the highly statistically signicant measures in Table II are given in Figure 4. They show the distributions of individual case values for the associations between Photo Attractiveness Rank on the y-axis and each of the cephalometric measures on the x-axis. Values for the least attractive patients are at the bottom of each plot, and values for the most attractive patients are at the top. Within ethnicity, the y-value for each patient is the same in all plots. On each plot, the ideal value of the cephalometric measure is represented by the vertical red line. It can be seen that, for almost all these measures, the ideal value falls within the distribution, but it is rarely near the distributions center. For the majority of cephalometric variables, the values for the less attractive patients tend to be farther from the ideal than the values for the more attractive patients. The patterns of the scatterplots for each of the 4 variables that met our high signicance criterion for both ethnicities (angle of convexity, chin prominence, H-angle, and Z-angle) were quite similar for both ethnicities. With regard to the goal of clinical prediction, it was noted that, even though when the correlations are highly statistically signicant, these real-world distributions showed considerable variability. We also thought it important that many cephalometric measures generally considered to be strongly associated with facial attractiveness failed to correlate signicantly with the judges rankings of facial attractiveness on photographs. Among these were lower incisor to mandibular plane angle, %lower face height, Wits appraisal, nose prominence, and all other measures in Table I whose probability of occurring by chance was greater than 0.05. At this point in the study, it was not possible to be sure whether the values for these variables actually had little association with Photo Attractiveness Rank or whether they had a curvilinear association that was undetectable by linear correlation unless the data were transformed. Hence, the relevance of Question 2. Question 2 was to what extent do the data support the idea that the patients with the highest Photo Attractiveness Ranks have cephalometric values close to the cephalometric ideals, whereas patients with cephalometric values either higher or lower than the cephalometric ideals tend to be judged less attractive? If we correlate Photo Attractiveness Rank with a cephalometric measure whose facial attractiveness is highest at its ideal value but lower both above and below the ideal value (a condition we expected to encounter for a number of cephalometric measures), the regression line would tend to approximate a parabolic curve rather than the straight line that results

when 2 variables maintain the same relationship to each other throughout their range. This is a problem because the Pearson correlation method as usually used is almost completely insensitive to this type of nonlinear distribution. To protect ourselves from failing to detect associations of this type between Photo Attractiveness Rank and one or more of the cephalometric measures under examination, a new value was calculated equal to the absolute distance (in degrees or millimeters) between each patients raw value for each measure and the ethnic ideal for that measure. The transformed cephalometric values for each cephalometric measure are then amenable to conventional linear correlation with the patients Photo Attractiveness Ranks. The ideal value for each measure for each ethnicity from Table I was used to transform the raw value of each subject by rescaling it in terms of its absolute distance from the ideal. Table III lists the transformed correlations and P values for the relationships between Photo Attractiveness Rank and each of the 21 cephalometric measures for both the Chinese and the US samples. Also listed is the increment or decrement in the strength of each correlation as compared with Table II. As can be seen in the table, only 1 measure that did not satisfy our P \0.01 criterion for statistical signicance in Table II achieved such signicance after transformation. This was the transformed value of %lower face height in the Chinese sample. For this cephalometric measure, the change in r value was dramatic, moving from r 5 0.09 to r 5 0.44, and from the 18th position among the 21 Chinese cephalometric variables to rst. The actual distributions of the raw and transformed values for this correlation are shown in Figure 5. This implies that %lower face height values that were larger or smaller than the Chinese ideal of 54% tended to be equally unattractive. No other cephalometric measure for either ethnicity whose correlation with Photo Attractiveness Rank did not satisfy the P \0.01 criterion before transformation increased sufciently to satisfy it after transformation. B-line to lower lip and B-line to upper lip in the US sample did show substantial improvements in r value after transformation but still did not reach our criterion for statistical signicance. In general, the transformations had less effect than we had expected. Three of the 4 cephalometric measures that had correlated most highly with Photo Attractiveness Rank in both the Chinese and US samples in Table I (chin prominence, angle of convexity, and Z-angle) remained highly signicant after transformation, with only very small changes in r value. For example,

American Journal of Orthodontics and Dentofacial Orthopedics Volume 136, Number 6

Oh et al 762.e9

762.e10

Oh et al

American Journal of Orthodontics and Dentofacial Orthopedics December 2009

the transformation increased the value of prole angle in the US sample from 0.59 to 0.61. It is theoretically possible that a curvilinear relationship might have been missed by our transformation if the ideal value for that cephalometric measure were far from the center of the distribution of sampled values. To control for this possibility, we examined the plots of all cephalometric measures vs Photo Attractiveness Rank. (All plots can be viewed on the Internet at www.cril.org.) Only for lower lip prominence and nasolabial angle, both for the US sample, were there any indications that this might be true. To explore this possibility further, we transformed the individual case values of these 2 measures by computing their absolute differences from their sample means rather than from the ideal for the cephalometric measure. The resulting correlations were 0.25 (P 5 0.098) for lower lip prominence and 0.39 (P 5 0.015) for nasolabial angle. The value for lower lip prominence represents a decrement with respect to the value in Table II. The nding for nasolabial angle is more interesting. In the past, our laboratory found the reliability of this measure to be low because its landmarks are located near each other and are ambiguous. The fact that the transformed value nevertheless came so close to meeting our criterion for statistical signicance makes it a reasonable target for further investigation. Question 3 was are the associations made stronger when combinations of cephalometric variables are compared with the rankings from facial photographs? In clinical orthodontics, practitioners frequently aggregate data from several measures. Hence, we asked in our third question whether combinations among the signicant variables in our study can more fully account for the observed variability in Photo Attractiveness Rank. To provide answers to this question, separate stepwise regression analyses were performed for the US and the Chinese samples. The dependent variable in each test was Photo Attractiveness Rank. The independent variables in each test were the most highly statistically signicant raw and transformed cephalometric variables for that ethnicity. We asked whether any combination of these variables yielded a substantial improvement in accounting for attractiveness in either ethnicity. For the US sample, the transformed value of prole angle was the only variable that entered into

the stepwise regression model. Its r2 value was 0.37, implying that slightly more than 2/5 of the variability in Photo Attractiveness Rank could be accounted for by the set of cephalometric variables. For the Chinese sample, Z-angle entered the model rst, followed by the transformed value of %lower face height. The cumulative r2 value for the 2 variables considered together was 0.41, implying that here too slightly more than 2/5 of the variability in Photo Attractiveness Rank could be accounted for by the pair of cephalometric variables. (The r2 value for Z-angle considered alone was 0.24.) It can be seen that adding together information from several raw and transformed cephalometric variables did not greatly improve our ability to explain the clinicians rankings of facial photographs for attractiveness. Although it is true that our sample sizes were small for conducting multiple regression analyses of this sort, our results do strongly imply the need for caution in treating ndings from different but related cephalometric measures as if information from them can be treated additively.
DISCUSSION

Lateral cephalograms and facial photographs have complementary roles in the evaluation of facial attractiveness by orthodontists. Photographs show the surface structures of the face in considerable detail, and x-ray images allow us to understand the relationship between those surface structures and the skeletal and dental armature that supports them. Certainly, photographs are much closer to the natural state of the subject than are lateral cephalograms. However, techniques for quantitative measurement of facial photographs and standardization of photographic orientation are much less well advanced in orthodontics.33-38 Ranking methods such as the one that we used are useful for the conduct of studies such as this one. But the numerical rank of each of our subjects was completely dependent on the judges evaluations of the other patients in our particular sample. For that reason, it cannot be compared quantitatively to the rank of a photograph from any other sample. Clearly, the development of reliable, clinically useful quantitative and objective measures of facial attractiveness on photographs would be desirable, but such measures have yet to be developed. Now, only relatively vague sets of rules and principles have been

Fig 4. Distributions of the actual relationship between Photo Attractiveness Rank and A, the 7 US and B, 6 Chinese cephalometric measures for which highly signicant associations are reported in Table II. The purportedly ideal value for each measure from Table I is indicated by a vertical red line. Even though these associations are highly statistically signicant, these real-world distributions are much more scattered than the perfect schematic distributions.

American Journal of Orthodontics and Dentofacial Orthopedics Volume 136, Number 6

Oh et al

762.e11

Table III.

Correlations between Photo Attractiveness Rank and 21 transformed cephalometric measures

US patients (n 5 45) Transformed value Chinese patients (n 5 48) Transformed value Difference* Order 0.02 0.05 0.01 0.05 0.03 0.02 0.18 0.11 0.03 0.06 1 2 3 4 5 6 7 8 9 10 Cephalometric measure %Lower face height Chin prominence (mm) Angle of convexity ( ) Z-angle ( ) ANB ( ) Facial plane angle ( ) H-angle ( ) Mandibular plane angle ( ) Nose prominence (mm) Prole angle ( ) r 0.44 0.43 0.38 0.37 0.36 0.35 0.33 0.33 0.31 0.28 P 0.002 0.002 0.008 0.009 0.01 0.01 0.02 0.02 0.03 0.06 Difference* 0.35 0.01 0.07 0.12 0.01 0.01 0.07 0.06 0.12 0.04

Order 1 2 3 4 5 6 7 8 9 10

Cephalometric measure Prole angle ( ) H-angle ( ) Chin prominence (mm) Angle of convexity ( ) Z-angle ( ) Lower lip prominence (mm) B-line to lower lip (mm) SN-pogonion angle ( ) Upper lip prominence ( ) Lower incisor to A-pogonion line (mm) B-line to upper lip (mm) Mandibular plane angle ( ) E-line to lower lip (mm) Nasolabial angle ( ) ANB ( )

r 0.61 0.53 0.49 0.46 0.45 0.40 0.37 0.35 0.35 0.35

P \0.0001 0.0002 0.001 0.002 0.003 0.008 0.01 0.02 0.02 0.02

11 12

0.32 0.28

0.04 0.06

0.20 0.10

11 12

13 14 15

0.26 0.25 0.25

0.10 0.13 0.10

0.03 0.00 0.19

13 14 15

16 17 18 19 20

Facial plane angle ( ) Wits appraisal (mm) %Lower face height Nose prominence (mm) Lower incisor to mandibular plane angle ( ) Lower incisor to A-pogonion line angle ( )

0.20 0.15 0.12 0.12 0.05

0.19 0.31 0.45 0.45 0.74

0.03 0.06 0.11 0.13 0.02

16 17 18 19 20

B-line to upper lip (mm) Lower incisor to mandibular plane angle ( ) SN-pogonion angle ( ) Lower lip prominence (mm) Lower incisor to A-pogonion line angle ( ) E-line to lower lip (mm) Upper lip prominence (mm) Nasolabial angle ( ) Wits appraisal (mm) B-line to lower lip (mm)

0.21 0.19

0.16 0.19

0.19 0.05

0.19 0.18 0.10

0.21 0.22 0.48

0.09 0.01 0.23

0.10 0.08 0.08 0.04 0.04

0.49 0.58 0.60 0.79 0.81

0.24 0.02 0.02 0.02 0.27

21

0.04

0.79

0.01

21

Lower incisor to A-pogonion line (mm)

0.01

0.92

0.01

*Difference 5 change in r compared to Table II; %Lower face height in the Chinese sample had the largest increase in absolute r value and is the only measure that increased sufcently after transformation to satisfy our criteria for high statistical signicance. Other large increases in absolute r values were found for B-line to lower lip and B-line to upper lip in the US sample, but these still did not meet our criteria for statistical signicance.

established, mostly imported from the elds of art (eg, divine proportions, 39-42 and vertical thirds and horizontal fths43-46) and psychology (eg, averageness, 47-50 youthfulness,51,52 and symmetry53,54). Those measures on photographs that have been thus far developed in orthodontics (eg, the smile line55-57) have been subjected to limited critical testing. And when tests have been done, the results have generally not been statistically signicant.58,59 Lateral cephalograms, on the other hand, are fairly well understood by the members of the orthodontic community. Although quantication of facial esthetics is certainly not the main use of cephalograms in ortho-

dontics, many cephalogrametric measurements have been proposed as reliable indexes of facial attractiveness. From the early work of Downs1 to the recent work of Arnett et al,13 almost every proposed cephalometric analysis contains some measures of facial attractiveness. Hence, it seems reasonable to investigate the concordance between the objective angular and linear measurements of x-ray cephalometry and the subjective but highly reliable ranking of facial photographs for attractiveness. The underlying approach of this study was to use such highly reliable but noninterval and hence only relative average rankings by clinicians of representative

762.e12

Oh et al

American Journal of Orthodontics and Dentofacial Orthopedics December 2009

Fig 5. A, Raw cephalometric values from Table II; B, transformed values from Table III. The high statistical signicance of plot B shows that patients who had raw %lower face height values close to the ethnic mean of 54% tended to have more favorable Photo Attractiveness Ranks than patients who had %lower face height values either higher or lower than the ideal value.

sets of facial photographs as a standard against which to test the association of 21 widely used quantitative cephalometric measures. The cephalometric variables chosen for this test included classical hard-tissue measures from Downs,1,2 Steiner,3 Riedel,20 and so on, as well as soft-tissue measures from the more recent analyses of Holdaway,21 Arnett and Bergman,12 Arnett et al,13 and Burstone.24 Separate assessments were made for samples of treated Chinese and US orthodontic patients. Approximately one third of the 21 cephalometric measures tested for patients of each ethnicity correlated with the clinicians rankings for facial attractiveness for at least 1 ethnicity with high statistical signicance (P\0.01). Chin prominence, angle of convexity, H-angle, and Z-angle correlated signicantly with Photo Attractiveness Rank among patients of both US and Chinese ethnicities. Some other measures correlated signicantly with Photo Attractiveness Rank among patients of 1 ethnicity but not of the other. Among the US patients, these measures included prole angle, ANB angle, and lower lip prominence. Among the Chinese patients, B-line to upper lip and mandibular plane angle were included. The statistically signicant relationships described in the previous paragraph were all linear; ie, facial attractiveness tended to increase (or decrease) in a fairly straight line throughout the full range of available values for a particular measure. Additional data-trans-

forming tests were performed for all 21 cephalometric measures for both ethnicities to detect any measures for which clinicians preferred some ideal value in the middle of the variable range, but values on either side of the ideal tended to be found less attractive. We believed at the outset of this study that many cephalometric measures would exhibit this pattern. But we found this pattern to be of consequential importance for only 1 measure, %lower face height among the Chinese patients. For this variable, there was substantial evidence that the judges considered values approximating the ideal of 54% (Table I) to be most attractive, with values either above or below 54% tending to be considered less attractive. This implies that, among the Chinese patients, the vertical orientation of the chin contributed consequentially to the evaluation of facial attractiveness (see the study of Johnston et al60 for comparison). The limited ability of our data-transforming procedure to increase the number of statistically signicant correlations between the cephalometric measures and the Photo Attractiveness Rank seems to imply that clinicians perception of attractiveness might not be as symmetrically distributed around the ideal values of most cephalometric measures as has been thought. Almost half of the 21 cephalometric measures commonly considered markers for facial attractiveness failed to correlate with Photo Attractiveness Rank better than chance in either their raw or transformed form. Among the poorer measures were E-line to lower lip, lower incisor to mandibular plane angle, and Wits appraisal. And even among cephalometric measures that did correlate highly with photographic rank, no single raw or transformed measure had a higher correlation than 0.61, and no combination of measures accounted for more than 41% of the variability in the ranking of facial photographs for attractiveness by orthodontists. In considering the relatively small portion of the total variance in Photo Attractiveness Rank accounted for by the cephalograms, it should be recalled that the photographs used in the ranking process included frontal and smiling views as well as a prole view. Recent studies of facial photographs of the type examined here have reported that frontal and frontal smiling images account for a larger part of a judges conclusions about facial attractiveness than do prole images.14,61,62 In general, it seems fair to infer that most information that viewers use in the evaluation of facial attractiveness is not available in lateral cephalograms. As in any study, the limitations of the sample must be considered when generalizing the study results. The images in this sample were all taken at the end of the orthodontic treatment of relatively attractive normal

American Journal of Orthodontics and Dentofacial Orthopedics Volume 136, Number 6

Oh et al

762.e13

subjects. The sample had been stratied to make it more representative of the adolescent population seeking orthodontic treatment than of the general population. In each sample at original presentation, half of the patients were Class I, and half were Class II; half were treated with extractions, and half without extractions; three quarters were girls, and a quarter were boys. These constraints might have tended to make the range of values of the cephalometric measures smaller than would have been the case for pretreatment cephalograms.
CONCLUSIONS

In general, correlations between cephalometric measures and rankings of facial attractiveness were less strong than had been expected. Soft-tissue prole angle appeared to be the most important single variable in accounting for differences in facial attractiveness rank on photographs of US patients, but it was not nearly as important in ranking the photographs of Chinese patients. In the Chinese patients, a combination of Z-angle and transformed %lower face height contributed statistically signicantly to the explanation of the observed variability in Photo Attractiveness Rank. However, no combination of cephalometric measures for either ethnicity accounted for more than 42% of the variance in Photo Attractiveness Rank, thus leaving more than half of the variance unexplained. In general, it seems fair to infer that most of the information that viewers use in the evaluation of facial attractiveness is not available in lateral cephalograms. For both Chinese and US patients, atter facial proles appeared to be preferred. In both ethnicities, greater chin prominence in the sagittal direction was preferred, and lip position was considered important. With the exception of %lower face height in the Chinese sample, close approximation to the ethnic ideal did not appear to be highly correlated with facial attractiveness rank.
REFERENCES 1. Downs WB. Variations in facial relationship. Their signicance in treatment and prognosis. Am J Orthod 1948;34:812-40. 2. Downs WB. The role of cephalometrics in orthodontic case analysis and diagnosis. Am J Orthod 1952;38:162-82. 3. Steiner CC. Cephalometrics for you and me. Am J Orthod 1953; 39:729-54. 4. Tweed CH. Evolutionary trends in orthodontics, past, present, and future. Am J Orthod 1953;39:81-108. 5. Tweed CH. The Frankfort-mandibular incisor angle (FMIA) in orthodontic diagnosis, treatment planning and prognosis. Angle Orthod 1954;24:121-69. 6. Jarabak JR, Fizzell JA. Technique and treatment with light-wire edgewise appliance. 2nd ed. St Louis: C.V. Mosby; 1972.

7. Sassouni V. A classication of skeletal facial types. Am J Orthod 1969;55:109-23. 8. Bjo rk A. The face in prole. Svennsk Tandla kare-Tidskrift 40 No. 5B, suppl lund, 1947. 9. Ricketts RM. Perspectives in the clinical application of cephalometrics. Angle Orthod 1981;51:115-50. 10. Ricketts RM. Planning treatment on the basis of facial pattern and an estimate of its growth. Angle Orthod 1957;27:14-37. 11. McNamara JA. A method of cephalometric evaluation. Am J Orthod 1984;86:449-69. 12. Arnett GW, Bergman RT. Facial keys to orthodontic diagnosis and treatment planning. Part I. Am J Orthod Dentofacial Orthop 1993; 103:299-31. 13. Arnett GW, Jelic JS, Kim J, Cummings DR, Beress A, Worley CM, et al. Soft tissue cephalometric analysis: diagnosis and treatment planning of dentofacial deformity. Am J Orthod Dentofacial Orthop 1999;116:239-53. 14. Matoula S, Pancherz H. Skeletofacial morphology of attractive and nonattractive faces. Angle Orthod 2006;76:204-10. 15. Tulloch C, Philips C, Dann C. Cephalometric measures as indicators of facial attractiveness. Int J Adult Orthod Orthognath Surg 1993;8:171-9. 16. Kiekens RMA, Maltha JC, Vant Hof MA, Kuijpers-Jagtman AM. Objective measures as indicators for facial esthetics in adolescents. Angle Orthod 2006;76:551-6. 17. Xu TM, Korn EL, Liu Y, Oh HS, Lee KH, Boyd RL, et al. Facial attractiveness: ranking of end-of-treatment facial photographs by pairs of Chinese and US orthodontists. Am J Orthod Dentofacial Orthop 2008;134:74-84. 18. Liu Y, Korn EL, Oh HS, Pearson H, Xu TM, Baumrind S. Comparison of Chinese and US orthodontists averaged evaluations of facial attractiveness from end-of-treatment facial photographs. Am J Orthod Dentofacial Orthop 2009;135: 621-34. 19. Salzman JA. Roentgenographic cephalometrics. Philadelphia: J.B. Lippincott; 1961. 20. Riedel RR. The relation of maxillary structures to cranium in malocclusion and in normal occlusion. Angle Orthod 1952;22: 142-5. 21. Holdway RA. A soft tissue cephalometric analysis and its use in orthodontic treatment planning. Part I. Am J Orthod 1983;84:1-28. 22. Holdaway RA. A soft tissue cephalometric analysis and its use in orthodontic treatment planning. Part II. Am J Orthod 1984;85:279-93. 23. Ricketts RM. Cephalometric analysis and synthesis. Angle Orthod 1961;31:141-56. 24. Burstone CJ. Integumental prole. Am J Orthod 1958;44:1-25. 25. Burstone CJ. Lip posture and its siginicance in treatment planning. Am J Orthod 1967;53:262-84. 26. Bell WH, Proft WR, White RP. Surgical correction of dentofacial deformities. Vol I. Philadelphia: W.B. Saunders; 1980. p. 137-50. 27. Merrield LL. The prole line as an aid in critically evaluating facial esthetics. Am J Orthod 1966;52:804-22. 28. Jacobson A. The Wits appraisal of jaw disharmony. Am J Orthod 1975;67:125-38. 29. Farkas LG. Anthropometry of the head and face in medicine. New York: Elsevier; 1981. 30. Yu XH. A radiographic cephalometric study of soft-tissue prole on 180 normal occlusions among Chinese population [thesis]. Beijing: Peking University; 1985. 31. Zhang XZ. Longitudinal study of craniofacial growth of 13-17 years adolescents with normal occlusion [thesis]. Beijing: Peking University; 1995.

762.e14

Oh et al

American Journal of Orthodontics and Dentofacial Orthopedics December 2009

32. Yuan DH, Zuo YP. The soft tissue prole analysis of normal occlusion in the early permanent dentitiona cephalometric study. Chinese Journal of Aesthetic Medicine 2007;1:95-8. 33. Lucker GW. Esthetics and a quantitative analysis of facial appearance. In: McNamara JA Jr, editor. Esthetics and the treatment of facial form. Monograph 28. Craniofacial Growth Series. Ann Arbor, MI: Center for Growth and Development, University of Michigan; 1993. 34. Edler RJ. Background considerations to facial aesthetics. J Orthod 2001;28:159-68. 35. Wahl N. Orthodontics in 3 millennia. Chapter 7: facial analysis before the advent of the cephalometer. Am J Orthod Dentofacial Orthop 2006;129:293-8. 36. Peck S, Peck L. Facial realities and oral esthetics. In: McNamara JA Jr, editor. Esthetics and the treatment of facial form. Monograph 28. Craniofacial Growth Series. Ann Arbor: Center for Human Growth and Development; University of Michigan; 1993. 37. Giddon DB, Sconzo R, Kinchen JA, Evans CA. Quantitative comparison of computerized discrete and animated prole preferences. Angle Orthod 1996;66:441-8. 38. Orsini MG, Huang GJ, Kiyak AH, Ramsay DS, Bollen AM, Anderson NK, et al. Methods to evaluate prole preferences for the anteroposterior position of the mandible. Am J Orthod Dentofacial Orthop 2006;130:283-91. 39. Ricketts RM. The biological signicance of the divine proportion and Fibonacci series. Am J Orthod 1982;81:351-70. 40. Olds C. Facial beauty in Western art. In: McNamara JA, editor. Esthetics and the treatment of facial form. Monograph 28. Craniofacial Growth Series. Ann Arbor: Center for Human Growth and Development; University of Michigan; 1993. 41. Naini FB, Moss JP, Gill DS. The enigma of facial beauty: esthetics, proportions, deformity, and controversy. Am J Orthod 2006;130:277-82. 42. Euklid. The Elements, Books I-XIII. Darmstadt, Germany: Wissenschaftliche Buchgesellschaft; 1991. 43. Powell N, Humphries B. Proportions of the aesthetic face. New York: C.M. Thieme-Stratton; 1984. 44. Clark K. Leonardo Da Vinci. Revised ed. New York: Penguin Books; 1988. 45. Panofsky E. Du rer as a theorist of art. In: The life and art of Albrecht Du rer. 4th ed. Princeton, NJ: Princeton University Press: 1955.

46. Gwilt J. The architecture of Marcus Vitruvius Pollio. London: Priestly and Weale; 1826. 47. Alley TR, Cunningham MR. Averaged faces are attractive, but very attractive faces are not average. Psychol Sci 1991;2:123-35. 48. Langlois JH, Roggman LA. Attractive faces are only average. Psychol Sci 1990;1:115-21. 49. Langlois JH, Roggman LA, Musselman L. What is average and what is not average about attractive faces? Psychol Sci 1994;5: 214-20. 50. Newcombe RG. Averageness as the ideal. J Orthod 2002;29:76-8. 51. Henss R. Perceiving age and attractiveness in facial photographs. J Appl Psychol 1991;21:933-46. 52. Mathes EW, Brennan SM, Haugen PM, Rice HB. Rating of physical attractiveness as a function of age. J Soc Psychol 1985;125: 157-68. 53. Grammer K, Thornhill R. Human (Homo sapiens) facial attractiveness and sexual selection: the role of symmetry and averageness. J Comp Psychol 1994;108:233-42. 54. Rhode G, Roberts J, Simmons L. Reections on symmetry and attractiveness. Psychology, Evolution, and Gender 1999;1:279-95. 55. Sarver DM. The importance of incisor positioning in the esthetic smile: the smile arc. Am J Orthod Dentofacial Orthop 2001;120: 98-111. 56. Sarver DM, Ackerman MB. Dynamic smile visualization and quantication: part 1. Evolution of the concept and dynamic records for smile capture. Am J Orthod Dentofacial Orthop 2003;124: 4-12. 57. Hulsey CM. An esthetic evaluation of lip-teeth relationships present in the smile. Am J Orthod 1970;57:132-44. 58. Is xksal E, Hazar S, Akyalc n S. Smile esthetics: perception and comparison of treated and untreated smiles. Am J Orthod Dentofacial Orthop 2006;129:8-16. 59. Vanderburg N. The horizontal smile line in adolescent malocclusions. IADR abstract 2583; 2004. 60. Johnston DJ, Hunt O, Burden CDE, Stevenson M, Hepper P. The inuence of lower face vertical proportion on facial attractiveness. Eur J Orthod 2005;27:349-54. 61. Shaee R, Korn EL, Pearson H, Boyd RL, Baumrind S. Evaluation of facial attractiveness from end-of-treatment facial photographs. Am J Orthod Dentofacial Orthop 2008;133:500-8. 62. Lundstrom A, Popovich F, Woodside DG. Panel assessments of the facial frontal view as related to mandibular growth direction. Eur J Orthod 1989;11:290-7.