Keywords
surface acquisition systems - 3D faces - 3D imaging - 3D analysis
Introduction
Over the years, many methods have been devised to capture three-dimensional (3D) facial
morphologies.[1]
[2]
[3]
[4] Most of the time, development ran concurrently with technology and from anthropological
studies to X-rays and finally 3D imaging was how development occurred.[5]
[6]
[7] At present, 2D X-ray imaging is still the mainstay of today’s diagnosis, but 3D
techniques are rapidly changing these.[8]
[9] The study of facial form normally uses lines and angles from a cephalogram. Very
few studies have discussed the 3D facial forms of subjects of African descent.[10]
[11]
African Cephalometric Norms
Connor and Moshiri in the mid-80s studied cephalograms of Caucasian and African American
(AA) people for orthognathic profile preference and found significant variation in
various landmarks such as the maxillary and mandibular jaw landmarks, mandibular length,
and the Wits appraisal. They performed other soft tissue analysis like teeth display,
upper lip length, lower lip length, throat length, and lip chin throat angle.[12] Another study by Flynn et al showed that AA had more maxillary skeletal prognathism,
increased lower facial height, increased skeletal facial convexity, lower incisor
proclination, increased upper and lower lip lengths, and increased soft tissue thickness
of lips and chin. These subjects also displayed less nasal depth, projection, and
smaller nasolabial angle in AA people.[13] Another study done by O’Reilly investigated soft tissue profile change in AAs after
dentoalveolar setback and concluded significant variation in horizontal and vertical
lip thickness.[14] As late as 2007 Beukes et al studied South African silhouette pictures and evaluated
most pleasing and unpleasant profiles and established typical soft tissue values.
The most salient feature they noted was upper lip prominence of 5 to 6 mm more than
their AA counterparts; however, other factors such as angular measurements of nose,
lip, and chin were in close proximity to those given by Naidoo and Miles.[15] Last but not least and possibly the most relevant to this study is the work of Dandajena
et al who have cephalometrically analyzed dentoalveolar relations and anterior facial
heights of the Shona people. He found that anterior facial height was higher in men
than in women. The older age groups had shorter anterior face height (AFH) than the
younger group. The AFH of the Shona was lower than that of the AAs. All AFHs for the
men and only TAFHs for the women were significantly shorter for the Shona than the
Caucasians.[16]
Another study by the same author involving the study of lateral cephalograms of 12
angular and 6 linear measurements showed that they had a low Frankfort-mandibular
plane angle with a receding chin. Both the maxilla and mandible sella-nasion to A
point (SNA) and sella-nasion to B point (SNB) were prognathic and the ANB difference
was large. The maxillary incisors were more upright as compared with Caucasian people
while measuring the maxillary incisor to NA; however, the lower incisor to mandibular
plane angle (IMPA) were relatively proclined at 105.8 ± 6.0 degrees and this proclination
was considered to be compensatory to the prognathic maxilla.[17]
Soft Tissue Hard Tissue Paradigm
The specialty of orthodontics has been based on the angle paradigm that involved treating
the hard tissues only, getting the teeth into perfect occlusion and allowing soft
tissue to follow. The soft and hard tissue are an intricate paired phenomenon, where
by one influences the other in an ever constant dynamic interaction.[18] Teeth usually occupy a neutral zone between muscular intra oral tissues such as
the tongue and extra oral soft tissue like the muscles of facial expression and their
whole related complex.[19] Treating only the hard tissues over the years created dental relapse due to displacement
from the neutral desired positions. However, the lips in general are also affected
by a change in lip thickness and also possible lip lengthening due to the constant
pull of gravity. As a result, it is important to consider lip support during extraction
versus nonextraction decision.[20]
The specific aims of the study were to determine the average morphology of the Zimbabwean
(Zim) people both male and female in a specific age group and compare facial morphological
differences of the Zim people to that of AA norms in Birmingham, Alabama, of a similar
age.
Materials and Methods
Subjects were selected from the capital city Harare, Zimbabwe, and Birmingham, Alabama,
United States. Both sample sizes were randomly selected from a specific age group
from the surrounding metropolitan area.
A questionnaire was handed out to determine demographical origin and other inclusion
criteria that include are as follows:
-
Subjects between the age of 18 to 30 years,
-
Ethnicity of Shona descent,
-
Subjects had no adverse skeletal deviations,
-
Subjects with no history of craniofacial deformation,
-
Body mass index within an acceptable reading,
-
No previous orthognathic surgery,
-
No history of orthodontics.
Imaging System
The 3dMDfaceTM system used for this comprised two infrared cameras and one color camera
that casts a random light pattern on the object. The image was captured from a set
series of angles to represent the image in the format of a 3D shell. Capture time
was 1.5 milliseconds with an accuracy of 1.5%.[21]
Image Acquisition
All subjects rested their head position in a natural way because it is the most attainable
and reproducible. The subjects sat on a chair that could be adjusted for the right
position and were asked to gaze into a mirror set in front of them. Their eyes were
levelled, by looking into the mirror and stool adjusted according to their heights
as required. The subjects were told to swallow and keep jaws relaxed just before image
capture.
Processing of Facial Shells
All images imported to Rapidform software 2006, INUS Technology, Seoul, Korea (RF6)
for analysis. Areas such as the hair, ears, neck, and shoulders were removed by a
computer tool and the surface defects filled in. One facial shell for each individual
was created.[22]
Average Face Construction
Each subject’s shell was carefully created and they were aligned and to the “best-fit
algorithm.” Subsequently, the shells were averaged by a computer algorithm that produced
an average shell, one for the females and the other for males.
The steps required to produce and average are summarized as follows:
-
All images were aligned to form the principal axis of rotation, also known as the
center of mass.
-
Manual alignment can also be used to improve the position.
-
The built-in “algorithm” in RF6 calculated the best fit for the shells precise registration.
-
The 3D datapoints of the images were averaged based on a facial template.
-
The resulting output of the average algorithm gave a point cloud reading, which was
then mathematically triangulated to obtain an average face.
-
Filling in or snubbing any mesh defects improves the average faces.
-
A color texture was applied onto the facial shell. In addition, shells with one positive
and one negative deviation were created.
Parameters Measured
Two average facial shells were generated: Zimbabwean male (Zim-M), Zimbabwean female
(Zim-F). These shells were carefully aligned on each other using the RF6 software.
A specialized technique to compare morphological differences was used. This process
comprised manually aligning five points of the facial scans, two points on outer canthus
of both eyes and two points on the inner canthus of both eyes, and one point on the
tip of the nose. The RF6 software then determines the best fit of the two shells.
The parameters used to study these results were as follows:
-
Linear measurements in millimeter, both absolute and signed, which measure the greatest
difference between and within the shells, respectively.
-
Color histograms as a percentage, which show similarities, positives and negatives.
-
Surface areas and shapes, which show range of distribution.
Linear Measurements
The differences are measured by a linear measurement at any given point of the facial
topography. The linear difference is the discrepancy between the two shells and the
value is used to quantify the deviation. A percentage similarity can also be calculated.
Color Histograms
The two shells are compared with each other with one being the base and the other
a superimposition. A positive difference is noted if the superimposed shell is more
positive than the base and a negative difference is noted if the superimposed shell
is more deficient than the base. The differences are identified by different color
depictions: black indicating similarity between the shells, red indicating a positive
change (prominence), and blue indicating a negative change (deficient). This allows
us to distinguish and analyze surface topography and its differences.
Surface Area and Shapes
This is automatically generated by the RF6 software with a tolerance level of 0.50
mm that was offset to the paired surface shells. The value here was obtained from
previous work that shows that 90% of created composite scans are within the 0.85 mm
error. Any variations within 0.50 mm were considered to be similar surfaces, while
surface areas outside this tolerance showed up as a color map of blues and reds.
Results
The final sample was 301 subjects. These included 201 people: 107 men and 94 women
from Zimbabwe. Average faces were constructed for each of the groups and compared
with each other and to that of AAs in Birmingham, Alabama (50 men and 50 women) ([Figs. 1] and [2]).
Fig. 1 Average facial constructions for the Zimbabwean males (row 1) and the Zimbabwean
females (row 2).
Fig. 2 Average African American males (row 1) and African American females (row 2).
Linear Measurements
The absolute linear measurement in gender differences ranged from 0.42 (AA-M vs. AA-F)
to 1.24 mm (Zim-F vs. Zim-M) as shown in [Table 1]. Further differences between Zim and AA are listed below. The maximum distance of
5.09 mm was seen in (Zim-F vs. Zim-M), while the minimum of 2.02 mm was seen in (Zim-M
vs. AA-M).
Table 1
Absolute linear measurements indicating differences between facial shells
|
Average distance (mm)
|
Standard deviation (mm)
|
Maximum distance (mm)
|
|
Abbreviation: AA, African American.
|
|
Zim-M vs. Zim-F
|
1.24
|
1.20
|
5.09
|
|
AA-F vs. Zim-F
|
1.16
|
0.87
|
4.21
|
|
Zim-F vs. AA-M
|
1.22
|
0.96
|
4.47
|
|
Zim-M vs. AA-M
|
0.51
|
0.41
|
2.02
|
|
AA-F vs. Zim-M
|
1.04
|
0.92
|
5.01
|
|
AA-F vs. AA-M
|
0.42
|
0.35
|
2.99
|
Color Histograms
The differences in color histograms between Zim-M and Zim-F are shown in [Table 2]. The % similarity is 27.08%. The greatest amount of similarity is seen between the
Zim-M versus AA-M at 58.75%, which is quite significant. The AA-M versus AA-F too
had a high similarity of 57.57%. The Zim-F sample seems to be the most dissimilar
group when compared with their male counterparts and also to that of the AA-M and
AA-F. The results in the signed color histograms show the similarities between the
same sex individual of different ethnicities but very dissimilar among their own ethnic
groups ([Figs. 3]
[4]
[5]
[6]
[7]
[8]).
Table 2
Signed color map measurement indicating differences in facial shells
|
Average distance
(mm)
|
Standard deviation
(mm)
|
Percentage
Similarity
|
|
Abbreviation: AA, African American.
|
|
Zim-M vs. Zim-F
|
0.35
|
1.62
|
27.08
|
|
AA-F vs. Zim-F
|
0.39
|
1.40
|
25.79
|
|
Zim-F vs. AA-M
|
0.55
|
1.45
|
28.79
|
|
Zim-M vs. AA-M
|
0.02
|
0.66
|
58.75
|
|
AA-F vs. Zim-M
|
0.04
|
1.39
|
36.19
|
|
AA-F vs. AA-M
|
–0.07
|
0.54
|
57.57
|
Fig. 3 Absolute and signed histograms of Zim-M versus Zim-F. The differences indicate that
average shells are uniquely different and represent each individual group. Zim-F,
Zimbabwean females; Zim-M, Zimbabwean males.
Fig. 4 Absolute and signed histogram of AA-F versus Zim-F. The average distance of the differences
was 1.16 ± 0.87 mm. The absolute difference ranged from 0 to 4.21 mm. Positive differences
are seen in red and negative differences in blue. The average difference was 0.39
± 1.40 mm and 25.79% of the faces were similar. AA-F, African American females; Zim-F,
Zimbabwean females.
Fig. 5 Absolute and signed histogram of Zim-F versus AA-M. AA-M, African American males;
Zim-F, Zimbabwean females.
Fig. 6 Absolute and color histogram of Zim-M versus AA-M. The average distance of the differences
was 0.51 ± 0±.41 mm. The absolute difference ranged from 0 to 4.47 mm. Positive differences
are seen in red and negative differences in blue. The average difference was 0.02
± 0.66 mm and 58.75% of the faces were similar. AA-M, African American males; Zim-M,
Zimbabwean males.
Fig. 7 Absolute and color histogram of AA-F versus Zim-M. AA-F, African American females;
Zim-M, Zimbabwean males.
Fig. 8 Absolute and color histogram of AA-F versus AA-M. The differences indicate that average
shells are uniquely different and represent each individual group. AA-F, African American
females; AA-M, African American males.
Discussion
Very few studies on 2D analysis have been done on native southern African people,
let alone 3D. Isiekwe et al studied the nose prominence relative to other structures
of an adult Nigerian population, using the Holdaway analysis.[23] A few photogrammetric studies of North African countries, such as Senegalese students
versus Moroccans students where linear measurements were taken to quantify their differences.[24] Another Sudanese study used a hand laser scanner in 653 subjects and analyzed 14
landmarks on the facial soft tissue.[25] Soft tissue cephalometric studies on Nigerians, Ghanaians, and Sudanese teenagers
were done to determine soft tissue pattern and compared with that of Caucasian counterparts.
Significant statistical difference was apparent.[26] Only one soft tissue profile study was done in South Africa to establish a profile
index for bimaxillary protrusion and soft tissue preference.[15]
Zimbabwean Male and Female Morphology
When looking at the color histograms, the Zim-M show a more protrusive forehead, supraorbital,
and nasal bridge area while being retrusive in the infraorbital and lateral alar region.
They also showed accentuated perioral region. The similarities between the two sexes
are the lateral parts of the nose, subnasal area superior portion of the lip extending
down to the corner of the mouth toward the parasymphysis area.
Comparison between Population Subgroups
When comparing histograms of the same sexes but of different population groups, the
similarities for the males were considerably higher than the females. The similarities
were in the inner canthus of the eyes, lateral nasal region, submental and lower borders
of the chin. The noticeable difference between them were nasal and perioral areas
that showed a more retrusive region, while the malar/zygomatic and lateral periorbital
ridges were more prominent. The females between the two subgroups were more dissimilar
with the forehead, inner canthus of the eyes, nasal bridge, upper and lower lips being
more prominent, while the lateral aspects of the face that is the malar region extending
down to the mandible seem more retrusive.
Clinical Implications
It is apparent that all along we have been diagnosing and categorizing various ethnic
groups into a general pool of Caucasian norms.[27]
[28]
[29] The current trend of treating facial profile first and dentition to follow has taken
great precedence. Understanding soft tissue norms first and then comparing them to
other average shells give us a perspective on variations. This allows us to plan incisor
positions according to average profiles for that age group and it also gives us a
basis to form data on which we can build a foundation to understand trends among ethnicities.
It is obvious in this study that despite the similarity in African descent in the
two populations that significant differences do exist for each of the facial averages.
It is, therefore, important as always to create normative data for each individual
population or race and to make it also specific to gender.
This study uses 3D averages to differentiate between two distinct populations. It
uses new technologies for the better understanding of facial differences and represents
a new way to quantify populations. While previous studies have been used in AAs,[10]
[11] this is the first time this type of analysis is used. Furthermore, in recent times,
genomic studies[29]
[30] have also been introduced to better quantify data obtained from 3D faces and these
will play a more significant role in the era of 3D facial imaging.[31]
Conclusion
The following conclusions may be drawn from this study.
-
When compared with the AA-M, the Zim-M have a prominent zygomatic, lateral supraorbital,
and infraorbital region. Bridge of the nose and perioral area around the lips were
also retrusive.
-
When compared with the AA-M, the Zim-M show a more protrusive forehead, supraorbital,
and nasal bridge area while being retrusive in the infraorbital and lateral alar region.
They also showed accentuated perioral region.
-
When compared with AA-M, the Zim-M show wider supraorbital regions but smaller more
retruded perioral regions.
-
When compared with AA-F, the Zim-F show a subtler malar region with less prominent
lips.
-
When compared with Zim-F, the AA-F are more protrusive in the middle forehead area,
alar base, and lateral perioral region, retrusive in the lateral zygomatic region,
gonial, and submental area.
