RECONSTRUCTIVE Reliability and Reproducibility of Landmarks on Three-Dimensional Soft-Tissue Cephalometrics Using Different Placement Methods Han Lin, B.S. Ping Zhu, M.D. Yi Lin, B.S. Yuxi Zheng, B.S. Yue Xu, M.D., Ph.D. Guangdong, People’s Republic of China; and Cleveland, Ohio
Background: Affixing markers on the face in vivo with an optical surface imaging system and placing landmarks on a computed tomographic reconstructive facial model represent two helpful approaches prevailing in three-dimensional cephalometric analysis of facial aesthetics. In this study, the authors determine the suitability of these methods for soft-tissue evaluation along with the reproducibility and reliability of landmark placement. Methods: Thirty-seven soft-tissue landmarks were investigated in 35 normal healthy volunteers who underwent cephalometric analysis by direct and indirect placement methods. Two operators performed the analysis twice for each method at a 1-week interval. Landmark positions were measured on the three-dimensional coordinates, and data were standardized and converted into landmark placement errors for the estimation, by two-way random intraclass correlation coefficients. Results: Using the direct method, 86.5 percent of the landmarks had a higher intraclass correlation coefficient than 0.75, and 67.6 percent were higher than 0.90. The authors found that 75.7 percent were higher than 0.75 and 43.2 percent were higher than 0.90 with the indirect method. Both methods showed good reliability in identifying midline-related structures, although the correlations of markers surrounding eyelashes and lips were better using the direct method. The direct method had a significantly smaller landmark placement error and better reproducibility than that identified indirectly (p < 0.05). Conclusions: When evaluating well-defined contours and the natural texture of the face, the direct placement method shows better accuracy than the indirect method. The three-dimensional evaluation emphasizes the eye, lip, and some structures more laterally, therefore providing comprehensive analysis for clinical diagnosis. (Plast. Reconstr. Surg. 134: 102e, 2014.) CLINICAL QUESTION/LEVEL OF EVIDENCE: Diagnostic, II.
S
oft-tissue adaptations and harmony are the main goals of modern orthodontic and orthognathic treatments.1–3 The soft-tissue morphology may not be fully expressed by the hard-tissue structure because of its compensation. Thus, increased focus on facial aesthetics evaluation rather than examination of tooth position From the Department of Orthodontics, Guanghua School of Stomatology, Hospital of Stomatology, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Stomatology; and Case Western Reserve University. Received for publication September 4, 2013; accepted December 12, 2013.
The first two authors contributed equally to this work and should be considered co–first authors. Copyright © 2014 by the American Society of Plastic Surgeons DOI: 10.1097/PRS.0000000000000279
102e
leads to different and helpful approaches to obtaining diagnostic information.4 It has been ascertained that accurate facial appearance evaluation relies on precise measurement. Direct anthropometry has been viewed as the criterion standard for facial soft-tissue assessment,5 although it is difficult to apply in the clinic. Landmark placement on the face in vivo is time consuming because it involves complicated manual measurement.4 After Hounsfield’s introduction of a three-dimensional technique for human imaging purposes in the 1970s,6 three-dimensional scanning techniques became more acceptable Disclosure: The authors have no financial interest in any of the products or devices mentioned in this article.
www.PRSJournal.com
Volume 134, Number 1 • Direct versus Indirect Cephalometrics among orthodontists than conventional cephalography. Using computed tomographic scanning– based indirect measurements, some bone-related soft-tissue landmarks can be identified because of a visualization of the underlying hard tissue.7 It is known that reliable and accurate landmark location is the foundation of a convincing measurement analytic system. Previous studies have shown that the principal errors in cephalometric analysis lie in the method of locating landmarks.8–10 The pattern of error varies in every landmark.11,12 The facial measurement results have been proved to be useful only if the landmarks are placed accurately.10 Several studies have been conducted establishing different soft-tissue analytic approaches since the advent of the “facial harmony theory” put forward by Angle.13 Systemic errors caused by the imaging acquisition process itself have not, however, been reported. Here, we introduce a direct method by affixing markers to the in vivo face and measuring them using a computer-aided system. By investigating the reproducibility and reliability of facial soft-tissue landmarks using a direct method and an indirect placement method based on computed tomographic facial reconstructed models, we aim to provide a guide for the suitable selection of landmarks and corresponding placement methods for diagnostic and treatment purposes in the orthodontic clinic.
PATIENTS AND METHODS A total of 35 normal healthy volunteers (18 men and 17 women aged 24.9 ± 2.6 years) were enrolled in the study. They had no facial asymmetry, protrusion, craniofacial disorders, or history of orthodontic treatments. All of the procedures were explained before data collection. The volunteers were informed of the experimental design and consented to the study, which was approved by the Sun Yat-sen University Ethical Committee. Selection of Landmarks A total of 37 soft-tissue landmarks most commonly applied in the profile assessment5,10,14 were collected (Table 1 and Fig. 1). They include five midline landmarks (pronasale, sellion, labiale superius, labiale inferius, and soft-tissue gnathion) and 16 paired landmarks (right and left side indicated as r and l, respectively): lateral ciliary, interciliary, exocanthion, endocanthion, upper eyelid, maxillary point, soft-tissue zygion, orbitale superius, soft-tissue orbitale, alar curvature, upper lip, cheilion, lower lip, soft-tissue gonion, middle
jaw, and cheek point. Four further markers (indicated as tr1, tr2, tr3, and tr4) on a headband were affixed to the forehead of a subject’s head along the hairline and with the front middle point of the headband bisecting the distance between the bilateral tragus. These four constantly positioned head landmarks were used to create a local coordinate system. This defined a head plane of reference that was used to standardize head positions in and between subjects. Figure 2 shows the defined local coordinate system on the head. Direct Method: Affixing Landmarks on the In Vivo Subject’s Face For each subject, the operator located a set of 37 small reflective markers (2-mm diameter; NaturalPoint, Inc., Corvallis, Ore.) by inspection and/ or palpation, and affixed them onto the cutaneous surface. Before the recording procedure, volunteers were asked to wash their faces, to reduce any chemical substances that might interfere with the landmark placement. During landmark placement, the subjects sat relaxed on a stool in the center of the working volume, with their heads in a natural position. To acquire the three-dimensional coordinates of the facial soft-tissue landmarks, a NaturalPoint facial motion capture system (Arena Expression; NaturalPoint) was used. The system captured the location of these small reflective markers within 1 second using six infrared cameras (V100:R2, OptiTrack; NaturalPoint). Before each data collection session, the instrument was calibrated. The calibration process allows the definition of a global reference frame for all the cameras used in the optical tracking system. Once the extrinsic and intrinsic parameters for each particular camera were computed, the relative position of these markers inside the working region was known. After the landmarks were captured, the threedimensional coordinates were converted in the local coordinate system created by four head landmarks and subsequently exported. Indirect Method: Marking on Three-Dimensional Reconstructed Computed Tomographic Model (Cone Beam Computed Tomography–Based Three-Dimensional Cephalometry) Computed tomographic images were obtained by means of a DCT Pro cone beam computed tomography device (Vatech, Co., Ltd., Hwasung, Republic of Korea) using the following scanning parameters: 90 kVp; 24 seconds;4 mA; voxel size, 0.4 mm; and field of view, 20 × 19 cm. The image covered the area from the highest
103e
Plastic and Reconstructive Surgery • July 2014 Table 1. Landmark Descriptions Landmark Midline landmarks Sellion Pronasale Labiale superius Labiale inferius Soft-tissue gnathion Paired landmarks: (right and left side noted r and l) Left lateral ciliary Right lateral ciliary Left interciliary Right interciliary Exocanthion (l)
Abbreviation se prn ls li st gn
Lateral ciliaryl Lateral ciliaryr interciliaryl interciliaryr exl
Exocanthion (r)
exr
Endocanthion (l)
enl
Endocanthion (r)
enr
Left upper eyelid Right upper eyelid Left maxillary point
u eyelipl u eyelidr maxl
Right maxillary point
maxr
Soft-tissue zygion (l)
st zyl
Soft-tissue zygion (r)
st zyr
Orbitale superius (l)
osl
Orbitale superius (r)
osr
Soft-tissue orbitale (l)
orl
Soft-tissue orbitale (r)
orr
Alar curvature (l) Alar curvature (r) Left upper lip Right upper lip Cheilion (l) Cheilion (r) Left lower lip Right lower lip Soft-tissue gonion (l)
acl acr u lipl u lipr chl chr l lipl l lipr st gol
Soft-tissue gonion (r)
st gor
Left middle jaw
middle jawl
Right middle jaw
middle jawr
Left cheek point
cheekl
Right cheek point
cheekr
104e
Definition The most posterior point of the frontonasal soft-tissue contour in the midline of the base of the nasal root The most anterior midpoint of the nasal tip The midpoint of the vermilion line of the upper lip The midpoint of the vermilion line of the lower lip The most inferior midpoint of the soft-tissue contour of the chin located at the level of the three-dimensional cephalometric hardtissue menton landmark Located above the most lateral aspect of the left eyebrow Located above the most lateral aspect of the right eyebrow Located above the medial aspect of the left eyebrow Located above the medial aspect of the right eyebrow The soft-tissue point located at the outer commissure of the left eye fissure The soft-tissue point located at the outer commissure of the right eye fissure The soft-tissue point located at the inner commissure of the left eye fissure The soft-tissue point located at the inner commissure of the right eye fissure Superior midportion of the upper left eyelid Superior midportion of the upper right eyelid Located on the cheek one-quarter of the distance between the left alar and left temporomandibular joint Located on the cheek one-quarter of the distance between the right alar and right temporomandibular joint The most lateral point on the soft-tissue contour of the left zygomatic arch, located at the level of the three-dimensional hard-tissue cephalometric zygion landmark The most lateral point on the soft-tissue contour of the right zygomatic arch, located at the level of the three-dimensional hard-tissue cephalometric zygion landmark The most superior soft-tissue point of the lower border of the left brow, located at the most superior level of the left supraorbital rim The most superior soft-tissue point of the lower border of the right brow, located at the most superior level of the right supraorbital rim The soft-tissue point located at the most inferior level of the left infraorbital rim, located at the level of the three-dimensional hardtissue cephalometric orbitale landmark The soft-tissue point located at the most inferior level of the right infraorbital rim, located at the level of the three-dimensional hardtissue cephalometric orbitale landmark The point located at the facial insertion of the left alar base The point located at the facial insertion of the right alar base Left upper lip point halfway between the cheilion and labiale superius Right upper lip point halfway between the cheilion and labiale superius The point located at the left labial commissure The point located at the right labial commissure Left lower lip point halfway between the cheilion and labiale inferius Right lower lip point halfway between the cheilion and labiale inferius The most lateral point on the soft-tissue contour of the left mandibular angle, located at the same level as the three-dimensional hard-tissue cephalometric gonion landmark The most lateral point on the soft-tissue contour of the right mandibular angle, located at the same level as the three-dimensional hardtissue cephalometric gonion landmark Left middle jaw point halfway between the soft-tissue gnathion and the soft-tissue gonion, along the inferior margin of the jaw Right middle jaw point halfway between the soft-tissue gnathion and the soft-tissue gonion, along the inferior margin of the jaw Left cheek point located on the cheek one-quarter of the distance between the left commissures and the left temporomandibular joint Right cheek point located on the cheek one-quarter of the distance between the right commissures and the right temporomandibular joint
Volume 134, Number 1 • Direct versus Indirect Cephalometrics
Fig. 1. Landmarks used for three-dimensional facial soft-tissue analysis. A total of 37 soft-tissue landmarks were applied in the assessment, including five midline landmarks (se, prn, ls, li, and st gn) and 16 paired landmarks.
point on the superior margin of the cranial vault (vertex) to the inferior border of the mandibular body. The gross data and the obtained slices were imported and reconstructed into a three-dimensional model by an interactive image analysis system (Mimics, 14.0; Materialise, Leuven, Belgium). The software automatically determined the origin of the coordinates (x = 0, y = 0, and z = 0) at the left anterior lower corner of the cube contained in the three-dimensional image. As the coordinate system had been standardized and uniquely defined, the spatial positions of each landmark were represented as numerical values (expressed in millimeters) on each axis. Landmarks were designed on the three-dimensional surface model and their positions verified in multiple planar reformat modes.
with an interval of 1 week between the two measurements for each method.
Comparison between the Two Placement Methods Two trained and qualified observers with 7 years’ experience in orthodontics analyzed the reliability and reproducibility of both placement methods. They independently performed the measurements using the placement methods,
where D is the total difference, Δx = xa − x b x + x b x ′a + x ′ b or a − in intraobserver or interob2 2 server error calculation and representing the difference in the x axis, and the same for Δy and Δz.15 Interobserver reliability for landmark placement errors in both direct and indirect measurement
Statistical Analysis Statistical analysis was performed using SPSS Version 16 (SPSS, Inc., Chicago, Ill.). The intraobserver error in each landmark was expressed as a distance between two measurements in three-dimensional coordinates for each placement method. The interobserver error was expressed as a distance between the averages of each observer’s two sets of three-dimensional coordinates of the same point. The equation is as follows: D=
( x ) ^ 2 + ( y ) ^ 2 + ( z ) ^ 2↵ ,
105e
Plastic and Reconstructive Surgery • July 2014
Fig. 2. The direct method, affixing landmarks on the in vivo subject’s face. A set of 37 small reflective markers were affixed on the cutaneous surface of the subject and then captured by six infrared cameras. The three-dimensional coordinates of markers were translated in a local coordinate system created by four head landmarks and subsequently exported. The 16 paired landmarks are illustrated with purple dots, the five midline landmarks are in green, and the four head landmarks are in gray.
methods was assessed by two-way random intraclass correlation coefficients with absolute agreement and their 95 percent confidence intervals. Reproducibility was defined as the mean of the total differences between two sets of measurement for interobserver or intraobserver reproducibility. Paired t tests were used to examine significant differences in the average landmark placement errors between observers and between the two placement methods for each observer. Values of p < 0.05 were considered statistically significant. The average errors were categorized as follows: less than 1 mm, highly precise; 1 to 1.5 mm, precise; 1.5 to 2 mm, moderately precise; and greater than 2 mm, imprecise.16
RESULTS The intraclass correlation coefficients and 95 percent confidence intervals for both landmark placement methods are listed in Table 2. An intraclass correlation coefficient of 0.75 or above is usually considered to be good and that above 0.9 to be excellent.17 For the direct placement method, 32 landmarks (86.5 percent) had an intraclass correlation coefficient higher than 0.75. Among them, 25 landmarks (67.6 percent) had an intraclass correlation coefficient higher than 0.90. For the indirect placement
106e
method, 28 landmarks (75.7 percent) had an intraclass correlation coefficient higher than 0.75, and 16 (43.2 percent) had an intraclass correlation coefficient higher than 0.90. Figure 3 shows the intraobserver and interobserver average landmark placement error distribution for each placement method as color maps. Areas of smaller landmark placement errors were mainly located close to the midline of the face. Landmark placement errors increased with increasing distance from the midline, symmetrically. A significant difference was found between the mean landmark placement errors of the two placement methods (Table 3). The mean landmark placement errors using the direct placement method was 1.093 ± 0.518 mm for observer 1 and 1.101 ± 0.642 mm for observer 2, and the interobserver variability was 1.259 ± 0.566 mm. These findings were all significantly smaller than those obtained by means of the indirect placement method (observer 1, 1.301 ± 0.653 mm; observer 2, 1.340 ± 0.667 mm; interobserver variability, 1.420 ± 0.638 mm). The placement error of several landmarks was less than 1 mm, and only 8 to 17 percent of the landmarks using the direct method had more than 2-mm errors, compared with 16 to 27 percent for the indirect method (Table 4).
Volume 134, Number 1 • Direct versus Indirect Cephalometrics Table 2. Interobserver Reliability of Landmark Placement Errors Using Direct and Indirect Placement Methods Mean Intraclass Correlation Coefficients (95% CI) Landmarks Midline landmark Sellion Pronasale Labiale superius Labiale inferius Soft-tissue gnathion Paired landmarks Left lateral ciliary Right lateral ciliary Left interciliary Right interciliary Exocanthion (l) Exocanthion (r) Endocanthion (l) Endocanthion (r) Left upper eyelid Right upper eyelid Left maxillary point Right maxillary point Soft-tissue zygion (l) Soft-tissue zygion (r) Orbitale superius (l) Orbitale superius (r) Soft-tissue orbitale (l) Soft-tissue orbitale (r) Alar curvature (l) Alar curvature (r) Left upper lip Right upper lip Cheilion (l) Cheilion (r)
Direct Placement Method
Indirect Placement Method
0.971 (0.946–0.986) 0.966 (0.938–0.984) 0.974 (0.943–0.989) 0.96 (0.935–0.985) 0.927 (0.901–0.964)
0.957 (0.916–0.983) 0.943 (0.941–0.992) 0.931 (0.907–0.959) 0.945 (0.922–0.979) 0.88 (0.627–0.947)
0.911 (0.884–0.946) 0.904 (0.757–0.962) 0.9 (0.745–0.945) 0.912 (0.813–0.934) 0.955 (0.912–0.983) 0.96 (0.937–0.986) 0.973 (0.946–0.986) 0.977 (0.953–0.993) 0.944 (0.894–0.967) 0.943 (0.901–0.978) 0.921 (0.887–0.953) 0.913 (0.884–0.943) 0.69 (0.414–0.937) 0.62 (0.397–0.946) 0.833 (0.712–0.934) 0.851 (0.610–0.937) 0.951 (0.918–0.986) 0.942 (0.904–0.978) 0.962 (0.928–0.991) 0.967 (0.945–0.987) 0.952 (0.921–0.983) 0.937 (0.913–0.974) 0.971 (0.948–0.991) 0.976 (0.955–0.993)
0.798 (0.501–0.913) 0.833 (0.528–0.927) 0.84 (0.511–0.907) 0.844 (0.557–0.914) 0.963 (0.921–0.981) 0.933 (0.913–0.954) 0.975 (0.938–0.990) 0.973 (0.933–0.989) 0.917 (0.887–0.953) 0.955 (0.932–0.974) 0.84 (0.563–0.939) 0.872 (0.518–0.908) 0.75 (0.310–0.926) 0.74 (0.486–0.942) 0.86 (0.625–0.921) 0.834 (0.633–0.927) 0.913 (0.885–0.946) 0.935 (0.901–0.958) 0.963 (0.945–0.986) 0.965 (0.933–0.984) 0.882 (0.568–0.931) 0.887 (0.694–0.945) 0.921 (0.886–0.938) 0.918 (0.897–0.952) (Continued )
Table 2. (Continued) Mean Intraclass Correlation Coefficients (95% CI) Landmarks Left lower lip Right lower lip Soft-tissue gonion (l) Soft-tissue gonion (r) Left middle jaw Right middle jaw Left cheek point Right cheek point
Direct Placement Method
Indirect Placement Method
0.948 (0.913–0.987) 0.957 (0.924–0.978) 0.81 (0.624–0.936) 0.85 (0.577–0.943) 0.8 (0.601–0.945) 0.83 (0.572–0.942) 0.8 (0.633–0.931) 0.74 (0.531–0.938)
0.873 (0.678–0.923) 0.867 (0.655–0.942) 0.58 (0.217–0.908) 0.63 (0.396–0.918) 0.58 (0.398–0.894) 0.56 (0.23–0.910) 0.74 (0.382–0.915) 0.72 (0.437–0.874)
l, left; r, right.
DISCUSSION Many researchers have attempted to develop suitable analytic approaches since the development of various techniques for the threedimensional documentation of facial surfaces,15 including laser scanning, computed tomography, and stereophotogrammetry.18 Despite the fact that measurement on the reconstructed model improves the accuracy of anthropometry, many researchers evaluated the linear distances or angles,19 emphasizing the errors caused by placement manipulation and anatomical features. Actually, the systemic error resulting from the imaging acquisition should be a prime consideration before any analytic process. Studies aimed at comparing three-dimensional direct and indirect placement method are still absent. In this study, we investigate the reproducibility and reliability of a series of landmarks, using a three-dimensional direct measurement method, and compared them with computed tomographic measurements. This is the first study aimed at comparing the two threedimensional facial soft-tissue evaluation methods. In both direct and indirect measurement methods, landmarks closer to the midline (i.e., nose bridge, sneer line, and mouth corner) are more reliable than those located on the bilateral face, as indicated by a significantly higher intraclass correlation coefficient (>0.9) and smaller landmark placement error (Fig. 3). Correlation decreases with the distance away from the midline, which reflects the complexity of facial soft tissue. As these midline structures have distinguishable anatomical features (i.e., convex or
107e
Plastic and Reconstructive Surgery • July 2014
Fig. 3. Color maps of landmark placement error distribution in the face mask. (Above) Observer 1, observer 2, and interobserver errors in the direct placement method. (Below) Observer 1, observer 2, and interobserver errors in the indirect placement method.
concave), locating landmarks is easier. Gwilliam et al.18 demonstrated that the surface contours might affect precision and reliability in anthropometry. Good identification in irregular structures by means of a comprehensive observation from different angles could be obtained by deft control of the three-dimensional reconstruction through rotation, translation, and magnification of the images in computed tomography,10 or by sensitive visual observation under natural light in the direct method. It has been suggested that the midline-related landmarks are accurate when establishing their position, whether using direct or indirect measurement methods. Referring to the paired landmarks, it was found that landmarks surrounding the eyelashes Table 3. Comparison of Landmark Placement Mean Errors Using Direct and Indirect Placement Methods Landmark Placement Method Direct Indirect p
108e
Landmark Placement Mean Errors (mean ± SD) Observer 1
Observer 2
Interobserver Variability
1.093 ± 0.518 1.301 ± 0.653 0.005
1.101 ± 0.642 1.340 ± 0.667 0.009
1.259 ± 0.566 1.420 ± 0.638 0.050
(i.e., left lateral ciliary, right lateral ciliary, left interciliary, and right interciliary) and lips (i.e., left upper lip, right upper lip, left lower lip, and right lower lip) showed excellent correlations (>0.90) using the direct method, compared with those found for the computed tomography method (0.75 to 0.90). The larger error and lower precision in computed tomographic marking can be explained by the difficulty in identifying the upper eyelid border or vermilion border in the lack of a colored texture in the three-dimensional reconstructed facial model. Because of the welldefined contours, the direct placement method can identify the eyebrow and lips border by inspection, thereby achieving accurate placement. In addition, the limited accuracy of computed tomography in revealing soft tissues contributes to a larger error. Even with multiangle control of images, the observation of a three-dimensional model from a two-dimensional computer screen cannot reflect reality, unlike direct observation. The registration of the natural texture of the face by advanced three-dimensional computed tomographic reconstruction could be a helpful solution to this problem. Computed tomography, the common imaging technique in orthodontic
Volume 134, Number 1 • Direct versus Indirect Cephalometrics Table 4. Number and Percentage of Landmarks in Relation to the Average Intraobserver and Interobserver Differences, Divided into Four Groups, of 37 Landmarks for Direct and Indirect Placement Methods Reproducibility Level Landmark Placement Method Direct No. % Indirect No. %
Highly Precise (≤1 mm)
Moderately Precise (1.5–2 mm)
Precise (1–1.5 mm)
Imprecise (>2mm)
Ob1
Ob2
Ob1–Ob2
Ob1
Ob2
Ob1–Ob2
Ob1
Ob2
Ob1–Ob2
Ob1
Ob2
Ob1–Ob2
17 45.95
18 48.65
13 35.14
11 29.73
8 21.62
13 35.14
6 16.22
6 16.22
5 13.51
3 8.11
5 13.51
6 16.22
16 43.24
15 40.54
14 37.84
11 29.73
8 21.62
6 16.22
3 8.11
6 16.22
6 16.22
6 16.22
7 18.92
10 27.03
Ob1, observer 1; Ob2, observer 2.
examination, has advantages in overlapping skin and the subcutaneous bone tissues, improving the identification of facial soft-tissue landmark locations. Several errors exist because of a certain distance between the surface of soft and hard tissues. Structural features do not completely match (i.e., the most convex point of soft tissue sometimes does not correspond to the most convex point of underlying hard tissue). Similarly, the location of landmarks representing the soft-tissue gonion and middle jaw is difficult using the indirect method. The intraclass correlation coefficient has been found to be lower than 0.75. The lack of color contrast and the potential for concealment of this area by soft tissue are possible reasons for its poor reliability. In contrast, even the smallest deflection of observation angle could bring false identification of these markers because of their locations at the edge of soft tissue. Reliability is a crucial factor in evaluating a placement method, but reproducibility, assessed by landmark placement error, is also an essential element. The majority of landmarks (>35 percent) in both methods are reproducible to less than 1 mm. The average landmark placement error in the direct methods is lower than in the indirect ones, with values of p = 0.005, p = 0.009, and p = 0.050 for within and between observers, respectively. A higher accuracy in landmark placement is indicated with the direct placement method. Although performing a direct measurement is time-consuming because of affixing landmarks manually on the face, once the landmarks are all set, the capture speed is so fast that even erratic movements by the youngest child or the most restless patient can be ignored. Therefore, the measurements could be more readily obtained. Furthermore, the direct placement method is noninvasive, without exposure to radiation. Images can be taken as frequently as needed, providing a safe strategy for repeated assessment
and testing. The capture of the landmarks pasted directly onto the face is of significance when studying facial kinematics, such as expressions and smile assessment. This is relevant to orthodontic treatment assessment.
CONCLUSIONS Three-dimensional dynamic evaluation of facial aesthetics based on the motion capture technique has the potential to transform the conventional static assessment in plastic surgery and related disciplines. This first study compared the direct and indirect three-dimensional facial evaluation methods and classified the 37 soft-tissue landmarks into four precise grades according to their landmark placement error values. Our findings mainly suggest that direct observation is more accurate in the perception of welldefined contours and natural textures of the face than the indirect method. The direct placement method had significantly lower errors than the indirect method for repeatedly placing markers. Meanwhile, similar reliability in evaluating midline-related structures was found with the two approaches. With the identification of the landmark placement error category of landmarks, we laid a sound basis for optimal selection of landmarks, which results in a convincing facial analytic system. Yue Xu, M.D., Ph.D. Guanghua School of Stomatology Sun Yat-sen University Lingyuan Xilu No.56 Guangzhou 510055, People’s Republic of China [email protected]
acknowledgments
This work was supported by the Projects of Integration of Industry, Education and Research of Guangdong Province (2012B091100453); Sun Yat-sen
109e
Plastic and Reconstructive Surgery • July 2014 University Clinical Research 5010 Program (2007050); and Key Laboratory of Sensing Technology and Biomedical Instruments of Guangdong Province (2011A 060901013). references 1. Goldstein RE. Study of need for esthetics in dentistry. J Prosthet Dent. 1969;216:589–598. 2. Paiva JB, Attizzani MF, Miasiro Júnior H, Rino Neto J. Facial harmony in orthodontic diagnosis and planning. Braz Oral Res. 2010;241:52–57. 3. Spyropoulos MN, Halazonetis DJ. Significance of the soft tissue profile on facial esthetics. Am J Orthod Dentofacial Orthop. 2001;1195:464–471. 4. Germec-Cakan D, Canter HI, Nur B, Arun T. Comparison of facial soft tissue measurements on three-dimensional images and models obtained with different methods. J Craniofac Surg. 2010;21:1393–1399. 5. de Menezes M, Rosati R, Allievi C, Sforza C. A photographic system for the three-dimensional study of facial morphology. Angle Orthod. 2009;79:1070–1077. 6. Swennen GRJ, Schutyser F, Hausamen JE. Three-Dimensional Cephalometry: A Color Atlas and Manual. Berlin: Springer; 2006. 7. Park SH, Yu HS, Kim KD, Lee KJ, Baik HS. A proposal for a new analysis of craniofacial morphology by 3-dimensional computed tomography. Am J Orthod Dentofacial Orthop. 2006;129:600.e23–600.e34. 8. Cevidanes LH, Bailey LJ, Tucker GR Jr, et al. Superimposition of 3D cone-beam CT models of orthognathic surgery patients. Dentomaxillofac Radiol. 2005;34:369–375. 9. Grauer D, Cevidanes LS, Styner MA, et al. Accuracy and landmark error calculation using cone-beam computed tomography-generated cephalograms. Angle Orthod. 2010;802: 286–294.
110e
10. Plooij JM, Swennen GR, Rangel FA, et al. Evaluation of reproducibility and reliability of 3D soft tissue analysis using 3D stereophotogrammetry. Int J Oral Maxillofac Surg. 2009;38:267–273. 11. Lou L, Lagravere MO, Compton S, Major PW, Flores-Mir C. Accuracy of measurements and reliability of landmark identification with computed tomography (CT) techniques in the maxillofacial area: A systematic review. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2007;104:402–411. 12. Muramatsu A, Nawa H, Kimura M, et al. Reproducibility of maxillofacial anatomic landmarks on 3-dimensional computed tomographic images determined with the 95% confidence ellipse method. Angle Orthod. 2008;78:396–402. 13. Proffit WR. Malocclusion and dentofacial deformity in contemporary society. In: Proffit WR, Fields HW Jr, Sarver DM, eds. Contemporary Orthodontics. 5th ed. St. Louis, Mo: Elsevier/Mosby; 2013:2–18. 14. Weeden JC, Trotman CA, Faraway JJ. Three dimensional analysis of facial movement in normal adults: Influence of sex and facial shape. Angle Orthod. 2001;71:132–140. 15. Toma AM, Zhurov A, Playle R, Ong E, Richmond S. Reproducibility of facial soft tissue landmarks on 3D laserscanned facial images. Orthod Craniofac Res. 2009;12:33–42. 16. Mutsvangwa TE, Veeraragoo M, Douglas TS. Precision assessment of stereo-photogrammetrically derived facial landmarks in infants. Ann Anat. 2011;193:100–105. 17. Sayinsu K, Isik F, Trakyali G, Arun T. An evaluation of the errors in cephalometric measurements on scanned cephalometric images and conventional tracings. Eur J Orthod. 2007;29:105–108. 18. Gwilliam JR, Cunningham SJ, Hutton T. Reproducibility of soft tissue landmarks on three-dimensional facial scans. Eur J Orthod. 2006;28:408–415. 19. Zamora N, Llamas JM, Cibrián R, Gandia JL, Paredes V. A study on the reproducibility of cephalometric landmarks when undertaking a three-dimensional (3D) cephalometric analysis. Med Oral Patol Oral Cir Bucal. 2012;17:e678–e688.
Background: Affixing markers on the face in vivo with an optical surface imaging system and placing landmarks on a computed tomographic reconstructive facial model represent two helpful approaches prevailing in three-dimensional cephalometric analysis of facial aesthetics. In this study, the authors determine the suitability of these methods for soft-tissue evaluation along with the reproducibility and reliability of landmark placement. Methods: Thirty-seven soft-tissue landmarks were investigated in 35 normal healthy volunteers who underwent cephalometric analysis by direct and indirect placement methods. Two operators performed the analysis twice for each method at a 1-week interval. Landmark positions were measured on the three-dimensional coordinates, and data were standardized and converted into landmark placement errors for the estimation, by two-way random intraclass correlation coefficients. Results: Using the direct method, 86.5 percent of the landmarks had a higher intraclass correlation coefficient than 0.75, and 67.6 percent were higher than 0.90. The authors found that 75.7 percent were higher than 0.75 and 43.2 percent were higher than 0.90 with the indirect method. Both methods showed good reliability in identifying midline-related structures, although the correlations of markers surrounding eyelashes and lips were better using the direct method. The direct method had a significantly smaller landmark placement error and better reproducibility than that identified indirectly (p < 0.05). Conclusions: When evaluating well-defined contours and the natural texture of the face, the direct placement method shows better accuracy than the indirect method. The three-dimensional evaluation emphasizes the eye, lip, and some structures more laterally, therefore providing comprehensive analysis for clinical diagnosis. (Plast. Reconstr. Surg. 134: 102e, 2014.) CLINICAL QUESTION/LEVEL OF EVIDENCE: Diagnostic, II.
S
oft-tissue adaptations and harmony are the main goals of modern orthodontic and orthognathic treatments.1–3 The soft-tissue morphology may not be fully expressed by the hard-tissue structure because of its compensation. Thus, increased focus on facial aesthetics evaluation rather than examination of tooth position From the Department of Orthodontics, Guanghua School of Stomatology, Hospital of Stomatology, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Stomatology; and Case Western Reserve University. Received for publication September 4, 2013; accepted December 12, 2013.
The first two authors contributed equally to this work and should be considered co–first authors. Copyright © 2014 by the American Society of Plastic Surgeons DOI: 10.1097/PRS.0000000000000279
102e
leads to different and helpful approaches to obtaining diagnostic information.4 It has been ascertained that accurate facial appearance evaluation relies on precise measurement. Direct anthropometry has been viewed as the criterion standard for facial soft-tissue assessment,5 although it is difficult to apply in the clinic. Landmark placement on the face in vivo is time consuming because it involves complicated manual measurement.4 After Hounsfield’s introduction of a three-dimensional technique for human imaging purposes in the 1970s,6 three-dimensional scanning techniques became more acceptable Disclosure: The authors have no financial interest in any of the products or devices mentioned in this article.
www.PRSJournal.com
Volume 134, Number 1 • Direct versus Indirect Cephalometrics among orthodontists than conventional cephalography. Using computed tomographic scanning– based indirect measurements, some bone-related soft-tissue landmarks can be identified because of a visualization of the underlying hard tissue.7 It is known that reliable and accurate landmark location is the foundation of a convincing measurement analytic system. Previous studies have shown that the principal errors in cephalometric analysis lie in the method of locating landmarks.8–10 The pattern of error varies in every landmark.11,12 The facial measurement results have been proved to be useful only if the landmarks are placed accurately.10 Several studies have been conducted establishing different soft-tissue analytic approaches since the advent of the “facial harmony theory” put forward by Angle.13 Systemic errors caused by the imaging acquisition process itself have not, however, been reported. Here, we introduce a direct method by affixing markers to the in vivo face and measuring them using a computer-aided system. By investigating the reproducibility and reliability of facial soft-tissue landmarks using a direct method and an indirect placement method based on computed tomographic facial reconstructed models, we aim to provide a guide for the suitable selection of landmarks and corresponding placement methods for diagnostic and treatment purposes in the orthodontic clinic.
PATIENTS AND METHODS A total of 35 normal healthy volunteers (18 men and 17 women aged 24.9 ± 2.6 years) were enrolled in the study. They had no facial asymmetry, protrusion, craniofacial disorders, or history of orthodontic treatments. All of the procedures were explained before data collection. The volunteers were informed of the experimental design and consented to the study, which was approved by the Sun Yat-sen University Ethical Committee. Selection of Landmarks A total of 37 soft-tissue landmarks most commonly applied in the profile assessment5,10,14 were collected (Table 1 and Fig. 1). They include five midline landmarks (pronasale, sellion, labiale superius, labiale inferius, and soft-tissue gnathion) and 16 paired landmarks (right and left side indicated as r and l, respectively): lateral ciliary, interciliary, exocanthion, endocanthion, upper eyelid, maxillary point, soft-tissue zygion, orbitale superius, soft-tissue orbitale, alar curvature, upper lip, cheilion, lower lip, soft-tissue gonion, middle
jaw, and cheek point. Four further markers (indicated as tr1, tr2, tr3, and tr4) on a headband were affixed to the forehead of a subject’s head along the hairline and with the front middle point of the headband bisecting the distance between the bilateral tragus. These four constantly positioned head landmarks were used to create a local coordinate system. This defined a head plane of reference that was used to standardize head positions in and between subjects. Figure 2 shows the defined local coordinate system on the head. Direct Method: Affixing Landmarks on the In Vivo Subject’s Face For each subject, the operator located a set of 37 small reflective markers (2-mm diameter; NaturalPoint, Inc., Corvallis, Ore.) by inspection and/ or palpation, and affixed them onto the cutaneous surface. Before the recording procedure, volunteers were asked to wash their faces, to reduce any chemical substances that might interfere with the landmark placement. During landmark placement, the subjects sat relaxed on a stool in the center of the working volume, with their heads in a natural position. To acquire the three-dimensional coordinates of the facial soft-tissue landmarks, a NaturalPoint facial motion capture system (Arena Expression; NaturalPoint) was used. The system captured the location of these small reflective markers within 1 second using six infrared cameras (V100:R2, OptiTrack; NaturalPoint). Before each data collection session, the instrument was calibrated. The calibration process allows the definition of a global reference frame for all the cameras used in the optical tracking system. Once the extrinsic and intrinsic parameters for each particular camera were computed, the relative position of these markers inside the working region was known. After the landmarks were captured, the threedimensional coordinates were converted in the local coordinate system created by four head landmarks and subsequently exported. Indirect Method: Marking on Three-Dimensional Reconstructed Computed Tomographic Model (Cone Beam Computed Tomography–Based Three-Dimensional Cephalometry) Computed tomographic images were obtained by means of a DCT Pro cone beam computed tomography device (Vatech, Co., Ltd., Hwasung, Republic of Korea) using the following scanning parameters: 90 kVp; 24 seconds;4 mA; voxel size, 0.4 mm; and field of view, 20 × 19 cm. The image covered the area from the highest
103e
Plastic and Reconstructive Surgery • July 2014 Table 1. Landmark Descriptions Landmark Midline landmarks Sellion Pronasale Labiale superius Labiale inferius Soft-tissue gnathion Paired landmarks: (right and left side noted r and l) Left lateral ciliary Right lateral ciliary Left interciliary Right interciliary Exocanthion (l)
Abbreviation se prn ls li st gn
Lateral ciliaryl Lateral ciliaryr interciliaryl interciliaryr exl
Exocanthion (r)
exr
Endocanthion (l)
enl
Endocanthion (r)
enr
Left upper eyelid Right upper eyelid Left maxillary point
u eyelipl u eyelidr maxl
Right maxillary point
maxr
Soft-tissue zygion (l)
st zyl
Soft-tissue zygion (r)
st zyr
Orbitale superius (l)
osl
Orbitale superius (r)
osr
Soft-tissue orbitale (l)
orl
Soft-tissue orbitale (r)
orr
Alar curvature (l) Alar curvature (r) Left upper lip Right upper lip Cheilion (l) Cheilion (r) Left lower lip Right lower lip Soft-tissue gonion (l)
acl acr u lipl u lipr chl chr l lipl l lipr st gol
Soft-tissue gonion (r)
st gor
Left middle jaw
middle jawl
Right middle jaw
middle jawr
Left cheek point
cheekl
Right cheek point
cheekr
104e
Definition The most posterior point of the frontonasal soft-tissue contour in the midline of the base of the nasal root The most anterior midpoint of the nasal tip The midpoint of the vermilion line of the upper lip The midpoint of the vermilion line of the lower lip The most inferior midpoint of the soft-tissue contour of the chin located at the level of the three-dimensional cephalometric hardtissue menton landmark Located above the most lateral aspect of the left eyebrow Located above the most lateral aspect of the right eyebrow Located above the medial aspect of the left eyebrow Located above the medial aspect of the right eyebrow The soft-tissue point located at the outer commissure of the left eye fissure The soft-tissue point located at the outer commissure of the right eye fissure The soft-tissue point located at the inner commissure of the left eye fissure The soft-tissue point located at the inner commissure of the right eye fissure Superior midportion of the upper left eyelid Superior midportion of the upper right eyelid Located on the cheek one-quarter of the distance between the left alar and left temporomandibular joint Located on the cheek one-quarter of the distance between the right alar and right temporomandibular joint The most lateral point on the soft-tissue contour of the left zygomatic arch, located at the level of the three-dimensional hard-tissue cephalometric zygion landmark The most lateral point on the soft-tissue contour of the right zygomatic arch, located at the level of the three-dimensional hard-tissue cephalometric zygion landmark The most superior soft-tissue point of the lower border of the left brow, located at the most superior level of the left supraorbital rim The most superior soft-tissue point of the lower border of the right brow, located at the most superior level of the right supraorbital rim The soft-tissue point located at the most inferior level of the left infraorbital rim, located at the level of the three-dimensional hardtissue cephalometric orbitale landmark The soft-tissue point located at the most inferior level of the right infraorbital rim, located at the level of the three-dimensional hardtissue cephalometric orbitale landmark The point located at the facial insertion of the left alar base The point located at the facial insertion of the right alar base Left upper lip point halfway between the cheilion and labiale superius Right upper lip point halfway between the cheilion and labiale superius The point located at the left labial commissure The point located at the right labial commissure Left lower lip point halfway between the cheilion and labiale inferius Right lower lip point halfway between the cheilion and labiale inferius The most lateral point on the soft-tissue contour of the left mandibular angle, located at the same level as the three-dimensional hard-tissue cephalometric gonion landmark The most lateral point on the soft-tissue contour of the right mandibular angle, located at the same level as the three-dimensional hardtissue cephalometric gonion landmark Left middle jaw point halfway between the soft-tissue gnathion and the soft-tissue gonion, along the inferior margin of the jaw Right middle jaw point halfway between the soft-tissue gnathion and the soft-tissue gonion, along the inferior margin of the jaw Left cheek point located on the cheek one-quarter of the distance between the left commissures and the left temporomandibular joint Right cheek point located on the cheek one-quarter of the distance between the right commissures and the right temporomandibular joint
Volume 134, Number 1 • Direct versus Indirect Cephalometrics
Fig. 1. Landmarks used for three-dimensional facial soft-tissue analysis. A total of 37 soft-tissue landmarks were applied in the assessment, including five midline landmarks (se, prn, ls, li, and st gn) and 16 paired landmarks.
point on the superior margin of the cranial vault (vertex) to the inferior border of the mandibular body. The gross data and the obtained slices were imported and reconstructed into a three-dimensional model by an interactive image analysis system (Mimics, 14.0; Materialise, Leuven, Belgium). The software automatically determined the origin of the coordinates (x = 0, y = 0, and z = 0) at the left anterior lower corner of the cube contained in the three-dimensional image. As the coordinate system had been standardized and uniquely defined, the spatial positions of each landmark were represented as numerical values (expressed in millimeters) on each axis. Landmarks were designed on the three-dimensional surface model and their positions verified in multiple planar reformat modes.
with an interval of 1 week between the two measurements for each method.
Comparison between the Two Placement Methods Two trained and qualified observers with 7 years’ experience in orthodontics analyzed the reliability and reproducibility of both placement methods. They independently performed the measurements using the placement methods,
where D is the total difference, Δx = xa − x b x + x b x ′a + x ′ b or a − in intraobserver or interob2 2 server error calculation and representing the difference in the x axis, and the same for Δy and Δz.15 Interobserver reliability for landmark placement errors in both direct and indirect measurement
Statistical Analysis Statistical analysis was performed using SPSS Version 16 (SPSS, Inc., Chicago, Ill.). The intraobserver error in each landmark was expressed as a distance between two measurements in three-dimensional coordinates for each placement method. The interobserver error was expressed as a distance between the averages of each observer’s two sets of three-dimensional coordinates of the same point. The equation is as follows: D=
( x ) ^ 2 + ( y ) ^ 2 + ( z ) ^ 2↵ ,
105e
Plastic and Reconstructive Surgery • July 2014
Fig. 2. The direct method, affixing landmarks on the in vivo subject’s face. A set of 37 small reflective markers were affixed on the cutaneous surface of the subject and then captured by six infrared cameras. The three-dimensional coordinates of markers were translated in a local coordinate system created by four head landmarks and subsequently exported. The 16 paired landmarks are illustrated with purple dots, the five midline landmarks are in green, and the four head landmarks are in gray.
methods was assessed by two-way random intraclass correlation coefficients with absolute agreement and their 95 percent confidence intervals. Reproducibility was defined as the mean of the total differences between two sets of measurement for interobserver or intraobserver reproducibility. Paired t tests were used to examine significant differences in the average landmark placement errors between observers and between the two placement methods for each observer. Values of p < 0.05 were considered statistically significant. The average errors were categorized as follows: less than 1 mm, highly precise; 1 to 1.5 mm, precise; 1.5 to 2 mm, moderately precise; and greater than 2 mm, imprecise.16
RESULTS The intraclass correlation coefficients and 95 percent confidence intervals for both landmark placement methods are listed in Table 2. An intraclass correlation coefficient of 0.75 or above is usually considered to be good and that above 0.9 to be excellent.17 For the direct placement method, 32 landmarks (86.5 percent) had an intraclass correlation coefficient higher than 0.75. Among them, 25 landmarks (67.6 percent) had an intraclass correlation coefficient higher than 0.90. For the indirect placement
106e
method, 28 landmarks (75.7 percent) had an intraclass correlation coefficient higher than 0.75, and 16 (43.2 percent) had an intraclass correlation coefficient higher than 0.90. Figure 3 shows the intraobserver and interobserver average landmark placement error distribution for each placement method as color maps. Areas of smaller landmark placement errors were mainly located close to the midline of the face. Landmark placement errors increased with increasing distance from the midline, symmetrically. A significant difference was found between the mean landmark placement errors of the two placement methods (Table 3). The mean landmark placement errors using the direct placement method was 1.093 ± 0.518 mm for observer 1 and 1.101 ± 0.642 mm for observer 2, and the interobserver variability was 1.259 ± 0.566 mm. These findings were all significantly smaller than those obtained by means of the indirect placement method (observer 1, 1.301 ± 0.653 mm; observer 2, 1.340 ± 0.667 mm; interobserver variability, 1.420 ± 0.638 mm). The placement error of several landmarks was less than 1 mm, and only 8 to 17 percent of the landmarks using the direct method had more than 2-mm errors, compared with 16 to 27 percent for the indirect method (Table 4).
Volume 134, Number 1 • Direct versus Indirect Cephalometrics Table 2. Interobserver Reliability of Landmark Placement Errors Using Direct and Indirect Placement Methods Mean Intraclass Correlation Coefficients (95% CI) Landmarks Midline landmark Sellion Pronasale Labiale superius Labiale inferius Soft-tissue gnathion Paired landmarks Left lateral ciliary Right lateral ciliary Left interciliary Right interciliary Exocanthion (l) Exocanthion (r) Endocanthion (l) Endocanthion (r) Left upper eyelid Right upper eyelid Left maxillary point Right maxillary point Soft-tissue zygion (l) Soft-tissue zygion (r) Orbitale superius (l) Orbitale superius (r) Soft-tissue orbitale (l) Soft-tissue orbitale (r) Alar curvature (l) Alar curvature (r) Left upper lip Right upper lip Cheilion (l) Cheilion (r)
Direct Placement Method
Indirect Placement Method
0.971 (0.946–0.986) 0.966 (0.938–0.984) 0.974 (0.943–0.989) 0.96 (0.935–0.985) 0.927 (0.901–0.964)
0.957 (0.916–0.983) 0.943 (0.941–0.992) 0.931 (0.907–0.959) 0.945 (0.922–0.979) 0.88 (0.627–0.947)
0.911 (0.884–0.946) 0.904 (0.757–0.962) 0.9 (0.745–0.945) 0.912 (0.813–0.934) 0.955 (0.912–0.983) 0.96 (0.937–0.986) 0.973 (0.946–0.986) 0.977 (0.953–0.993) 0.944 (0.894–0.967) 0.943 (0.901–0.978) 0.921 (0.887–0.953) 0.913 (0.884–0.943) 0.69 (0.414–0.937) 0.62 (0.397–0.946) 0.833 (0.712–0.934) 0.851 (0.610–0.937) 0.951 (0.918–0.986) 0.942 (0.904–0.978) 0.962 (0.928–0.991) 0.967 (0.945–0.987) 0.952 (0.921–0.983) 0.937 (0.913–0.974) 0.971 (0.948–0.991) 0.976 (0.955–0.993)
0.798 (0.501–0.913) 0.833 (0.528–0.927) 0.84 (0.511–0.907) 0.844 (0.557–0.914) 0.963 (0.921–0.981) 0.933 (0.913–0.954) 0.975 (0.938–0.990) 0.973 (0.933–0.989) 0.917 (0.887–0.953) 0.955 (0.932–0.974) 0.84 (0.563–0.939) 0.872 (0.518–0.908) 0.75 (0.310–0.926) 0.74 (0.486–0.942) 0.86 (0.625–0.921) 0.834 (0.633–0.927) 0.913 (0.885–0.946) 0.935 (0.901–0.958) 0.963 (0.945–0.986) 0.965 (0.933–0.984) 0.882 (0.568–0.931) 0.887 (0.694–0.945) 0.921 (0.886–0.938) 0.918 (0.897–0.952) (Continued )
Table 2. (Continued) Mean Intraclass Correlation Coefficients (95% CI) Landmarks Left lower lip Right lower lip Soft-tissue gonion (l) Soft-tissue gonion (r) Left middle jaw Right middle jaw Left cheek point Right cheek point
Direct Placement Method
Indirect Placement Method
0.948 (0.913–0.987) 0.957 (0.924–0.978) 0.81 (0.624–0.936) 0.85 (0.577–0.943) 0.8 (0.601–0.945) 0.83 (0.572–0.942) 0.8 (0.633–0.931) 0.74 (0.531–0.938)
0.873 (0.678–0.923) 0.867 (0.655–0.942) 0.58 (0.217–0.908) 0.63 (0.396–0.918) 0.58 (0.398–0.894) 0.56 (0.23–0.910) 0.74 (0.382–0.915) 0.72 (0.437–0.874)
l, left; r, right.
DISCUSSION Many researchers have attempted to develop suitable analytic approaches since the development of various techniques for the threedimensional documentation of facial surfaces,15 including laser scanning, computed tomography, and stereophotogrammetry.18 Despite the fact that measurement on the reconstructed model improves the accuracy of anthropometry, many researchers evaluated the linear distances or angles,19 emphasizing the errors caused by placement manipulation and anatomical features. Actually, the systemic error resulting from the imaging acquisition should be a prime consideration before any analytic process. Studies aimed at comparing three-dimensional direct and indirect placement method are still absent. In this study, we investigate the reproducibility and reliability of a series of landmarks, using a three-dimensional direct measurement method, and compared them with computed tomographic measurements. This is the first study aimed at comparing the two threedimensional facial soft-tissue evaluation methods. In both direct and indirect measurement methods, landmarks closer to the midline (i.e., nose bridge, sneer line, and mouth corner) are more reliable than those located on the bilateral face, as indicated by a significantly higher intraclass correlation coefficient (>0.9) and smaller landmark placement error (Fig. 3). Correlation decreases with the distance away from the midline, which reflects the complexity of facial soft tissue. As these midline structures have distinguishable anatomical features (i.e., convex or
107e
Plastic and Reconstructive Surgery • July 2014
Fig. 3. Color maps of landmark placement error distribution in the face mask. (Above) Observer 1, observer 2, and interobserver errors in the direct placement method. (Below) Observer 1, observer 2, and interobserver errors in the indirect placement method.
concave), locating landmarks is easier. Gwilliam et al.18 demonstrated that the surface contours might affect precision and reliability in anthropometry. Good identification in irregular structures by means of a comprehensive observation from different angles could be obtained by deft control of the three-dimensional reconstruction through rotation, translation, and magnification of the images in computed tomography,10 or by sensitive visual observation under natural light in the direct method. It has been suggested that the midline-related landmarks are accurate when establishing their position, whether using direct or indirect measurement methods. Referring to the paired landmarks, it was found that landmarks surrounding the eyelashes Table 3. Comparison of Landmark Placement Mean Errors Using Direct and Indirect Placement Methods Landmark Placement Method Direct Indirect p
108e
Landmark Placement Mean Errors (mean ± SD) Observer 1
Observer 2
Interobserver Variability
1.093 ± 0.518 1.301 ± 0.653 0.005
1.101 ± 0.642 1.340 ± 0.667 0.009
1.259 ± 0.566 1.420 ± 0.638 0.050
(i.e., left lateral ciliary, right lateral ciliary, left interciliary, and right interciliary) and lips (i.e., left upper lip, right upper lip, left lower lip, and right lower lip) showed excellent correlations (>0.90) using the direct method, compared with those found for the computed tomography method (0.75 to 0.90). The larger error and lower precision in computed tomographic marking can be explained by the difficulty in identifying the upper eyelid border or vermilion border in the lack of a colored texture in the three-dimensional reconstructed facial model. Because of the welldefined contours, the direct placement method can identify the eyebrow and lips border by inspection, thereby achieving accurate placement. In addition, the limited accuracy of computed tomography in revealing soft tissues contributes to a larger error. Even with multiangle control of images, the observation of a three-dimensional model from a two-dimensional computer screen cannot reflect reality, unlike direct observation. The registration of the natural texture of the face by advanced three-dimensional computed tomographic reconstruction could be a helpful solution to this problem. Computed tomography, the common imaging technique in orthodontic
Volume 134, Number 1 • Direct versus Indirect Cephalometrics Table 4. Number and Percentage of Landmarks in Relation to the Average Intraobserver and Interobserver Differences, Divided into Four Groups, of 37 Landmarks for Direct and Indirect Placement Methods Reproducibility Level Landmark Placement Method Direct No. % Indirect No. %
Highly Precise (≤1 mm)
Moderately Precise (1.5–2 mm)
Precise (1–1.5 mm)
Imprecise (>2mm)
Ob1
Ob2
Ob1–Ob2
Ob1
Ob2
Ob1–Ob2
Ob1
Ob2
Ob1–Ob2
Ob1
Ob2
Ob1–Ob2
17 45.95
18 48.65
13 35.14
11 29.73
8 21.62
13 35.14
6 16.22
6 16.22
5 13.51
3 8.11
5 13.51
6 16.22
16 43.24
15 40.54
14 37.84
11 29.73
8 21.62
6 16.22
3 8.11
6 16.22
6 16.22
6 16.22
7 18.92
10 27.03
Ob1, observer 1; Ob2, observer 2.
examination, has advantages in overlapping skin and the subcutaneous bone tissues, improving the identification of facial soft-tissue landmark locations. Several errors exist because of a certain distance between the surface of soft and hard tissues. Structural features do not completely match (i.e., the most convex point of soft tissue sometimes does not correspond to the most convex point of underlying hard tissue). Similarly, the location of landmarks representing the soft-tissue gonion and middle jaw is difficult using the indirect method. The intraclass correlation coefficient has been found to be lower than 0.75. The lack of color contrast and the potential for concealment of this area by soft tissue are possible reasons for its poor reliability. In contrast, even the smallest deflection of observation angle could bring false identification of these markers because of their locations at the edge of soft tissue. Reliability is a crucial factor in evaluating a placement method, but reproducibility, assessed by landmark placement error, is also an essential element. The majority of landmarks (>35 percent) in both methods are reproducible to less than 1 mm. The average landmark placement error in the direct methods is lower than in the indirect ones, with values of p = 0.005, p = 0.009, and p = 0.050 for within and between observers, respectively. A higher accuracy in landmark placement is indicated with the direct placement method. Although performing a direct measurement is time-consuming because of affixing landmarks manually on the face, once the landmarks are all set, the capture speed is so fast that even erratic movements by the youngest child or the most restless patient can be ignored. Therefore, the measurements could be more readily obtained. Furthermore, the direct placement method is noninvasive, without exposure to radiation. Images can be taken as frequently as needed, providing a safe strategy for repeated assessment
and testing. The capture of the landmarks pasted directly onto the face is of significance when studying facial kinematics, such as expressions and smile assessment. This is relevant to orthodontic treatment assessment.
CONCLUSIONS Three-dimensional dynamic evaluation of facial aesthetics based on the motion capture technique has the potential to transform the conventional static assessment in plastic surgery and related disciplines. This first study compared the direct and indirect three-dimensional facial evaluation methods and classified the 37 soft-tissue landmarks into four precise grades according to their landmark placement error values. Our findings mainly suggest that direct observation is more accurate in the perception of welldefined contours and natural textures of the face than the indirect method. The direct placement method had significantly lower errors than the indirect method for repeatedly placing markers. Meanwhile, similar reliability in evaluating midline-related structures was found with the two approaches. With the identification of the landmark placement error category of landmarks, we laid a sound basis for optimal selection of landmarks, which results in a convincing facial analytic system. Yue Xu, M.D., Ph.D. Guanghua School of Stomatology Sun Yat-sen University Lingyuan Xilu No.56 Guangzhou 510055, People’s Republic of China [email protected]
acknowledgments
This work was supported by the Projects of Integration of Industry, Education and Research of Guangdong Province (2012B091100453); Sun Yat-sen
109e
Plastic and Reconstructive Surgery • July 2014 University Clinical Research 5010 Program (2007050); and Key Laboratory of Sensing Technology and Biomedical Instruments of Guangdong Province (2011A 060901013). references 1. Goldstein RE. Study of need for esthetics in dentistry. J Prosthet Dent. 1969;216:589–598. 2. Paiva JB, Attizzani MF, Miasiro Júnior H, Rino Neto J. Facial harmony in orthodontic diagnosis and planning. Braz Oral Res. 2010;241:52–57. 3. Spyropoulos MN, Halazonetis DJ. Significance of the soft tissue profile on facial esthetics. Am J Orthod Dentofacial Orthop. 2001;1195:464–471. 4. Germec-Cakan D, Canter HI, Nur B, Arun T. Comparison of facial soft tissue measurements on three-dimensional images and models obtained with different methods. J Craniofac Surg. 2010;21:1393–1399. 5. de Menezes M, Rosati R, Allievi C, Sforza C. A photographic system for the three-dimensional study of facial morphology. Angle Orthod. 2009;79:1070–1077. 6. Swennen GRJ, Schutyser F, Hausamen JE. Three-Dimensional Cephalometry: A Color Atlas and Manual. Berlin: Springer; 2006. 7. Park SH, Yu HS, Kim KD, Lee KJ, Baik HS. A proposal for a new analysis of craniofacial morphology by 3-dimensional computed tomography. Am J Orthod Dentofacial Orthop. 2006;129:600.e23–600.e34. 8. Cevidanes LH, Bailey LJ, Tucker GR Jr, et al. Superimposition of 3D cone-beam CT models of orthognathic surgery patients. Dentomaxillofac Radiol. 2005;34:369–375. 9. Grauer D, Cevidanes LS, Styner MA, et al. Accuracy and landmark error calculation using cone-beam computed tomography-generated cephalograms. Angle Orthod. 2010;802: 286–294.
110e
10. Plooij JM, Swennen GR, Rangel FA, et al. Evaluation of reproducibility and reliability of 3D soft tissue analysis using 3D stereophotogrammetry. Int J Oral Maxillofac Surg. 2009;38:267–273. 11. Lou L, Lagravere MO, Compton S, Major PW, Flores-Mir C. Accuracy of measurements and reliability of landmark identification with computed tomography (CT) techniques in the maxillofacial area: A systematic review. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2007;104:402–411. 12. Muramatsu A, Nawa H, Kimura M, et al. Reproducibility of maxillofacial anatomic landmarks on 3-dimensional computed tomographic images determined with the 95% confidence ellipse method. Angle Orthod. 2008;78:396–402. 13. Proffit WR. Malocclusion and dentofacial deformity in contemporary society. In: Proffit WR, Fields HW Jr, Sarver DM, eds. Contemporary Orthodontics. 5th ed. St. Louis, Mo: Elsevier/Mosby; 2013:2–18. 14. Weeden JC, Trotman CA, Faraway JJ. Three dimensional analysis of facial movement in normal adults: Influence of sex and facial shape. Angle Orthod. 2001;71:132–140. 15. Toma AM, Zhurov A, Playle R, Ong E, Richmond S. Reproducibility of facial soft tissue landmarks on 3D laserscanned facial images. Orthod Craniofac Res. 2009;12:33–42. 16. Mutsvangwa TE, Veeraragoo M, Douglas TS. Precision assessment of stereo-photogrammetrically derived facial landmarks in infants. Ann Anat. 2011;193:100–105. 17. Sayinsu K, Isik F, Trakyali G, Arun T. An evaluation of the errors in cephalometric measurements on scanned cephalometric images and conventional tracings. Eur J Orthod. 2007;29:105–108. 18. Gwilliam JR, Cunningham SJ, Hutton T. Reproducibility of soft tissue landmarks on three-dimensional facial scans. Eur J Orthod. 2006;28:408–415. 19. Zamora N, Llamas JM, Cibrián R, Gandia JL, Paredes V. A study on the reproducibility of cephalometric landmarks when undertaking a three-dimensional (3D) cephalometric analysis. Med Oral Patol Oral Cir Bucal. 2012;17:e678–e688.
