INTRODUCTION

The extent to which an individual's face is recognized by others exerts a great sociopsychological influence on that individual's sense of acceptance by his or her community. The lips are located in the lower third division of the face, and they play a key role in sustaining the quality of our life (ie, not only in terms of sustenance of biological life, but also in relation to proper pride as social beings).

When an orthodontic treatment plan is developed, evaluation of lip vermilion profile forms is indispensable in the decision of whether orthodontic movement of the incisors alone (camouflage) or a combined surgical orthodontic approach will more precisely create an improved posttreatment facial appearance. On the basis of these clinical demands, an objective and quantitative evaluation of morphologic traits of the lip vermilion in the lateral view would be of great value for making orthodontic diagnoses and for optimizing predictions of possible treatment outcomes. To date, several reports1–6 have documented quantitative evaluations of the lip vermilion performed by means of linear and/or angular measurements with cephalograms (eg, distances between incisors and soft tissues, degrees of protrusion or retrusion). The results of these reports are meaningful because the use of cephalograms allows simultaneous quantitative evaluation of hard and soft tissues; it is difficult, however, to identify lip commissures and lip vermilion borders from the cephalograms. A second problem is that the shape of the lip vermilion on cephalograms differed from that observed during visual inspection of the face (ie, the actual shape) because of a radial-tangential effect.7 To resolve these issues, photographs were used in several reports to analyze the lip vermilion shape.8–10 Because these analyses were conducted primarily with the use of linear and/or angular measurements, detailed morphologic characteristics of the entire lip vermilion (eg, pouty lip vermilion, flat lip vermilion, flaps of the lip vermilion) were not considered. Although Fourier series computations were found to be useful in obtaining and analyzing approximate entire human profiles,11 they have not yet been applied to evaluation of the entire lip vermilion shape with a fine enough resolution to detect morphologic traits that, although they may be subtle, are significant in terms of an orthodontic diagnosis. Recently, a technique for modeling the feature extraction process that takes into account the knowledge and thought processes employed by experts in linguistically describing the morphologic traits of nose profiles has been established and confirmed to be of great value in orthodontic diagnoses.12

The purposes of the present study were (1) to determine the numbers and shapes of human lip vermilion profiles that are mathematically and thus objectively classifiable, and (2) to examine whether any unique morphologic characteristics of dentoskeletal patterns are specific to each classified soft tissue pattern.

MATERIALS AND METHODS

Two hundred thirty-four Japanese females who had never before received any orthodontic treatment were studied (mean age, 26 years 2 months; age range, 18 years 2 months to 59 years 1 month). These participants were selected consecutively from the patient database in order of their dates of registration at the hospital. The subjects had permanent dentition; none of them had any congenital anomalies, history of surgery, or trauma or injury to the face.

Conventional lateral facial photographs (right-side views) were employed. They were taken with the head fixed by ear rods and the Frankfort horizontal plane parallel to the ground, and with the upper and lower teeth in light intercuspation and the lips in repose. To examine the craniofacial morphologic characteristics of subjects, standard digitally recorded lateral cephalograms of each patient were used. Dentoskeletal traits of the subjects are summarized in Table 1. Proportions of subjects in terms of conventional sagittal skeletal classification13 were as follows: skeletal Class I, 45.0%; skeletal Class II, 30.0%; and skeletal Class III, 25.0%.

Table 1.  Summary of Subjects' Dentoskeletal Morphologic Traits^a

View Fullscreen

The methods used for collection and analysis of data in the current study, including the definition of the coordinate system used in measuring the lateral face and the mathematical extraction of morphologic traits on the basis of human knowledge, have been described in detail elsewhere.12 In this article, only a brief summary of the methods is provided.

An orthodontist visually located the positions of seven facial landmarks912 (Table 2)—porion (po), exocanthion (ex), subnasale (sn), labiale superius (ls), labiale inferius (li), cheilion (ch), and stomion (sto)—on each photograph. The methods12 used in segmenting the facial profile and in defining the coordinate system employed in the present study are schematically illustrated in Figure 1.

Table 2.  Definitions of the Landmarks Employed in the Present Study

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Twenty-eight descriptive parameters, expressed linguistically by t (t = 1, 2, …, 28) and used in classifying the configuration of lip vermilions in the lateral view, were collected from a panel of three orthodontists, who provided their judgments based on experience and knowledge (Table 3).

Table 3.  Twenty-eight Kinds of Knowledge Described in Linguistic Forms That Were Used for Classifying Shapes of Lip Vermilions

View Fullscreen

On the basis of linguistic descriptions of the configurations of lip vermilions, 18 mathematical descriptions (ie, vector elements v1, v2, …, v18 and subsets of vector elements) were defined (Table 4, Figure 2).

Table 4.  Definitions and Interpretations of 18 Vector Elements

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Sixteen feature vector sets V^(w) (w = 1, 2, …, 16) that feature the configuration of the lip vermilion thus were determined with the use of vector elements (v1, v2, …, v18) or their combinations, with the conditions previously reported.12 Figure 3 shows the feature vector set that met these conditions. Shapes of the lip vermilion in the lateral view were classified, and the optimum vector set was identified.

View Figure

• Download as Powerpoint
• Download Figure

A detailed description of methods used to calculate the optimum number of configurations of lip vermilion profile patterns (codes) and to identify an optimum feature vector set has been given.12 In short, the vector quantization (VQ) method14 was applied to feature vector sets to obtain the number of lip vermilion profile patterns that were mathematically optimized. Then, so that an optimum feature vector set could be identified out of V^(w) (w = 1, 2, …, 16), the matching score12 for each feature vector set was calculated. The w that achieved the highest matching score was defined as the optimum feature vector extraction method.

Mean lip vermilion profile shapes reconstructed by averaging the approximated curves that connected ls and li and the linear curves that connected ch and ls, ch and sto, and ch and li, respectively for each code vector, were computed. Classified lip vermilion profile shapes were compared, as were the corresponding mean dentoskeletal patterns.

In examining the record sets that belonged to each extracted element of the feature vector, one-way analysis of variance (ANOVA) was performed to determine whether a significant difference could be seen between each element's code and each cephalometric variable as measured. In addition, the Tukey-Kramer test was performed to identify the codes that showed significant differences between the vector elements and between the cephalometric variables. A P < .01 level was assigned as significant.

RESULTS

The highest matching score was achieved when w = 1. In other words, the optimum feature vector was found to be a 13-dimensional vector whose elements were v1, v2, v3, v4, v5, v6, v7, v8, v9, v12, v13, v15, and v16. The shapes of lip vermilions in the current sample were found to be most effectively distinguished by the 13-dimensional feature vector thus determined in accordance with linguistic descriptions (t = 1, 2, …, 28) that were assumed to reflect the knowledge and judgment shared by oral health practitioners who were experts on lateral lip shapes.

Seven mathematically optimized codes were identified. This means that the current samples typically exhibit seven types of lip vermilion profile shape. Proportions of the numbers of subjects classified into each code were, from Code 1 to Code 7, 23.5%, 18.4%, 15.8%, 15.4%, 14.5%, 9.8%, and 2.6%, respectively. Mean shapes of the lip vermilion profile patterns that corresponded to each of the seven codes are shown in Figure 4. Codes that were expressed in the real space are given in Table 5. Figure 5 reveals comparisons between codes for each vector element. A detailed description of differences in the shape of lip vermilions between codes is given in the Appendix.

View Figure

• Download as Powerpoint
• Download Figure

Table 5.  Means and Their Standard Deviations for Vector Elements Expressed in the Real Space Determined for Each Code^a,b

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Figure 6 shows comparisons of sample means and their standard deviations between the codes calculated for each cephalometric variable. Codes that were expressed in the real space are given in Table 6.

View Figure

• Download as Powerpoint
• Download Figure

Table 6.  Means and Their Standard Deviations Computed for Cephalometric Variables Determined for Each Code^a,b,c

View Fullscreen

As for the length SN, Code 4 showed a significantly greater mean value when compared with Code 1 (P < .01). Regarding the maxilla, significant differences were not found between any codes for the angle SNA and the length A-Ptm/PP (P < .01). With regard to the mandible, Code 1, Code 3, Code 4, and Code 7 had significantly greater mean values than Code 5 for the length Ar-Me (P < .01). Code 5 had a significantly greater mean value than Code 6 for angle SNMP (P < .01). Code 1 had a significantly greater mean value than Code 6 for the length Me/PP (P < .01). As for the angle ANB, Code 1, Code 3, Code 4, and Code 7 showed significantly smaller mean values than were seen in Code 5 (P < .01).

With regard to upper incisor inclination, Code 1 exhibited a significantly greater mean value than Code 5 for the angle U1-SN (P < .01). For the lower incisors, Code 2 and Code 5 represented significantly smaller mean values than Code 1 for the angle L1-FH (P < .01). Code 2, Code 5, and Code 6 showed significantly increased means compared with Code 1, Code 3, and Code 7 (P < .01). As for the overbite, Code 1 showed a significantly reduced mean value when compared with Code 2, Code 4, Code 5, and Code 6 (P < .01).

DISCUSSION

Relationships between dentoskeletal morphology and lip vermilion profile configuration have been documented most often on the basis of subjective observation.8 This is the first report that classified the human lip vermilion profile configuration objectively using the vector quantization method and determined its relationship to the underlying dentoskeletal morphology.

Code 2, Code 5, and Code 6 were characterized by forward posturing of the upper lip vermilion relative to the lower lip (see v13) and were associated with an increased overjet and a retruded mandible relative to the anterior cranial base. In other words, the results suggested that individuals with an increased overjet and the skeletal Class II tendency are likely to show more protrusive upper lip vermilions than lower lip vermilions. Code 2 was characterized by upper/lower lip vermilions whose anterior portions are large (often referred to as “protruded lips”; see v6, v8, v16). Code 6 represented the vertically long (often called a “thick or full lip”; see v4) upper lip vermilion whose anterior portion is large (see v6, v15) and the vertically shorter (“thin”; see v5) and flatter lower lip vermilion (see v16). Lip vermilion profile configurations of Code 2 and Code 6 were similar, but Code 6 exhibited a thicker and more protruded mean upper lip vermilion shape and a thinner lower lip vermilion shape, in contrast to Code 2. Such differences may be explained by the dentoskeletal morphologic traits as determined for Code 6 (ie, the tendency of lower incisors toward retroclination). It can be inferred that the tipped-in lower incisors as found in Code 6 subjects may likely lead to a backward positioning of the lower lip vermilion, forming a thinner appearance, whereas its upper counterpart may be looked upon as thicker (or vertically longer) because the upper lip vermilion receives, in theory, less vertical pressure from the lower lip vermilion. Code 5 was characterized by a lowered positioning of the most protruded point of the lower lip vermilion (see v9) and a straightforwardly oriented lip fissure (see v12). Mean dentoskeletal patterns characteristic of Code 5 included high mandibular angle tendency, palatally tipped upper incisors, and labially tipped lower incisors. It can be speculated that the lip vermilion in Code 5 is strained because the lip fissure tends to go upward in its posterior part when the subject must strain to bring the lips together. Code 5 is supposed to have inherent lip shortness in both upper and lower lips because of the clockwise-rotated mandible, which leads to incongruity of lip structure with the skeletal pattern.8

Code 1, Code 3, and Code 7 exhibited a mean lip configuration that was characterized by a more forward position of the lower lip vermilion relative to the upper lip (see v13). Dentoskeletally, these are characterized by a decreased overjet and a skeletal Class III tendency accompanied by a longer mandibular effective length. It thus can be inferred that individuals who have a decreased incisor overjet and/or skeletal Class III morphologic traits are likely to show the more protruded lower lip vermilion when compared with the upper lip vermilion.

Both Code 3 and Code 7 had vertically shorter upper lip vermilions (“thin lip”; see v4) and longer lower lip vermilions (“chubby lip”; see v5). Two possible reasons can be put forth for such features. First, in individuals with skeletal mandibular prognathism, the lower lip vermilion, positioned more forward than its opponent, is likely to receive less vertically directed pressure from its upper counterpart (but rather greater pressure from the upper lip vermilion toward the forward direction), which results in the vertically thicker and horizontally protruded lower lip vermilion. In contrast, the upper lip vermilion acquires the vertically short (“thin”) and flat appearance. This may occur because the lower lip vermilion horizontally pushes the upper lip vermilion backward, and a part of the upper lip vermilion goes behind the lower lip vermilion. A second possibility is that Code 3 and Code 7 showed shortness in the upper lip, which leads to a thinner upper lip and thus a thicker lower lip as the result of less pressure from the upper lip vermilion. Code 1 does not exhibit the vertically shorter (i.e., thin) upper lip vermilion; this feature is unique to Code 1 among the lip configurations whose underlying hard tissue structures showed skeletal Class III tendencies. This may be so because the severity of the mandibular prognathism in Code 1 was weaker than that in Code 3 or Code 7. In addition, subjects who constituted Code 1 showed labially inclined upper incisors. It can be speculated that the upper lip vermilion posture in Code 1 is dependent more on the underlying upper incisors than is the posture of those of other codes that showed skeletal Class III tendency; thus it pushes its lower counterpart vertically, which leads to the vertically even proportion of the upper and lower lip vermilions.

Code 4 was characterized by anteroposteriorly long lip vermilions (see v1, v2, v3), a flat lower lip vermilion (see v8, v9, v16), and the straightforwardly oriented fissure (see v12). As for the dentoskeletal features, the subjects who belong to Code 4 exhibited skeletal Class I relationships, a tendency toward the long mandibular effective length, and a long anterior cranial base, which translates to a horizontally large face. It can be speculated that, in a similar way as is found in Code 5, horizontal lip shortness in the upper and lower lips makes lips strained because of the incongruity of the lip structure with the horizontally large face; thus the lip fissure tends to go upward in its posterior part in strain, and the lip vermilion tends to be anteroposteriorly longer when the subject brings the lips together and purses the lips.8

In summary, lip vermilion profile configurations were found to be associated with horizontal lengths of the anterior cranial base, horizontal/vertical positions, inclination and length of the mandible, and horizontal positions and labio-lingual inclinations of the upper and lower incisors.

CONCLUSIONS

• The optimum knowledge description for distinguishing lip vermilion profile configurations was provided by a 13-dimensional feature vector representation.

• Classifying lip vermilion profile shapes collected from 234 women into seven representative patterns was optimum in maximizing differences in configuration of the lip vermilion.

• Proportions for each code (pattern) were 23.5%, 18.4%, 15.8%, 15.4%, 14.5%, 9.8%, and 2.6% from Code 1 to Code 7, respectively.

• Lip vermilion profile configurations were found to be associated with horizontal lengths of the anterior cranial base, horizontal/vertical positions, inclination and length of the mandible, and horizontal positions and labio-lingual inclinations of the upper and lower incisors.