Three-dimensional dynamic evaluation of facial soft tissue changes following anterior traction in growing Angle Class III malocclusion patients

Abstract

Background

The aesthetic improvement of the midface is a critical goal in treating growing children with Angle Class III malocclusion. While the skeletal effects of maxillary anterior traction are well-established, predicting the final aesthetic outcome requires a precise understanding of the accompanying three-dimensional (3D) soft tissue response. Current literature lacks detailed, dynamic quantification of these adaptive changes using high-precision methodology. This study aims to address this gap by employing dynamic stereophotogrammetry to sequentially analyze the 3D soft tissue kinematics associated with maxillary advancement.

Methods

In this study, the three-dimensional dynamic facial imaging system (3dMD) was employed to evaluate the effects of maxillary anterior traction treatment in 28 adolescents and children with Angle Class III malocclusion. Quantitative changes in facial root mean square (RMS) values and landmark displacements were compared across different treatment stages. Key measurements included RMS values at the initial (Ta), middle (Tb), and final (Tc) frames of four facial expression sequences (smile lips closed, smile lips open, lip purse, and cheek puff) along with the corresponding changes in anatomical landmark positions.

Results

Analysis revealed that the RMS values for each sequence frame of facial dynamic expressions increased from baseline (pre-treatment) levels at both the 3-month follow-up (T1-T0) and the end of treatment (T2-T0). RMS values exhibited differential evolution across distinct phases of each facial expression. For the smile lips closed expression, the overall change did not reach statistical significance (p = 0.064). In contrast, the smile lips open expression exhibited statistically significant increases over the full treatment course (p = 0.015). More pronounced changes were observed in the lip purse and cheek puff expressions, both of which exhibited highly significant increases in RMS amplitude across the entire treatment period (p = 0.002 and p < 0.001, respectively). With regards to landmark displacements, statistically significant anterior movements were observed for landmarks b, c, d, e, f and h, while landmark a exhibited a significant posterior displacement. In contrast, no significant positional changes were detected for landmarks i and j.

Conclusion

Anterior traction effectively improved the class III facial profile while dynamic 3D assessment revealed a non-linear adaptation pattern with enhanced midfacial convexity. These objective data are crucial for predicting outcomes in early-phase patient management.

Introduction

Skeletal class III malocclusion is one of the common complex malocclusion malformations in the orthodontic clinic, with a high incidence in Asian populations. These malformations not only affect a patient’s facial aesthetics and oral function but may also cause psychological disorders [1]. Consequently, early intervention is particularly important for patients with skeletal class III malocclusion. Anterior traction treatment remains an effective method with which to correct under-development of the maxilla in children during growth and development [2]. This method stimulates the reconstruction of bone sutures around the maxilla by applying orthopedic force and promotes the forward development of the maxilla [3].

Traditionally, evaluation of the efficacy of anterior traction treatment has been predominantly based on two-dimensional lateral cranial films and static three-dimensional images [4]. Although cone beam computed tomography (CT) provides information relating to bone changes from a three-dimensional perspective, these methods are still limited to static assessment [5, 6]. Facial form is the most complex and socially recognizable structure in the human body. Movement changes among various structures of the face are highly coordinated and refined, providing the face with significant diversity in beauty and harmony [7]. However, facial expression is a dynamic process, and traditional static analysis cannot fully reflect the impact of treatment on the dynamics of facial function [8]. The three-dimensional dynamic facial imaging system (3dMD) can capture the sequence of natural facial expression changes, thus providing a new perspective for evaluating treatment effects [9]. 3dMD system, developed in the early 2000s, is a validated high-precision stereophotogrammetric tool capable of capturing dynamic facial expression sequences in milliseconds, thereby eliminating motion artifact. It has been extensively used in craniofacial surgery and syndrome phenotyping to quantify static morphology and, increasingly, dynamic function. While previous orthodontic research has utilized its static imaging capability [10], this study leverages its dynamic capture to perform a novel kinematic analysis of the soft tissue response to maxillary anterior traction—a methodology that addresses the limitation of static snapshots and provides unprecedented insight into the timing and pattern of soft tissue adaptation.

Over recent years, finite element analysis has further revealed the biomechanical mechanisms underlying forward traction. Previous studies have shown that several factors can influence treatment effect, including the direction of traction (at an angle of 30° to 40° from the plane), the value of force (300–500 g on each side) and the method of anchorage (implant or traditional dental anchorage) [11, 12]. However, most of these previous studies focus on skeletal and dental changes and lack a quantitative assessment of dynamic changes in facial soft tissue [13]. The dynamic analysis of facial soft tissue can accurately evaluate and quantify changes in movement of the facial soft tissue, thus providing important reference and guiding significance for the establishment of aesthetic evaluation standards for oral clinical practice and the accurate assessment of soft tissue changes before and after treatment [14].

In this study, we utilized a 3dMD dynamic imaging system to quantify the characteristics of dynamic displacement of facial soft tissue landmarks and the improvement of facial soft tissue dynamic aesthetics in pediatric patients with skeletal class III malocclusion after anterior traction treatment through fixed-point measurement and quantitative analysis and further evaluated the correlation between these dynamic changes and correction effect.

Methods

Subjects

The study sample comprised a cohort of 28 patients (Table 1) who were referred to the Department of Orthodontics, School of Stomatology, the Fourth Military Medical University, Xi’an, Shaanxi, China. The inclusion criteria were as follows: 5–12 years of age; Angle Class III malocclusion (ANB < 0°); no previous history of motor and sensory neurological disorders in the face; no previous history of facial surgery; and no history of orthodontic treatment. The exclusion criteria were as follows: a previous history of orthodontic or orthopedic treatment; facial asymmetry or deformity; a previous history of maxillofacial trauma and surgery; temporomandibular joint disorders with obvious clinical symptoms, systemic disease that could influence the dentition, poor cooperation and/or inadequate oral hygiene.

Table 1.

Sample demographics

This study followed the guidelines of the Declaration of Helsinki (2013). Ethical approval was obtained from the Fourth Military Medical University, Xi’an, Shaanxi Province, China (Reference: KQ-YJ-2025-179) and all participants provided signed and informed approval to participate.

3D dynamic motion-capture imaging

Dynamic facial expressions were captured using a 3dMDface Dynamic System (3Q Technologies, Atlanta, GA, USA). This system utilizes a multi-camera setup to acquire high-resolution and synchronized 3D surface data at a rate of 60 frames per second. Data acquisition was performed at three time points: T0 (pre-treatment), T1 (anterior traction treatment for three months), and T2 (at the end of treatment). During each recording session, the patient was seated in a natural head position and instructed by a single investigator to perform a series of standardized facial animations [15]. The sequence included: smile lips closed, smile lips open, lip purse, and cheek puff (Fig. 1). Each expression was performed naturally and held steady for 2–3 s to ensure adequate data capture. The entire dynamic sequence for each subject was captured in a single and continuous take lasting approximately 15 s.

Fig. 1.

Facial expressions analyzed in this study. A Smile lips closed; B smile lips open; C lip purse; D cheek puff

Image analysis

Three-dimensional facial expression image sequence analysis was performed using Geomagic Wrap2021 (3D Systems, USA) software. Three key frames were selected during each expression process. The rest state was defined as the initial frame (Ta). The maximum motion state was defined as an intermediate frame (Tb), and the end of motion was defined as an end frame (Tc) (Fig. 2). Relatively stable reference points were selected on the forehead and the nasal root area of each patient. The pre-treatment initial frame (Ta) was overlapped with the initial frames taken after three months of treatment (Ta’) and upon treatment completion (Ta”) to generate a composite image [16, 17]. The specific sites used to acquire 3dMD facial image measurements are given in Fig. 3 and Table 2.

Fig. 2.

Sequence frame for smile lips open (A); the rest state was the initial frame Ta; B The maximum motion state was an intermediate frame Tb; C the end of motion was the end frame Tc

Fig. 3.

Facial soft tissue landmarks assessed in the study. a Glabella; b Pronasale; c Alare right; d Alareleft; e Subnasale; f Labiale superius; g Cheilion right; h Cheilion left; i Labiale inferius; j Pogonion

Table 2.

List of all facial landmarks evaluated in this study

Analysis was performed by color-coded mapping (Fig. 4), in which green corresponded to soft tissue changes below the tolerance threshold, blue corresponded to flattening or concave deformation (negative displacement), and red corresponded to a more convex deformation (positive displacement). The root mean square (RMS) value represented the magnitude of change across stages, with T1-T0 representing the overall facial change after three months of treatment and T2-T0 representing the overall facial change at the end of treatment. The 10 landmarks were used to describe point-specific changes across the corresponding treatment stage sequence (Ta).

Fig. 4.

Generation of absolute color mapping and the measurement of landmarks

Bias

To evaluate the repeatability of this method for image selection and superimposition, encompassing both intra-operator and inter-operator reliability, 80 3dMD images from 10 subjects were randomly selected from a pool of 360 images. First, to assess intra-operator reliability, Operator A performed the image selection and superimposition procedure independently at two time points (baseline and two weeks apart). Second, to assess inter-operator reliability, Operator B independently performed the procedure once, and the results were compared with the first measurement from Operator A. Intraclass and interclass correlation coefficients (ICC), technical error of measurement (TEM), and relative TEM (rTEM) were calculated separately for each reliability dimension. Analysis demonstrated excellent reliability for both measures: intra-operator reliability (ICC = 0.965, p < 0.001; TEM = 0.03 mm; rTEM = 4.85%) and inter-operator reliability (ICC = 0.971, p < 0.001; TEM = 0.04 mm; rTEM = 3.76%).

Study size

A clinically significant difference in landmark position between males and females was set at 2 mm, as described previously [18]. The expected variability of the differences was ± 1.9 mm [19.]. Using a significance level of 0.05 and a power of 80%, a sample size of 21 subjects was required. In our final cohort, we included a total of 28 patients to account for attrition.

Statistical analysis

SPSS version 23.0 (IBM, Armonk, NY, USA) was used for statistical analysis. Quantitative data were first assessed for normality using the Shapiro-Wilk test and Q-Q plots. Normally distributed data are presented as mean ± standard deviation, while non-normally distributed data are presented as median (interquartile range). The significance level was set at a two-tailed α of 0.05.

To evaluate both the intra-operator and inter-operator reliability of the image selection and superimposition procedure, intraclass and interclass correlation coefficients (ICC) were calculated using a two-way random-effects model for absolute agreement. To assess the temporal stability of the registration algorithm, the same operator performed three independent registrations on an identical dataset (T1-T0) using Geomagic Wrap software at three distinct time points (baseline, one week, and two weeks). The consistency of the overall facial change values, defined as the root mean square (RMS), across these time points was tested using one-way repeated-measures analysis of variance (ANOVA). The assumption of sphericity was verified using Mauchly’s test (p > 0.05). ANOVA revealed that there was no statistically significant main effect of time point (p > 0.05), with a very small effect size (partialη² < 0.01).

To quantify overall motion accuracy and analyze stage-wise differences, the mean RMS value across three sequential frames (initial, middle, and end) of each facial expression was calculated as a composite score. This yielded the early-stage change (ΔT1) and the full-course change (ΔT2). The paired difference (D = ΔT2-ΔT1) was then computed for each expression. The normality of the difference scores (D) for each expression was confirmed using the Shapiro-Wilk test (all p > 0.05). Consequently, paired-sample t-tests were used to compareΔT1 andΔT2 for each expression. Corresponding effect sizes were calculated using Cohen’s d.

Results

Each of the 28 subjects was evaluated using 10 landmarks, with each subject being evaluated on three occasions. A total of 936 3dMD dynamic sequence frames were acquired. There were no statistically significant differences in inter-examiner variability or intra-examiner variability. Neither tester encountered any difficulty setting up landmarks.

Table 3 presents the RMS values of dynamic facial expression movements at different follow-up stages relative to baseline (T0). Analysis showed that at the 3-month follow-up (T1-T0), the RMS values for all measured facial expressions were approximately 1 mm, with standard deviations reflecting inter‑subject variability. By the end of treatment (T2-T0), the RMS values for all expressions and measurement sequences (Ta, Tb, Tc) had increased, ranging from approximately 1.2 mm to 1.5 mm, indicating a progressive and widespread enlargement in the magnitude of soft-tissue dynamic movement during therapy. Table 4; Fig. 5 compare the RMS changes induced by treatment between the early phase (ΔT1) and the full-course phase (ΔT2). Statistical analysis revealed a consistent and significant increasing trend in RMS values from ΔT1 to ΔT2 across all four facial expressions. Specifically, the increases in RMS for smile lips open, lip purse, and cheek puff were statistically significant (p = 0.015, 0.002, and < 0.001, respectively), whereas the change for smile lips closed did not reach statistical significance (p = 0.064). According to Cohen’s d effect size estimates, the magnitude of the increase ranged from moderate (smile lips closed, d = 0.5) to very large (cheek puff, d = 1.6). Of these, the RMS change for cheek puff was the most pronounced, increasing from 1.025 mm at ΔT1 to 1.315 mm at ΔT2. These findings suggest that the treatment effect did not plateau after the initial phase but continued to develop dynamically throughout the entire treatment course. The sex‑based subgroup analysis of treatment-induced RMS improvement (T2-T0) is presented in Table 5. These results indicate that the magnitude of treatment-related improvement in dynamic facial movements did not differ significantly between sexes.

Table 3.

RMS during dynamic expressions at different treatments

Values are expressed as M ± SD. T1-T0: overall facial change at 3-month follow-up relative to baseline; T2-T0: final overall facial change at treatment completion relative to baseline

Table 4.

Comparison of treatment-induced changes in different facial expressions between early and full-course phases

ΔT1: Early-stage change (mean RMS of T1-T0).ΔT2: Full-course change (mean RMS of T2-T0). *p < 0.05; **p < 0.01; ***p < 0.001

Fig. 5.

Comparison of treatment-induced changes in different facial expressions between early and full-course phases

Table 5.

Subgroup analysis of T2-T0 RMS improvement by sex

Values are expressed as M ± SD

The 3D facial landmark displacements at different treatment stages are presented in Tables 6 and 7 and illustrated in Figs. 6, 7 and 8. Analysis revealed significant changes in the positions of multiple facial landmarks following intervention. Significant anterior displacements were identified in the midfacial and perinasal region. Landmarks b, c, and d all exhibited significant anterior displacement at both time points: b (T1-T0: 0.650 ± 0.814 mm, p = 0.014; T2-T0: 1.469 ± 1.383 mm, p < 0.001), c (T1-T0: 0.409 ± 0.529 mm, p = 0.016; T2-T0: 0.816 ± 0.730 mm, p = 0.003) and d (T1-T0: 0.722 ± 0.829 mm, p = 0.009; T2-T0: 1.018 ± 0.907 mm, p = 0.003). Landmark e, which showed no significant displacement T1-T0, exhibited a significant anterior displacement T2-T0 (0.865 ± 1.336 mm, p = 0.047). With regards to perioral landmarks, landmark f exhibited a significant anterior displacement that increased by more than two-fold in magnitude T2-T0, from 0.928 ± 1.007 mm (p = 0.006) to 1.899 ± 1.566 mm (p = 0.002). Landmark h also showed significant anterior displacement at T2-T0 (1.037 ± 1.333 mm, p = 0.021). In summary, treatment induced statistically significant anterior displacements in the nasal (landmarks b, c, d and e) and superior labial (landmarks f and h) regions, accompanied by a significant posterior displacement of landmark a at T2-T0 (-0.783 ± 0.520 mm, P < 0.001). There were no significant positional changes for landmarks g, i and j.

Table 6.

Midface landmark displacements with different treatments

*p < 0.05; **p < 0.01; ***p < 0.001

Table 7.

Lower facial third landmark displacements with different treatments

*p < 0.05; **p < 0.01

Fig. 6.

The displacement of midfacial landmarks observed at T2-T0

Fig. 7.

The displacement of lower facial third landmarks observed at T2-T0

Fig. 8.

Landmark displacements at T2-T0

Discussion

The main purpose of this study was to conduct a comprehensive static and dynamic analysis of facial soft tissue in children and adolescents with Angle Class III malocclusion undergoing maxillary anterior traction, by creatively applying the 3dMD dynamic facial capture system. For many years, the evaluation system utilized in this field has relied heavily on cephalometric technology, with a specific focus on the static improvement of profile lines [20]. However, there remains a notable lack of effective quantitative tools and in-depth discussion relating to the dynamic aesthetic changes presented by the face in the key life activity of facial expression [21].

Facial form is the most complex and recognizable structure in humans, and the finely coordinated movement of facial form provides the face with a wide diversity of beauty and harmony. As the requirement for aesthetic appearance increases in modern society, clinicians should not only consider static aesthetics, but also include dynamic aesthetic indicators such as speech communication and changes in expression when utilizing an evaluation system to evaluate the results of treatment [22].

Previous studies have found that children and adolescents with Angle Class III malocclusion have experienced significant positive changes in the soft tissue of the midface after receiving anterior traction. This result is reflected in the forward displacement of landmark points such as the pronasale, alare and labiale superius in the static 3D shape, visually demonstrating the contour optimization generated by the forward growth of the maxilla [23, 24]. Furthermore, by adopting the 3dMD dynamic system, we captured, for the first time, a significant increase in RMS values during key facial movements, including smile lips open, lip purse, and cheek puff, following traction treatment. As an orthopedic treatment, anterior traction provides a structural basis for the forward development of the maxilla by stimulating the patient’s own growth potential and promoting bone suture reconstruction [25–27]. The soft tissue above the maxilla was then reshaped and placed in a biomechanical environment more conducive to muscle function, thereby achieving a coordinated improvement from static form to dynamic function [28].

In the past, the use of traditional two-dimensional imaging technology in facial kinematics research has been associated with inherent limitations. For example, marker points have been shown to cause skin deformation, some patients experience difficulty when maintaining expressions, resulting in poor data consistency, and the 3D motion trajectory of landmark points cannot be restored [29–31]. The 3dMD dynamic analysis system successfully overcomes these shortcomings and provides reliable data support for this research with its non-contact, high precision, and high time resolution characteristics [32].

In this study, we also provided patients and parents with visual and scientifically supported renderings of changes in ‘face value’. The 3dMD dynamic analysis system makes the treatment results more intuitive while significantly enhancing the confidence and cooperation of patients with treatment, thereby indirectly improving the treatment success rate [33]. Moreover, by analyzing soft tissue data from different treatments, we can inversely infer the adaptive growth and remodeling of the hard tissue underneath the soft tissue, thus providing a new perspective for understanding the biological mechanism of correction [34, 35]. In this study, we combined fixed-point measurement with dynamic RMS analysis to provide a solid data foundation for establishing mathematical models and prediction models for craniofacial soft and hard tissue changes, and to promote the evaluation of orthodontic efficacy towards the integration of a morphology-function-dynamic approach.

This study has several limitations that need to be considered. First, the primary limitation is the absence of a control group. As a pre-post study design, our analysis compared changes within subjects across treatment stages but did not include a parallel comparison to untreated patients or alternative treatment modalities. This limits our ability to definitively attribute all observed changes solely to the anterior traction therapy, as natural growth in untreated Class III patients could not be accounted for. Second, the sample size was relatively limited, and the follow-up period was short. Future studies should incorporate well-designed control groups and employ multi-center collaborations with larger sample sizes and longer follow-up durations to verify the generalizability and stability of our findings and to further investigate the relationship between dynamic parameters and the long-term stability of orthodontic correction [36].

Conclusion

Our findings reveal that the facial adaptation of our patients to anterior traction was both favorable and non-linear, culminating in improved midfacial convexity. By utilizing 3dMD technology, we captured the precision and dynamics of these morphological changes, thus highlighting the importance of early treatment. Furthermore, this system not only provides patients with objective, visual feedback that clearly demonstrates treatment progress—a highly patient-friendly approach—but also offers clinicians a vital, data-driven prognostic tool, thus greatly enhancing the predictability and personalization of patient management.

Author contributions

J.H., Q.L. and L.W. were responsible for manuscript preparation and data collection; J.Z., B.L. and H.Z. prepared figures; J.L. contributed to experimental design and data analysis; K.Q. was in charge of article review and approval of the final version.

Funding

This work was supported by the National Clinical Research Center for Oral Diseases (Reference: LCA202202) and the Key Project of the Shaanxi Provincial Key Research and Development Program (Reference: 2025SF-YBXM-411).

Data availability

The data featured in this article will be made available at reasonable request to the corresponding author.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of the Stomatological Hospital of the Fourth Military Medical University (Approval Number: KQ-YJ-2025-179). All methods adhered to the principles of the Declaration of Helsinki.

Consent for publication

Before initiating the study, informed consent for publication was obtained from all patients.

Competing interests

The authors declare no competing interests.