Robust and regional 3D facial asymmetry assessment in hemimandibular hyperplasia and hemimandibular elongation anomalies

Abstract

Hemimandibular hyperplasia (HH) and hemimandibular elongation (HE) anomalies present with facial asymmetry and deranged occlusion. Currently, diagnosis and assessment of the facial dysmorphology is based on subjective clinical evaluation, supported by radiological scans. Advancements in objective assessments of facial asymmetry from three-dimensional (3D) facial scans facilitate a re-evaluation of the patterns of facial dysmorphology. Automated, robust and localised asymmetry assessments were obtained by comparing a 3D facial scan with its reflected image using a weighted least-squares superimposition. This robust superimposition is insensitive to severe asymmetries. This provides an estimation of the anatomical midline and a spatially dense vector map visualising localised directional differences between the left and right hemifaces. Analysis was conducted on three condylar hyperplasia phenotypes confirmed by clinical and CT evaluation: HH; HE; and hybrid phenotype. The midline extraction revealed chin point displacements in all cases. The upper lip philtrum and nose tip deviation to the affected side and a marked asymmetry of the mid face was noted in cases involving HE. Downward and medial rotation of the mandible with minor involvement of the midface was seen in the HH associated deformity. The hybrid phenotype case exhibited asymmetry features of both HH and HE cases.

Condylar hyperplasias can present a range of phenotypes including hemimandibular hyperplasia (HH), elongation (HE) and hybrid forms. They are rare anomalies mainly seen in early adulthood that can result in significant facial asymmetry. In the HH phenotype, the hyperplastic condyle can lead to increased height of the ascending ramus, displacing the body of the mandible inferiorly, coupled with a medial rotation. The lower border of the mandible is hyperplastic and terminated at the mental symphysis. Subsequent sequelae of dental alveolar compensation on the affected side can result in vertical facial discrepancy and a cant in the occlusal plane. The lip commissure slopes downwards towards the affected hemi-mandible and can remain evident. The combination of these skeletal dysmorphologies with a chin point that is often displaced to the unaffected side results in clinical presentation of a rotated facial form.

The clinical presentation of the HE phenotype includes a marked lateral displacement of the chin point to the unaffected side and depressed commissure of the lips on the affected side. Computed tomography (CT) or orthopantomogram (OPG) imaging reveal a unilateral slender elongated condylar neck, differentially to HH, the mandibular lower border is not hyperplastic but can be characterised by an open gonial angle with little discrepancy in the form of the rami of the affected side compared to the unaffected sides. Some dental alveolar compensation can occur, but is not as characteristic as in HH, as the tendency is a crossbite with lateral displacement, not an open bite with elongation in the ascending ramus as in HH.

Hybrid forms of these conditions also occur where a combination of unilateral HH and HE can lead to extensive facial anomaly associated with both lateral displacement and increased ramus height with a medially rotated hyperplastic lower border of the mandible. The anomaly in the mid and upper face can be quite striking with an increased height and sloping rima oris.

Conventionally, the differential diagnosis of these conditions is made with clinical examination supported by OPG, lateral cephalogram and CT imaging. The radiographic appearance of the hyperplastic condyle, the form of the mandibular ramus and angle, and the degree of hyperplasia are the essential phenotypic features used in diagnosis. Treatment planning is evolving to utilise emerging three-dimensional (3D) technologies. Primarily these involve rendered volumes of 3D CT imaging. Anatomical planes of symmetry are chosen to guide the reconstructive procedures coupled with orthodontic preparation to resolve the malocclusion.

The objective of the planned resolution of the dental–skeletal misalignment is to treat the patient’s complaint, primarily the presenting facial anomaly and deranged occlusion. A more direct and facial emphasised planning approach would require the means to assess the presented facial asymmetry objectively in HH and HE cases. Facial symmetry, or bilateral symmetry, is defined with respect to reflection or mirroring of the face across the median sagittal plane of the face. This midsagittal plane, by definition, divides a perfectly bilaterally symmetrical face into equal right and left halves. Since asymmetry involves absence of, or violation of symmetry, the unambiguous definition of this midsagittal plane becomes problematic. Different ad hoc approaches to calculating facial asymmetry and the associated midsagittal plane have been presented. Most of them are sensitive to and fail under high degrees of asymmetry. Recently, a robust (insensitive to severe asymmetries), automated and regional facial asymmetry assessment protocol was proposed by Claes et al. and has been applied to normal healthy individuals and those with abnormal and disordered development.

In respect to unilateral HH and HE with the corresponding sequelae of facial distortion to the midline and bilateral asymmetry, the means to provide a robust estimation of the true midline as well as quantify the degree and direction of the facial asymmetry would be invaluable. In this work the authors present, analyse and relate facial asymmetry assessments on three phenotypes associated with condylar hyperplasia: HH; HE; and hybrid phenotype.

Materials and methods

Preoperative 3D CT and facial scans in centric relation were obtained from 3 female patients presenting with significant facial asymmetry for corrective surgery. Based on clinical examination, a differential diagnosis of HH, HE and a patient with hybrid phenotype was made based on the criteria.

The first case, a 19-year-old woman presented with marked facial asymmetry, which had developed from late puberty. The patient’s facial appearance was typical of that reported in the literature with a confirmed diagnosed of HH. She presented with vertical elongation of the right side of the face, and downwards sloping lip commissure to the right. The occlusal plane was canted to the right with the dental midline coincident with the mid philtral line. There was a slight deviation to the chin point to the left, the right mandibular lower border was medially rotated and the left angle laterally displaced.

The second case, a 19-year-old woman presented with right side posterior open bite and crossbite, lateral chin displacement to the left and vertical orbital dystopia, consistent with HE dysmorphology.

The third case, a 26-year-old female patient presented with significant facial asymmetry with lateral chin displacement, inferiorly displaced left lower border of the mandible, vertical orbital dystopia, and occlusal cant. This presentation with features of HH and HE was diagnosed as a condylar hyperplasia exhibiting a hybrid phenotype.

Skeletal assessment

Fused CT with high resolution dental crown laser scan data were rendered into 3D manifold and imported into reverse engineering software (Geomagic Qualify version 11, Geomagic Inc., NC, USA) to extract Euclidean measures and volumes ( Fig. 1 ) of the skeletal dysplasias to quantify the ratio of bilateral discrepancy. Ratios of measures of affected and unaffected sides of the maxillary height ( Z -axis dimension between the most inferior point rima oris and talonid basin of the maxillary first molar); mandibular body height (greatest dimension from the talonid basin first molar to the lower border of the mandible); condylar neck length (the greatest dimension from the posterior–inferior contour of the condyle defined by the superior tangent of the ramal plane to the perpendicular line coincident with greatest concavity of sigmoid notch), ascending ramus height (the greatest dimension from the most superior aspect of the condyle to the inferior border of the angle of the mandible) and occlusal cant (angle subtended by the maxillary occlusal and Frankfort plane). A volumetric measure of the condyle was also made.

Fig. 1
A 19-year-old female presenting with HH. (A) 3D facial image with automated estimation of facial midline. (B) Asymmetry vector field mapped to facial image, arrows illustrating principal axes of deformation vectors. (C) Frontal and lateral rendered 3D CT images of craniofacial skeletal dysmorphology and condylar hyperplasia. Skeletal assessment variables are marked as follows: (1) maxillary height, (2) mandibular body height, (3) condylar neck length, (4) occlusal cant, and (5) condylar volume.

Facial asymmetry assessment

The facial images were acquired using a 3dMDface™ (two pod) system and consisted of a spatially dense set of 3D points connected to form a wireframe made of polygons representing the facial surface as a manifold. The precision and repeatability of the 3dMDface™ (two pod) system had been previously tested and validated with sub-millimetre resolutions.

The facial asymmetry assessment used in this study was adopted from previous and related work. An asymmetry assessment protocol, grounded in geometric morphometrics, to compare a face with its reflected image using a superimposition was originally proposed by Klingenberg et al. Facial shape is represented using the coordinates of a set of landmarks. These landmarks are defined as ‘points of correspondence on an object that matches between and within populations’. A Procrustes alignment of the original landmark configuration and its reflected image with respect to an arbitrary plane is then undertaken. Procrustes rigid alignment minimises the mean squared distance between paired landmarks by a rigid (only translations and rotations) superimposition. The reflected configuration is simply a reflection by changing the sign of the x coordinate of every landmark assuming that the images have been grossly aligned with the coordinate axes, the x -axis representing the left–right axis. The mean positions of the superimposed original and reflected points are, by construction, symmetrical with respect to a plane being an estimate of the mid-sagittal plane.

The original Klingenberg protocol was extended to obtain regional and robust assessments. Regional localisation of asymmetries was achieved using an anthropometric mask consisting of spatially dense and uniformly sampled quasi-landmarks that were automatically indicated on all 3D facial images. Robustness, in the sense of relative insensitivity of the analysis to increasing asymmetry, was ensured by an adaptive weighted least-squares superimposition.

The magnitude and direction of spatial discrepancies between corresponding quasi-landmarks in the original and reflected configurations after superimposition was calculated and visualised as a colour vector map projected onto the original manifold. These outputs reflect the localised degree and direction of asymmetry found in an individual. Summary statistics on the relative statistical asymmetry (RSA) given as a percentage of the face with asymmetry set at an arbitrary value of >2 standard deviations of the distribution of asymmetries between corresponding quasi-landmarks. As a measure of severity of the asymmetry, the root mean squared error (RMSE), which incorporates the variance and bias (average) of the distribution of Euclidean distances between quasi-landmarks is given in mm. The estimate of the facial midline is calculated from the robust statistical fitting of the anthropometric mask original to the reflected scans. This was superimposed onto the original facial scan to facilitate the assessment of the deviation of midline structures.

Results

CT evaluation

The mandibular ratios and condylar volumes extracted from the 3D CT data of all three cases can be found in Table 1 . All demonstrated an expanded condyle (ratio range 1.67–2.84) on the affected hemimandible with the HH and hybrid phenotype exhibiting the greatest discrepancy (2.01; 2.84, respectively). Similarly the condylar neck was longer (ratio range 1.32–1.52) in all cases, but it was the HE and hybrid presentation that had the largest elongation (1.5; 1.52, respectively). Likewise, ramal height is expanded (ratio range 1.12–1.27) on the affected side; the HE case demonstrated the least bilateral variance (1.12). The hybrid phenotype case had similar ratio values to the HH case in maxillary height (1.15 compared to 1.16) and mandibular height (1.21 compared to 1.15). These ratios were reduced in HE to 1.08 in the midface and 0.96 in mandibular height.

Table 1
Ratios of affected over contralateral Euclidean distances and volumes extracted from 3D CT images of the maxilla and mandible of three condylar hyperplasia cases: HH; HE; and a hybrid phenotype (hybrid).
Condylar hypoplasia cases Affected/contralateral ratios Maxillary occlusal cant (degrees)
Condylar volume Condylar neck length Ascending ramus height Mandibular height Mid facial height
HH 2.01 1.32 1.27 1.15 1.15 6.64
HE 1.67 1.51 1.12 0.96 1.08 5.04
Hybrid 2.84 1.52 1.26 1.21 1.16 8.71

Morphologically the mandible in HH and hybrid cases had an acute and rounded angle of the mandible ( Figs. 1 and 2 , respectively), inferiorly deflected and medially rotated about the symphysis. This contrasted to the presentation of HE with an obtuse angle of the mandible and the mandibular lower border on a similar plane to the contralateral side ( Fig. 3 ). All patients exhibited occlusal cants with the most severe exhibited in the hybrid phenotype case (8.71°) and the HE presentation the least (5.04°).

Jan 24, 2018 | Posted by in Oral and Maxillofacial Surgery | Comments Off on Robust and regional 3D facial asymmetry assessment in hemimandibular hyperplasia and hemimandibular elongation anomalies

VIDEdental - Online dental courses

Get VIDEdental app for watching clinical videos