Three-Dimensional Photogrammetric Analysis of Soft Tissue Facial Morphology in Autism

Background: Subtle dysmorphogenesis of the craniofacial region may constitute important corroborating evidence of the neuro-developmental aspects of various disorders. The etiology of autism remains unclear, but some data suggest aberrant circuitry in the cerebral cortex and impairments in brainstem, cerebellar, thalamic and basal gangliar connections. Effects on motor neurons which innervate the facial musculature and soft tissues suggest possible consequences related to the development of facial morphological patterns. Previous studies have associated facial features with neuro-developmental disorders, but there is paucity of data on facial morphometrics in autism.  Assessment of facial features also has been challenging due to the subjective nature of conventional analytical methods.

Objectives: The aim of this study is to objectively analyze the facial characteristics in autism using three-dimensional (3D) computerized surface tissue models (stereophotogrammetry) in comparing facial features of children with autistic disorder (AD) and those with typical development (TD).  

Methods: Twenty-nine Caucasian AD children (21 males, 8 females, age range: 2-7 years, mean: 4.2 years) and 29 Caucasian age- and gender-matched TD children were imaged using stereophotogrammetry (3dMDface System™).  Autism diagnoses were based on ADI-R, ADOS and clinical judgment. The following age groupings were formulated for both AD and TD: Group I, age 2-3 years, n=17; 12 males, 5 females; Group II, age 4-7 years, n=12; 9 males, 3 females.  Computerized dense surface models (DSM) were marked with 14 homologous facial landmarks and, using Procrustean superimposition, were combined to form mean 3-D facial models.  Analyses were performed using the following Morphostudio^TM algorithms: (1) finite-element analysis (FEA), which compared deviations in facial forms as a function of volume; (2) function manager analysis (FMA), which calculated numerical linear and angular parameters; and (3) principal components analysis (PCA)), which formulated graphical outputs calculated from the position of individual DSM’s in the X-and-Y axis modal space.

Results: Comparisons of mean AD and TD facial models yielded no significantly measurable differences between the two groups.  FEA-calculated volume and deformation factors revealed no statistically significant differences between AD and TD in both age groups (p>0.05).  Linear dimensions (e.g., intercanthal distance, nose width, upper-, mid- and lower-facial lengths) and angular measurements (e.g., lower facial profile, maxillary complex profile) generated by FMA yielded no numerical parameters that exceeded levels of significance.  PCA-generated graphical outputs (formed by enclosing ellipses around the groupings) revealed no significant differences in the geometrical dimensions outlined in the modal spaces.  The near complete merging of AD and TD ellipses for both age groups demonstrated the absence of meaningful deviations between the two groups.

Conclusions: Computerized morphometrics objectively quantified the facial features in autistic children. Our analyses revealed no statistically significant deviations in any of the measurements or indices between AD and TD. As this study has been conducted in very young subjects undergoing periods of active facial growth, future analyses may pursue possible significant facial patterns that emerge in older autistic populations.