An interactive atlas of three-dimensional syndromic facial morphology

Abstract

Craniofacial phenotyping is critical for both syndrome delineation and diagnosis because craniofacial abnormalities occur in 30% of characterized genetic syndromes. Clinical reports, textbooks, and available software tools typically provide two-dimensional, static images and illustrations of the characteristic phenotypes of genetic syndromes. In this work, we provide an interactive web application that provides three-dimensional, dynamic visualizations for the characteristic craniofacial effects of 95 syndromes. Users can visualize syndrome facial appearance estimates quantified from data and easily compare craniofacial phenotypes of different syndromes. Our application also provides a map of morphological similarity between a target syndrome and other syndromes. Finally, users can upload 3D facial scans of individuals and compare them to our syndrome atlas estimates. In summary, we provide an interactive reference for the craniofacial phenotypes of syndromes that allows for precise, individual-specific comparisons of dysmorphology.

Keywords: 3D facial imaging; diagnosis; genetic disease; human; morphometrics; penetrance; visualization.

Figure 1

Syndromic heterogeneity and measurements of syndromic gestalts using achondroplasia as an example (A) The most-severe, average, and least-severe shape gestalts from a sample of 50 individuals with achondroplasia. Heatmaps represent the per-vertex deformation relative to the achondroplasia mean shape. (B) An image of “facial archetype” generated with non-linear deformation of 2D screenshots to the achondroplasia mean.

Figure 2

Age- and sex-specific syndrome atlas estimates for Crouzon syndrome and Sotos syndrome The left column shows the age- and sex-specific estimated facial shape, and the right column shows a heatmap comparison between the syndrome and the age-/sex-matched non-syndromic expectation.

Figure 3

Comparison of age- and sex-matched model predictions for Nager syndrome and Van der Woude syndrome Each point in the plot represents the mean of the residuals across all mesh vertices for one syndromic individual. The heatmaps represent the differences between the model prediction and an age-, sex-, and syndrome-matched individual for the prediction with the least error (left) and most error (right). The mesh shows the shape of the individuals, and the heatmap colors show the differences to the model estimate with red showing areas that project outwards from the average shape mesh and blue showing areas that project inwards from the mesh.

Figure 4

Classification of syndrome atlas predictions and determinants of classification sensitivity (A) Classification of interpolated syndrome atlas predictions. Predicted shapes for syndromes were constrained to the age range observed for each syndrome and classified. (B) Classification of syndrome atlas prediction solely for age ranges not observed for each syndrome. (C) Relationship between classification sensitivity and sample size, average Procrustes distance to the non-syndromic mean face, the standardized variance of the eigenvalues, and the trace of the covariance matrix for each syndrome. Procrustes distance measures shape differences among configurations of landmark coordinates. The variance of eigenvalues measures the extent to which variation in shape is concentrated on the first few principal components. Each point represents an individual syndrome.

Figure 5

Panel views of the syndrome atlas application The application supports a mobile view driven by touch interactions (A–C) as well as a desktop view with mouse/keyboard controls (Figures S9–S11). (A) The gestalt tab focuses on individual syndrome visualization. Here we see an estimated face that would correspond to a 6-year-old male with severe craniofrontonasal dysplasia. (B) The comparison tab allows for the comparison between any syndrome or non-syndromic group. Comparisons can be limited to subsets of the facial morphology shown in the bottom box. The morphospace plot on the bottom shows the similarity of all syndromes to the specified syndrome. (C) The submitted face tab allows for the submission and registration of a novel mesh. The registered mesh is then run through a syndrome classifier (middle row) and projected onto syndromic principal component axes (bottom row). In the principal component space, the registered mesh is highlighted among age- and sex-matched projected atlas predictions for each syndrome.