Facial Imaging-Based Genetic Diagnoses Appears to Get Boost With Three-Dimensional Approach

A University of Calgary-led research team reporting in the European Journal of Human Genetics considers facial phenotype-based strategies for diagnosing genetic syndrome with computational approaches relying on two-dimensional or three-dimensional facial imaging representations. Based on data for more than 1,900 faces from individuals affected by 43 genetic syndromes, the investigators found that a three-dimensional facial representation-based approach appeared to provide more reliable face-based diagnoses relative to two-dimensional representations of the same set of faces. "Our experiments isolate the effects of using different facial representations on diagnostic model accuracy, providing empirical justification for continued research, data collection, and model development using 3D facial imaging modalities," the authors write, noting that "3D surface-based syndrome classification models significantly outperformed 2D image-based models trained and evaluated on the same subject faces.