The human face is complex three-dimensional structure that makes each of us uniquely distinguishable, but strongly determined by genetic factors. Consequently, many developmental, psychiatric and genetic abnormalities have defined facial morphological features. However, the underlying complexity of facial morphology cannot be fully captured by simple geometric measures. Rather, it is now increasingly clear that the genetic determination of facial morphology and its relation with health outcomes requires more sophisticated quantitative approaches for capturing facial morphology. Recent advances in computational and methodological approaches have made possible accurate and precise derivation of facial traits.
This project will focus on developing methods (based on machine learning and deep learning technologies) to derive complex facial measurements in a unique dataset comprising thousands of 3D facial images of children from the prospective, longitudinal, population-based
Generation R birth cohort and elderly population-based
Rotterdam Study. 3D facial morphology exists in a high-dimensional feature space, which cannot be analyzed using simple Euclidean metrics. Accordingly, the ultimate aim of this project is to leverage the large-scale 3D facial imaging, which provides extensive genetic and epidemiological measures, to unravel the complexity between genetics, facial morphology and health outcomes.
The project leader is Prof.dr. Eppo Wolvius, Department of Oral & Maxillofacial Surgery, Special Dental Care, and Orthodontics. In collaboration with Prof.Dr. Wiro Niessen, Biomedical Imaging Group Rotterdam (BIGR); Dr. Fernando Rivadeneira, Department of Internal Medicine; Prof.Dr. Manfred Kayser, Department of Genetic Identification; Prof. Dr. Steven Kushner, Department of Psychiatry; Dr. Gennady Roshchupkin, Department of Medical Informatics, Department of Epidemiology; Dr. Stefan Boehringer, Biomedical Data Sciences, Leiden University Medical Center.