Computer vision and machine learning to identify Down syndrome phenotypes from facial biomarkers

From June 6^th to 9^th, Barcelona hosted the 3^rd International Conference of the Trisomy 21 Research Society, the premier scientific meeting for Down syndrome research, attracting basic and clinical scientists and practitioners from around the world.

In the first poster session of the conference, GTM researchers Alejandro González and Xavier Sevillano presented their poster entitled “Comparing 2D, 2.5D and 3D Facial Biomarkers Based on Geometric Morphometrics to Classify Down Syndrome Individuals Using Machine Learning”. This work explores the use of facial biomarkers as a non-invasive diagnosis tool for conditions associated with minor facial dysmorphologies, using Down syndrome as a case study.

In particular, the study compares the use of 3D and 2D models to obtain facial biomarkers that represent face shape accurately enough to allow machine learning algorithms classify an individual as Down syndrome or not.

This work is part of the collaboration between GTM, the European Molecular Biology Laboratory, the Centre for Genomic Regulation and the Institut Hospital del Mar d'Investigacions Mèdiques in the context of their research for ameliorating the lives of people with Down syndrome.

Acknowledgements

Xavier Sevillano acknowledges the support from Secretaria d'Universitats i Recerca del Departament d'Empresa i Coneixement de la Generalitat de Catalunya under grant ref. 2018-URL-Proj-022.