DiGeorge syndrome, or 22q11.2 deletion syndrome, is a genetic disorder that can display itself in a variety of ways. It’s quite rare and children with the disease may have heart conditions, a poorly functioning immune system, and other medical problems. It has been difficult to diagnose because of its rarity and variety of symptoms, but researchers have known for a long time that DiGeorge syndrome also affects the look of the faces of the kids with the disease.
Because the visual cues are slight, and due to our multi-ethnic society, a physician can easily miss the telltale signs of DiGeorge, potentially leading to a wrong diagnosis and inappropriate followup care. Children of non-European ethnicities are apparently particularly difficult to diagnose.
Now researchers at the National Human Genome Research Institute have developed a computer vision algorithm that can be used to screen children for DiGeorge syndrome. They had access to over one hundred photos of DiGeorge children from Africa, Asia, and Latin America. Using this data they built their algorithm, which they then subjected to a test involving over 150 photos of children, including of European descent, with and without the condition. Impressively, the software made the correct diagnostic prediction 96.6% of the time.
The software doesn’t require anything beyond a photo, so should be easily portable to smartphones that can then be used by physicians right at the point of care.
Study in American Journal of Medical Genetics: 22q11.2 deletion syndrome in diverse populations…