Computer Program Uses Pictures To Identify Genetic Disorders
June 26, 2014 in News
There are more than 7,000 known genetic disorders, but only a small number of affected patients receive a clinical or genetic diagnosis, according to Medical News Today (Ellis, Medical News Today, 6/24). Genetic tests are unavailable for many genetic disorders, as the gene variants associated with them have yet to be discovered. As a result, doctors often rely on facial features to make diagnoses, NewScientist reports (Coghlan, NewScientist, 6/24).
Details of Study
The study was funded by the MRC, the Wellcome Trust, the National Institute for Health Research Oxford Biomedical Research Centre and the European Research Council (Medical News Today, 6/24).
The computer software was developed by researchers led by Christoffer Nellåker and Andrew Zisserman of the University of Oxford (NewScientist, 6/24).
To create the software, researchers first collected a database of 2,878 images that included:
- 1,515 healthy individuals’ pictures as controls (Medical News Today, 6/24); and
- 1,363 pictures of individuals diagnosed with one or more of eight known developmental disorders, such as Down’s syndrome, fragile X syndrome and progeria (NewScientist, 6/24).
The researchers’ algorithm then used the photos to progressively learn to identify which facial features to consider — such as corners of the eyes, nose and mouth — and which to ignore.
After recognizing relevant facial features, the program created a description of the face structure and compared the description with what it recognized in other photos.
Finally, it automatically grouped together patients who were found to have the same condition (Medical News Today, 6/24).
During initial testing, the software identified the correct disorder 93% of the time.
Since the initial testing, researchers have expanded the software to recognize 90 genetic disorders.
Based on the 2,754 faces currently in the database, researchers said the software makes it nearly 30 times more likely to make a correct diagnosis than by chance alone.
David FitzPatrick, a clinical geneticist at United Kingdom-based Western General Hospital, validated all the cases used during system training. He said the biggest benefit of the program is its ease of access, adding, “Worldwide, its main use will be in countries where you don’t have any access to clinical genetics at all.”
Nellåker noted that the system is “not sufficiently accurate to provide a rock-solid diagnosis, but it helps narrow down the possibilities” (NewScientist, 6/24).
He added, “A doctor should in future, anywhere in the world, be able to take a smartphone picture of a patient and run the computer analysis to quickly find out which genetic disorder the person might have” (Medical News Today, 6/24).