Professor Peter Hammond from the Institute of Child Health at the University College London has developed software that can scan facial features of children, and offer a likely diagnosis from over 700 genetic conditions. According to Dr. Hammond’s report, presented at the ongoing BA Festival of Science by the British Association for the Advancement of Science, his software is pretty accurate in making the diagnosis:
There are over 700 genetic conditions associated with characteristic facial features. Fragile X, the most common form of inherited mental impairment, affects one birth in 4,000. Children with the condition have longer, narrower faces, with bigger jaws and ears that stick out very slightly.
Smith-Magenis syndrome affects one birth in 25,000. It produces a nose with a flat bridge and an upper lip that looks as though it’s been lifted up from the inside. Children with this condition can be aggressive and self-harming. They sleep in the daytime and are awake at night.
The new software has identified Fragile X faces with an accuracy of 92%; Smith-Magenis syndrome at 91% and Williams syndrome at 98%.
A snippet on how the software works:
The specially written software is based on dense surface modelling techniques developed at UCL and compares the child’s face to groups of individuals with known conditions and selects which syndromes look most similar. In order to do this, extensive collections of 3D face images of children and adults with the same genetic condition had to be gathered, as well as controls or individuals with no known genetic condition. Each image contains 25,000 or so points on a face surface capturing even the most subtle contours in 3D. The images are then converted to a compact form that requires only a 100 or so numeric values to represent each face in the subsequent analysis.
Once the software has narrowed down conditions with similar facial features, molecular testing can then be used to confirm the diagnosis. Testing for fewer conditions will save money, time and reduce the amount of stress the child and the parents are put under.
So far the technique has proved fruitful, Professor Hammond says: ‘The technique is currently being applied to over 30 conditions with an underlying genetic abnormality. The discriminatory capability of the approach has proven highly accurate in identifying the characteristic facial features of a variety of genetic conditions, including Cornelia de Lange, Fragile X, Noonan, Smith-Magenis and Velocardiofacial syndromes. It has identified unusual facial asymmetry in children with autism spectrum disorder reflecting known brain asymmetry and has helped to identify genes affecting facial development in Williams syndrome.’
Press release @ the-BA: Faces reveal genetic conditions …
Press release @ UCL: Facial characteristics offer insights into genetic conditions …