The symptoms of some genetic diseases are so nonspecific that it can take years for a child to be diagnosed. Even when a doctor suspects a specific disorder, such as Noonan syndrome, a developmental disorder that can affect the skeletal system, heart, eyes, language, and speech, but usually not intelligence, testing can cost thousands of dollars and sometimes requires months to complete. Peter Hammond of University College London is developing a faster, potentially cheaper approach that uses computer analysis to spot facial characteristics associated with a variety of genetic disorders.
Hammond first projects a pattern of thousands of dots onto a patient’s face and then takes photos of the face from different angles with a digital camera, capturing the positions of the dots. Software converts the data into a three-dimensional “map” of the person’s face and compares this map to models of the face shapes linked with various genetic syndromes, including Noonan, Williams, and Fragile X. If the computer analysis indicates, for example, wide-set eyes, low ears, a small jaw, and drooping eyelids, the program might match it with Noonan syndrome, which often includes these features. Among children with one of the genetic disorders for which Hammond has compiled a face shape model, this technique has demonstrated greater than 90 percent accuracy.
So far, Hammond has modeled 12 of the 30 disorders he has studied. But your hospital might not be able to afford a scan machine yet. Custom-made to Hammond’s specs, they cost $40,000 to $60,000 each.