Can a Brain Scan Diagnose Autism?


The Journal of Neuroscience reported a few days ago that researchers from King’s College Institute of Psychiatry in London have used brain scans to correctly identify adult males with Autism Spectrum Disorder with a 90 percent success rate.

The brains of individuals with autism were found to have different patterns of thickness in the grey matter (the densely packed neural tissue of the cerebral cortex) when compared to the brains of control subjects. The differences were found in regions that handle language and behavior.

The researchers’ next step is to test the scans with children, hoping that this quick, objective measure could eventually be used to diagnose kids on the autism spectrum.

A 90 percent success rate for diagnosing autism—pretty good, right? But not so fast.As pointed out in an article in the Guardian’s science blog, a 90 percent success rate sounds accurate, but it actually translates into a very high “false positive” rate when you factor in the incidence of autism (around one percent). The example given showed that in a population of 10,000, around 2,000 children would be falsely diagnosed with autism using this test.

Of course it would be amazing if children could be diagnosed in the first year of life instead of the preschool years, as many kids are. But if there ever proves to be a quick, accurate biological test for autism, it’s going to be a while down the road.  As I describe in my recent column on the causes of autism, the disorder is so much more complicated and varied than we realize—hundreds of genes could be involved and in a different mixture depending on the child.  Unlikely that a simple test of any kind could pick up on something so complex.

Image: Flickr/Heungsub