J. Neurosci. 30, 10612–10623 (2010)

Brain scans may be sufficient to identify people with autism spectrum disorder (ASD), thanks to a new application of a type of data analysis.

Christine Ecker and her colleagues at the Institute of Psychiatry at King's College London scanned the brains of 20 adults diagnosed with ASD and 20 other volunteers using magnetic resonance imaging. The researchers' support vector machine (SVM) analysis — which is also used in face recognition — searched the data for subtle differences in cortex morphology between the two groups, using several parameters previously linked to ASD, such as cortical thickness and cortical folding. This identified several small, mostly non-overlapping, differences.

When participants were compared individually with data from the groups, SVM analysis identified ASD in as many as 90% of cases, which is comparable to the accuracy of behavioural diagnosis.