Autism affects 1.5 per cent of all 10- and 11-year-olds and 2.5 per cent of four- and five-year-olds in Australia every year. Siblings of children diagnosed with autism have a higher risk of developing the disorder.
Early diagnosis and early intervention can help improve outcomes for children with autism. However, at the moment there is no method to diagnose the disorder in children before they start showing the symptoms.
Previous studies have indicated brain-related changes which occur before behavioural symptoms emerge and based on this, researchers at the University of North Carolina at Chapel Hill and Washington University School of Medicine in St. Louis decided to investigate and detect brain differences.
=Q=
The study focussed on brain functional connectivity which indicates how the brain works together during different tasks and during rest.
The researchers scanned 59 high-risk six month old infants while they slept, using functional connectivity magnetic resonance imaging (fcMRI). The infants were deemed as high risk as they had older siblings with autism.
The fcMRI viewed activities across 230 neural regions of the brain. The scientists looked for areas with coordinated activities and focussed on connections associated with features of autism – language skills, repetitive behaviour and social behaviour.
The researchers used a computer technology called machine learning which trains itself to look for differences and separate the neuroimaging results into two groups – autism or non-autism. Machine learning can also predict future diagnosis.
One analysis predicted each infant’s future diagnosis based on 58 infants’ data to train the computer program.
With this method, the researchers identified that 82 per cent or 9 out of 11 infants would go on to have autism and also correctly identified which of the infants would develop autism.
11 out of the 59 infants in this group developed autism at age 2.
In another analysis which tested how well the results could be applied to other cases, the computer program predicted diagnoses for a group of 10 infants at an accuracy of 93 per cent.
Although further investigation is needed to replicate this study, neuroimaging may be a vital tool in early detection of autism in high risk infants and this can assist health care providers and parents to better prepare strategies needed to work with autism in children.
Source: Science Translational Medicine
