Imagine knowing that your infant is on the autism spectrum before they could even sit-up on their own. The researchers at the UNC Autism Research Center are doing just that.
It’s been long understood that children with autism spectrum disorder have increased brain volume. But a new discovery from UNC researchers narrows diagnosis of brain overgrowth to 12 months old, much earlier than the previous diagnostic test at 2 years old.
In February 2017, Nature (https://www.nature.com/articles/nature21369) published UNC research showing with MRI imaging and artificial intelligence, a deep-learning algorithm predicted the diagnosis of autism in high-risk children with MRI scans at 6 and 12 months with an 81% accuracy. By looking at brain growth, researchers were able to see the markers of brain over-expansion in autism by 12 months, significantly before behavior based diagnosis at 2 years old.
Having the knowledge that a child has autism by 12 months gives parents an 12-month jump on early interventions that could help keep a child on track to developing with their peers.
The diagnostic test is best used for high-risk children who have a sibling with an autism diagnosis, but this diagnostic test could potentially be used by pediatricians with an early hunch. This research could also lead to more predictive diagnoses of the range of severity of autism and cognitive function.
Martin Styner, co-director of the Neuro Image Analysis and Research Lab at UNC, is the brains behind the deep-learning algorithm on the expansive team of researchers at the UNC Autism Research Center. His interest and study has been in autism and brain development for a long time but having a son with autism 11 years ago has shaped his research.
“I wish I had this diagnostic test when my second child was born nine years ago,” Styner said. “While my second son is not on the spectrum, this information would have helped me predict more about my son’s development, what to expect, and prepare for early interventions.”
This research has come to light because of a strong collaborative team at UNC. “I don’t think this could have been possible any where but at UNC,” Styner said. The combination of machine learning algorithms with the clinical power that we have as well as the interventional knowledge is unique to the team at the UNC Autism Research Center.
Through more funding, this important research can forge ahead to discover more information to help children with autism. Styner’s drive to enable the work of his peers with new tools and new ways to start early intervention. “This is an exciting time for machine learning. This work is distinctively new with the artificial intelligence work. We could not have predicted with this nuance with techniques of the past.”
To learn more about how you can support the UNC Autism Research Center contact Aron Johnson at aron_johnson@med.unc.edu or 919.843.9902.