A newly developed artificial intelligence system which analyzes specialized MRI scans of the brain, accurately diagnosed children ages 24 to 48 months with autism with an accuracy rate of 98.5%.

The results are being presented next week at the annual meeting of the Radiological Society of North America.

The interdisciplinary team developed a three-step system to analyze and classify diffusion MRI, a special technique that detects how water travels along white matter pathways in the brain.

The artificial intelligence system involves the isolation of images of brain tissue from MRI scans and extracting imaging markers that indicate the level of connectivity between brain regions. A machine learning algorithm, trained to spot areas of deviation to diagnose whether someone is autistic or neurotypical, compares the patterns of markers in the brains of children with autism to those of normally developing brains.

The researchers applied their methodology to the MRI scans of 226 children aged 24 to 48 months, of which 126 children have autism and 100 children are developing normally. The system demonstrated 98.5% accuracy in identifying children with autism.

“Our approach is a new advance that allows early detection of autism in infants as young as two years old,” points out Mohamed Koudri, a visiting researcher at the University of Louisville in Kentucky and a member of the research team. He adds: “We believe that therapeutic intervention before the age of three can lead to better outcomes, including enabling people with autism to achieve greater independence and higher IQ.”

According to the US Centers for Disease Control and Prevention (CDC) report on autism, in 2023 fewer than half of children with autism spectrum disorder received a developmental assessment by age three, while 30% of children who met criteria for autism spectrum disorder did not receive a formal diagnosis by age eight.

“The idea behind early intervention is to take advantage of brain plasticity, or the brain’s ability to normalize its function with treatment,” says Gregory Barnes, a Louisville professor of neurology and one of the study’s authors.