A new artificial intelligence system may diagnose autism in children earlier, according to research presented this week at the Radiological Society of North America’s annual meeting in Chicago.
A newly developed system that analyzes specialized brain MRIs accurately diagnosed children between 24 and 48 months with autism with 98.5% accuracy, researchers said.
A multidisciplinary team at the University of Louisville has developed a three-step system for analyzing and classifying Diffusion Tensor MRI, or DT-MRI, of the brain. DT-MRI is a special technique that looks at how water travels along white matter pathways in the brain, according to a news release.
“Current tools for diagnosing autism are subjective, especially when assessing individuals near the borderline between autism and typical development,” study co-author Ayman El-Baz, professor and chair of the Department of Bioengineering at the University of Louisville, told UPI. by e-mail.
“Therefore, there is an urgent need to develop a new, objective technology for the early diagnosis of autism.”
The AI system involves isolating images of brain tissue from DT-MRI scans and extracting imaging markers that indicate the level of connectivity between brain regions.
A machine learning algorithm compares marker patterns in the brains of children with autism to those of people with normal brain development.
“Autism “It’s primarily a disease of miswires in the brain,” said co-author Dr. Gregory N. Barnes, professor of neurology at the University of Louisville and director of the Norton Children’s Autism Center in Louisville.
“DT-MRI captures these abnormal connections that lead to common symptoms in children with autism, such as impaired social interaction and repetitive behaviors.”
The researchers applied their technique to DT-MRI brain scans of 226 children aged 24 to 48 months from the Autism Brain Imaging Data Exchange-II. The dataset included scans of 126 children with autism and 100 typically developing children.
Researchers have found that therapeutic intervention as early as age three can lead to better outcomes, including that people with autism can achieve greater independence and higher IQs.
According to the CDC’s 2023 Autism Community Report, less than half of children with autism spectrum disorder received a developmental assessment by age 3, and 30% of children meeting criteria for autism spectrum disorder did not receive a formal diagnosis by age 8.
“Early intensive behavioral intervention between one and three years of age can be very effective because of a phenomenon called neuroplasticity in the infant’s brain,” El-Baz said.
The autism assessment begins with the researchers’ AI system, followed by a brief session with a psychologist to confirm the results and guide parents on next steps. They say that this can reduce the workload of psychologists by up to 30%.
Investigators are looking to commercialize and obtain approval from it Food and Drug Administration for their AI software.
Dr. Leandra Berry, associate director of clinical services at the Texas Children’s Hospital Autism Center in Houston, said: “I support any diagnostic technology that helps children get an earlier and more reliable diagnosis of autism, which helps children get faster access to evidence-based interventions,” told UPI in a phone interview.
However, he added, “More studies need to replicate this finding before we can really adopt this technology.”
Because the patients in the study ranged in age from 24 to 48 months, he noted, the study could not decide whether the technology would benefit younger children.
Berry also noted in research studies that MRIs and other scans are typically paid for by grants, but insurers cannot cover expensive imaging tests, adding that access to the technology is limited, especially in rural areas.
Although these data may indicate a diagnosis, the expert must communicate this information to the family. “There’s still a human physician component that’s critical,” Berry said.
Diana Robins, professor and director of the AJ Drexel Autism Institute at Drexel University in Philadelphia, told UPI in a phone interview that “early detection is critical to reducing autism-related disability and improving positive outcomes for autistic people.”
However, “you need to include children with other developmental delays in your sample before you can conclude that you have autism,” said Robins, Ph.D.
“It’s important to distinguish between co-occurring diagnoses,” Dr. Susan Hyman, a professor of developmental and behavioral pediatrics at the University of Rochester’s Golisano Children’s Hospital in Rochester, N.Y., told UPI in an email.
He said that about half of autistic children have attention deficit/hyperactivity disorder, 40% of autistic people with autism have mental retardation, and 25% of autistic people have seizures that usually start in early childhood or adolescence.
“MRIs for autistic children without seizures or neurological findings in the community rarely provide information that informs care at an individual level,” Hyman said.
It is also difficult for young children to remain still for an MRI.
“You have to try multiple times, you have to try to scan them while they sleep before going to bed, or you have to sedate them, and that comes with a medical risk,” Robins said, adding that parents may not consent.
Robins noted that there is no data to show how much the technology will reduce the workload of psychologists. “It (the technology) is a long way from being ready for the public,” he said.