Differences in behavior among people with autism spectrum disorder (ASD) are closely linked to differences in neuroanatomy – the shape of the brain – a team of neurologists from Boston College reported today in the journal science. This finding can help understand the causes of ASD and develop customized interventions.
The team uses artificial intelligence (AI) to study magnetic resonance imaging data from more than 1,000 people with ASD and compares these images with AI-generated simulations of what brains would look like without ASD.
We found that different people with ASD may have different brain areas affected, and thanks to AI-simulated brains, we were able to identify which specific brain regions differ in individuals with ASD. In addition, separating ASD-related variations in brain anatomy from unrelated variations reveals hidden links between individual differences in brain anatomy and symptoms.
Aidas Aglinskas, postdoctoral fellow at Boston College and co-author of the report
Autism varies from person to person, both in terms of symptoms and neuroanatomy. Previous research has suggested that there may not be a single set of neuroanatomical correlates common to all individuals with ASD.
Confirming these proposals was difficult because identifying ASD-specific neural changes is a challenging task, Aglinskas said. Brains are different because of many factors, including genetic variation, and not because of ASD, which is difficult to control in a research study.
The team overcame this barrier by using AI to identify ASD-specific models of neural variability, which then allowed the team to identify neural pathways specifically affected by ASD, said Aglinskas, who conducted the study with assistant professors of neurology. at Boston College Joshua Hartshorn and Stefano Anzelotti.
“ASD-related differences in brain anatomy can be ‘hidden’ between non-ASD-related differences,” Aglinskas said. “As a result, it was difficult to identify differences in brain anatomy that are related to differences in symptoms. We used AI to separate ASD-related differences from unrelated differences. “
The team set out to determine whether ASD-specific characteristics of brain anatomy varied in individuals in a way that was related to their symptoms. Previous studies examining individual differences in brain anatomy within ASD do not separate ASD-specific characteristics from other, unrelated individual differences in neuroanatomy, making it difficult to study the links between neuroanatomy and symptoms, Aglinskas said.
With MRI data from 1,103 study participants, the team uses an analytical method somewhat similar to “deep fakes” – difficult-to-detect simulated photos, videos and other images created using visual data models involving study participants, according to the report. .
Instead, the team uses computer-detected models to simulate what each ASD individual’s brain would look like without ASD. This was activated by a new AI technique that divides individual differences in brain anatomy into ASD-specific and non-ASD-specific characteristics, the team said.
“We were surprised to find that although we observed a large number of variations in brain anatomy between individuals with ASD across multiple dimensions, individuals were not grouped into separate, definite subtypes, as previously thought,” said Aglinskas. “At the level of brain anatomy, individual differences within ASD can be better captured by continuous measurements than by categorical subtypes, but importantly, this does not preclude the possibility of explicit subtypes being detected by other types of brain measurements, such as functional images. ”
Going forward, researchers point to the need to understand in more detail how these neuroanatomical differences affect behavior.
Angelotti said the team plans to use artificial intelligence tools to look beyond the brain structure for ways to better understand the diagnoses of ASD and the behavior of people with ASD.
“Two brains can be shaped very similarly, but they still work differently,” said Anzelotti. “There are a number of other aspects of the brain that we will need to look at to get a complete picture. We are currently focused on functional connectivity – a measure of how the brain is “connected”. The big question is whether this will show us something new about individual differences in ASD. The goal of this type of work is to be able to use brain imaging data to help develop personalized health approaches for those with ASD.
Reference in the magazine:
Aglinskas, A., et al. (2022) Contrasting machine learning reveals the structure of neuroanatomical variations within autism. science. doi.org/10.1126/science.abm2461.