Causal Inference Method Attenuates Motion Bias in Autism Imaging Studies | Spectrum | Autism Research News

Film: A typical approach to analyzing differences in functional connectivity between autistic and non-autistic children shows fewer group differences (left) than a new statistical approach guided by machine learning (right).

Young children with autism and those with important traits tend to be excluded from functional magnetic resonance imaging (fMRI) studies because they move your head too much during scans, according to new research.

The study also shows that applying a “missing data” approach that combines statistical modeling with machine learning can help solve the problem.

Findings suggest findings from autism imaging studies are biased, says Marie Nebelprincipal investigator of the study and a researcher at the Kennedy Krieger Institute in Baltimore, Maryland.

Because researchers often discard fMRI data from participants whose movements exceed a threshold, “we know we’re capturing some unrepresentative subsets of children who have an autism diagnosis,” says Kami Koldewynreader in psychology at the University of Bangor in Gwynedd, Wales, who was not involved in the work.

“It’s really nice to see this kind of confirmation of something that, to some degree, we all know in the field,” she says of the new work.

NOTEbel and his colleagues analyzed the resting-state fMRIs of 545 children aged 8 to 13 years. None of the children have intellectual disabilities and 148 of the 173 autistic children are boys. The researchers also measured characteristics of autism and attention-deficit/hyperactivity disorder in children using the Autism Diagnostic Observation Program and standard parent questionnaires.

Two common quality control measures of head movement had excluded more autistic children than non-autistic children, the researchers found. The least selective metric excluded about 29% of autistic children, compared to 16% of non-autistic children, and the strictest metric excluded 81 and 60%, respectively.

The excluded children tended to be younger and to have greater cognitive and social difficulties, as well as poorer motor control. Each factor related to differences in functional connectivity, the researchers found, supporting their hypothesis that patterns in the types of participants excluded from fMRI datasets could bias the conclusions scientists draw about connectivity in the brains of people with autism. The book was published in May in NeuroImage.

VSAusal inference, a way of analyzing data, can help correct for bias in large observational studies in which “missing data is a big problem,” says Nebel.

By applying a ‘propensity’ causal inference model to fMRI data, Nebel and colleagues characterized the relationship between a participant’s traits and the usability of that person’s data; an “outcome” model captured the link between a participant’s traits and their functional connectivity.

The two-pronged approach essentially served to increase the usable data weight of underrepresented children in the sample, Nebel says. “Once you fit the model, you can extrapolate to make predictions about children who didn’t have usable functional connectivity data.”

For fMRI data collected using the less selective motion criteria, bias corrections were small, the authors wrote. The stricter criteria resulted in too few autistic children with usable data, so they could not test their approach. The stricter criteria would likely have introduced greater bias and therefore resulted in greater bias corrections using their approach, Nebel and colleagues wrote.

“It’s a very important document,” says Max Bertoleroa former researcher who is a software architect in neuroinformatics at We Imaging, a Missouri-based medical imaging software company. “Even if it’s not the final solution, they at least come up with something and see if it works.” Bertolero did not participate in the study.

Still, “there’s quite a bit of evidence in the field that children with autism are quite idiosyncratic,” Koldewyn says. Regarding brain changes, researchers must assume that the relationships between functional connectivity and other characteristics of high-movement children who pass quality control will be the same across all subsets. “We don’t know,” she said.

AWhile approaches such as Nebel and his team’s are new and important, they are no substitute for efforts to reduce motion during fMRI studies in the first place, Bertolero says. “It can be part of a toolkit to manage movement in addition to collecting more data, which is very expensive.”

We Imaging, for example, has had some success in combining biofeedback and engaging games to help participants stay still in an fMRI machine. Koldewyn also discovered that Short breaks during imaging can reduce the amount of movement a participant has during scans.

Post-imaging approaches and strategies to reduce movement during imaging are both important, Nebel says, not only for children with autism, but also for people with other conditions who are prone to movement during scans.

Cite this article:

#Causal #Inference #Method #Attenuates #Motion #Bias #Autism #Imaging #Studies #Spectrum #Autism #Research #News

Leave a Comment

Your email address will not be published. Required fields are marked *