As children grow, brambles of short brain connections are gradually pruned down to longer, stronger neural pathways. Research has shown this trend to follow a fairly standard curve during normal development to adulthood, and scientists are now using this information to create predictive models of brain maturation.
This approach allows for calculations of "brain age" that are based not on structural development, but rather on how well structures are communicating with each other. A new study, published online September 9 in Science, shows the new connection-based model to be at least 92 percent accurate in predicting whether a person was a child or an adult based just on the neural communication patterns in their brain.
The research team scanned the brains—using functional connectivity magnetic resonance imaging (fcMRI)—of 238 normally developing subjects aged 7 to 30, for five minutes. By comparing 200 of 12,720 key functional brain connections and assessing them through multivariate pattern analysis, researchers then predicted volunteer subjects' developmental status.
Some groundwork for the approach was described in a 2009 study, published in PLoS Computational Biology, in which some of the same researchers described ways in which children's brains were organized differently from those of adults. Younger brains, they had found, had more close connections among the more physically proximate brain areas, whereas older brains had stronger ties among brain regions that were far apart. This adult organization "lets you connect one node with another in a relatively short number of steps via special nodes," Damien Fair, of Oregon Health and Science University and lead author of that study, noted in a prepared statement last year.
The new analysis demonstrated that rather than resulting from a mix of trimming and building connections, development was most easily predicted (about 68 percent) by just the trimming of the vast number of childhood connections.
The scientists behind the study liken this neural connection curve to the standard height and weight measurements kids get each time they go to the doctor for a check-up. "When the patient deviates too strongly from the standardized ranges or veers suddenly from one developmental path to another, the physician knows there's a need to start asking why," Bradley Schlagger, a pediatric neurologist at Washington University in St. Louis and coauthor of the new study, said in a prepared statement. With enough fcMRI data across individuals at different developmental periods, the standard curve of a brain's connectivity changes could be used to look for abnormalities, similar to the way in which doctors assess physical measurements—especially if a child is already showing other indications that something is amiss.
Children with cognitive irregularities are often already subjected to a barrage of tests, including MRI scans. But, as Schlagger pointed out, these scans are "typically looking at the data from a structural point of view—what's different about the shapes of various brain regions." And in many instances, those tests can come back puzzlingly normal. Scientists have known for decades that connections among brain regions are just as important as the health of the regions themselves, and Schlagger's new research has shown that "MRI also offers ways to analyze how different parts of the brain work together functionally," he noted in the recent prepared statement.
Testing is never cheap, but adding five minutes' worth of fcMRI to assess brain connections "won't add that much cost"—especially if a child is already undergoing scans to look for structural abnormalities, Nico Dosenbach, a pediatric neurology resident at St. Louis Children's Hospital and lead author on the new paper, noted in a prepared statement.
Now that these baseline measurements have been established for a range of normal development, the team members hope to use the curve to study groups of individuals at risk for developmental disorders. "When a fraction of them later develop that disorder, you can go back and construct an analysis like this one that will help predict the characteristics of the next child at highest risk of developing the disorder," Schlagger said. And as a passive measurement it does not rely on a subject's ability or willingness to perform a task, a common current gauge for different disabilities.
Such an approach might also eventually be able to help parse out a collection of indicators for various developmental disorders. "The beauty of this approach is that it lets you ask what's different in the way that children with autism, for example, are off the normal development curve versus the way children with attention deficit disorder are off that curve," Schlagger said. "That's very powerful both clinically and from the perspective of understanding the causes of these disorders."
Images of brain (with areas of connection that grow stronger with development are in orange and those that are weakened during maturation are in green) courtesy of Science/AAAS