Researchers at the Center for Translational Research in Neuroimaging and Data Science (TReNDs) at Georgia State have identified important new methods for pinpointing potential biomarkers in the adolescent brain that can reliably predict cognitive developments and psychiatric issues.
A new study, published in Nature Mental Health, represents the first large-scale analysis of its kind in which researchers analyzed functional network connectivity (FNC) across scans and identified associations with a wide range of health measures in children. The researchers believe that inferences about early cognitive and psychiatric behaviors in children can be made using these within-subject variables as a useful biomarker.
The researchers studied four scans from more than 9,000 people aged 9 to 11 years.
Neuroscientist, University Distinguished Professor and head of the TReNDS Center at Georgia State Vince Calhoun worked with the research team to develop the study. He said the research shows that, regardless of brain growth and development, a child’s FNC is robust and stable with high similarity across scans and can serve as a fingerprint to identify an individual child from a large group.
“This study is quite exciting as it shows the promise of using advanced machine learning to identify brain patterns that could help us intervene early in children who are at higher risk for cognitive or psychiatric problems,” said Calhoun, who is the senior author of the study.
Functional brain connectivity derived from functional magnetic resonance imaging (fMRI) is commonly used as a potential blueprint for adults, the researchers say. But they believe that intra-subject variation in FNC may convey biologically important information, especially during adolescence, which is a time of significant change in the brain.
Lead researcher Zening Fu said the study demonstrates that variability in functional connectivity can predict a wide range of children’s behavior, including cognitive function, mental health and sleep conditions.
“Most previous fMRI studies believe that resting-state functional connectivity can provide a fingerprint of an individual, and that variability in connectivity is due to noise or other confounding effects,” Fu said. “However, we found that variations in individualized FNC across scans are remarkable and convey psychological and physiological information underlying distinct behavioral phenotypes in children. Multivariate methods could help capture much larger effects between FNC stability and behavior of the children.”
The research team was able to predict with surprising accuracy a number of conditions or outcomes, including cognitive performance and psychiatric problems. The researchers were also able to predict sleep conditions and screen use based on FNC stability. Additionally, they were able to identify brain-behavior associations with parental psychopathology and prenatal exposure to marijuana and other drugs.
Fu explained how they are able to read the results and, in many cases, predict the results in the children based on the scans over time.
“FNC stability in our present work is defined as the variability or changes in resting-state functional connectivity across scans (measurements),” Fu said. “That is, if a subject has been collected using resting-state fMRI scans multiple times, the functional connectivity estimated using each fMRI scan should be different, even if they are from the same subject. This difference or variability is not trivial; but biologically Individuals with greater FNC variability (less stability) may tend to have lower cognitive performance and more mental health problems.”
In a second study, published in Biol Psychiatry, research conducted at the TReNDS Center led by Weizheng Yan finds that functional network connectivity, which is consistently remodeled over time, contains potentially rich information for psychiatric risk assessment. Yan is a former TReNDS Center postdoctoral research fellow and now works with the National Institutes of Health.
As part of the study, the researchers developed a brain-level risk score (BRS), a new FNC-based measure that contrasts the relative distances of an individual’s FNC to that of psychiatric disorders versus healthy control references.
The research team found that the BRS revealed a distinct, reproducible gradient of FNC patterns for each psychiatric disorder in more than 8,000 unaffected adolescents, ranging from low to high risk. The BRS could also identify individuals with early psychosis from healthy controls and predict psychosis scores.
To generate group-level reports of disorder and healthy controls, the researchers used a large brain imaging dataset containing more than 5,000 individuals diagnosed with schizophrenia, autism spectrum, major depressive and bipolar disorders and their matched healthy controls.
The findings suggest that the BRS could be a novel image-based tool for assessing psychiatric vulnerability over time and in unaffected individuals and could also serve as a potential biomarker, facilitating early screening and follow-up interventions .
Both studies used a multimodal database known as the Study of Adolescent Brain Development (ABCD). The dataset contains a wide range of measures of mental health, cognitive ability, and other health-related factors that have been found to be useful in examining the link between adolescent behaviors and brain function.
The Translational Research in Neuroimaging and Data Science Center (TReNDS) is a collaboration between Georgia State University, the Georgia Institute of Technology, and Emory University. It focuses on the development, application, and sharing of advanced analytical approaches and neuroinformatics tools that leverage state-of-the-art brain imaging and large-scale data analysis with the goal of translating these approaches into biomarkers that can help address relevant areas of brain health and disease.