Summary: A new deep learning method that uses graphical convolutional neural networks (gCNNs) can predict cognitive function based on brain size and structure. The algorithm can provide insight into the relationship between brain morphology and different cognitive functions, as well as decline in cognitive function.
Source: Northwestern University
Northwestern researchers have developed a deep learning-based method that can predict cognitive function ability based on brain shape and structure, detailed in a study published in Scientific reports.
The method, which uses graphical convolutional neural networks (gCNNs), could also reveal new insights into the relationship between brain morphology and different cognitive functions as well as declines in brain function.
“When we apply the rich capabilities of CNNs to graphical representation of the brain, we can explore the brain as an image in ways previously unexplored,” said S. Kathleen Bandt, MD, assistant professor of neurological surgery. and co-author of the study.
Understanding how the relationship between brain structure and cognitive function changes throughout life has remained elusive. However, previous work suggests that fluid intelligence – the ability to problem solve, think and reason abstractly – is highly dependent on two regions of the brain: the prefrontal cortex and the parietal cortex, both of which are involved in decision-making and sensory perception. other functions.
Additionally, studying the association between brain structure and cognitive function may provide additional insights into brain maturation and aging, as well as the physiological causes of cognitive impairment.
In the present study, researchers developed novel gCNNs, a specialized deep learning model that extracts distinct morphological features such as cortical thickness and subcortical structure from converted brain MRIs to predict the fluid intelligence.
“We reduce the brain to its surfaces only, which means we are able to capture information not only about folding patterns, but also about curvature and relationships between surfaces across tens of thousands of nodes, including the outer cortical surface, the inner cortical surface, and the surface of seven subcortical brain structures,” Bandt said.
Using their new gCNNs, the scientists extracted morphological information of cortical ribbons and subcortical structures from two large MRI datasets involving patients of different age groups.
Using this approach, the researchers were able to demonstrate that their model significantly outperformed other similar state-of-the-art methods and that using a combination of cortical and subcortical structures provided the most accurate predictions.
Additionally, in both datasets, they found that structural features of the amygdala, hippocampus, and nucleus accumbens (NAc), as well as the temporal, parietal, and cingulate cortex, drove the prediction of the fluid intelligence.
“Previous work investigating the neuroanatomical substrate of fluid intelligence has identified associations between extensive cortical areas, but relatively few relationships have been reported with subcortical structures.
“Our study added to these studies by identifying the involvement of the bilateral NAc, the hippocampus in the prediction of fluid intelligence, which has been linked to aspects of cognitive science such as reward processing in judgment and decision-making as well as emotion regulation,” said Yunan Wu, Ph.D., a graduate student in the McCormick School of Engineering’s Department of Electrical and Computer Engineering and lead author of the study.
According to the authors, their surface-based gCNNs offer the potential to map identified relationships between neurocognition and brain anatomy for multiple research purposes. The method also requires less training and computational time, making it more efficient to apply to other full datasets.
For example, another recent study led by Bandt used gCNN analysis for aging and dementia, finding that the rate of brain aging differs between healthy individuals and patients with dementia.
“We are now looking to see if similar cognitive measures can be predicted using this method, as has been done in our work here on fluid intelligence, but can we also predict the onset of dementia and potentially prevent it? or delay it before it starts,” Bandt said.
About this deep learning and cognition research news
Author: Melissa Rohman
Source: Northwestern University
Contact: Melissa Rohman – Northwestern University
Image: Image is in public domain
See also
Original research: Free access.
“A multicohort geometric deep learning study of age-dependent cortical and subcortical morphological interactions for the prediction of fluid intelligence” by Yunan Wu et al. Scientific reports
Summary
A multicohort geometric deep learning study of age-dependent cortical and subcortical morphological interactions for the prediction of fluid intelligence
The relationship between human brain structure and cognitive function is complex, and how this relationship differs between childhood and adulthood is poorly understood.
A strong hypothesis suggests that the cognitive function of Fluid Intelligence (Gf) depends on the prefrontal cortex and the parietal cortex.
In this work, we developed a novel graphical convolutional neural network (gCNN) for localized anatomical shape analysis and Gf prediction. Morphological information of cortical ribbons and subcortical structures was extracted from T1-weighted MRI scans in two independent cohorts, the Adolescent Brain Cognitive Development Study (ABCD; age: 9.93 ± 0.62 years) d children and the Human Connectome Project (HCP; age: 28.81 ± 3.70 years).
The prediction combining the cortical and subcortical surfaces yielded the highest accuracy of Gf for the ABCD (R=0.314) and HCP (R=0.454) datasets, outperforming the peak prediction of Gf from any other measurement brain in the literature.
In both datasets, the morphology of the amygdala, hippocampus, and nucleus accumbens, as well as the temporal, parietal, and cingulate cortex consistently led to the prediction of Gf, suggesting significant reframing of the relationship. between brain morphology and Gf to include systems involved in reward/aversion processing, judgment and decision-making, motivation and emotion.