Our previous study on the nature of changes in spatial and temporal features in rsfMRI across the lifespan show that cognitive maturation leads to network stabilization and thereby decreased cortical employment during functional activation. We believe that these observed changes in network topology and functional connectivity are driven by changes in the spectral composition of rsfMRI in a logarithmic scale. Understandably, there are smaller time-scale dynamics that manifest in rsfMRI data. This project seeks to explore how the time-varying functional connectivity between resting-state networks modulates across the lifespan. We use a novel data-driven approach, conditional inference decision trees with bagging to characterize the relationship between resting-state dynamics measures such as transitions between states, the frequency of occurrence of states, and age. We found that these dynamic features in rs-fMRI data exhibit a midlife shift. In the ages between 40-46, sudden changes in the number of transitions between states has been observed. Additionally, the frequency of occurrence of the most commonly occurring state decreases till the age of 39 and then stabilizes.
In neuropsychological applications such as autism, studies have conventionally examined participants with a range of ages in the same group to compare against age-matched healthy controls. The normalizing effect of averaging across age could possibly result in loss of statistical strength in such comparisons. For this, I aim to use the knowledge of typical variability as a basis to explore variability patterns associated with age in autism to help identify the effect of autism on the variability in functional activation over age.
Each person is different. This difference is scientifically termed inter-subject variability. In terms of functional activation patterns, these differences have been verified at network as well as local or individual spatial cluster levels. Studies have also shown that these differences in functional activation patterns are related to behavioral measures such as executive functioning, cognitive flexibility, etc.. Does this mean that a sample of participants, composed of even only healthy adults, can be grouped such that similarities in cognitive abilities are related to the functional patterns observed? This is the primary aim of this project. To identify neural patterns that can distinctly identify cognitive sub-groups. We are employing a data-driven approach to identifying sub-groups in resting state features that exhibit inter-subject variability and see if these sub-groups are related to executive functioning groups. This information will in fact provide the field of autism sub-groups using functional neuroimaging with significant bases to explore sub-groups in clinical disorders. Accordingly, we are aiming to extend the data-driven analysis of sub-groups in resting-state features that exhibit inter-subject variability in autism and see if these sub-groups can characterize symptomatic and behavioral sub-groups.