Shella Keilholz (Georgia Institute of Technology and Emory University)
Dr Bruce Crosson (Emory University and Georgia State University)
Dr Eric Schumacher (Georgia Institute of Technology)
Dr Anabelle Singer (Georgia Institute of Technology and Emory University)
Dr Garth Thompson (Shanghai Tech University)
Quasi-periodic patterns of brain intrinsic activity
dominantly coordinate the functional connections in humans
Brain is a complex self-organizing biophysical system and intrinsically very active. How such intrinsic activity serves the purpose of self-organization in humans can be examined during resting-state with the functional magnetic resonance imaging (rsfMRI). As a metric of average coherent activity, the Pearson correlation between rsfMRI timeseries of brain areas, called functional connectivity (FC), can reflect aspects of self-organization. For example, based on the FC between pairs of areas, the cerebral cortex can be parcellated into a few resting-state networks (RSNs) or exhibit a few functional connectivity gradients (FCGs). Varied dynamic regimes of intrinsic coherent activity exist as well. If often and strong enough, they can give rise to the average metric of FC yet might entail aspects about self-organization not captured by FC, and their independent characterization can be insightful. Among such dynamic regimes are the spatiotemporal quasiperiodic patterns (QPPs). Each QPP is obtained by identifying and simply averaging a set of similar segments of rsfMRI scan. Whole-brain QPPs in humans are ~20s long and each involve a cycle of activation and deactivation of different areas with different timings, overall reminiscent of RSNs and FCGs, suggesting contribution to FC.
To robustly detect multiple QPPs, method improvements were implemented and applied to ~800 individuals of the Human Connectome Project dataset. Three primary QPPs were thoroughly characterized. Within these QPPs activity propagates along the functional gradients at the cerebral cortex and most subcortical regions, in a well-coordinated way, because of the consistencies and synchronies across all brain regions which reasonably accord with the consensus on the structural connections. Nuanced timing differences between regions and the closed flow of activity throughout the brain suggest drivers for these patterns. QPPs reflect neuronal activity, however, they also exhibit a principled relation with the slow variations in the respiration and heart rate and might basically be neurophysiological patterns. When three QPPs are removed from rsfMRI timeseries, FC within and particularly between RSNs remarkably reduces, illustrating their dominant contribution. Together, our results suggest a few recurring spatiotemporal patterns of intrinsic activity can dominantly coordinate the functional connections across the whole brain and serve self-organization. These intrinsic patterns possibly interact with the external tasks, affecting performance, or might provide more sensitive biomarkers in certain disorders and diseases.