Unsupervised characterization of dynamic functional connectivity reveals age-associated differences in temporal stability and connectivity states during rest and task

Abstract

Understanding brain functions as an outcome of underlying neuro-cognitive network mechanisms in rest and task requires accurate spatiotemporal characterization of the relevant functional brain networks. Recent endeavours of the Neuroimaging community to develop the notion of dynamic functional connectivity is a step in this direction. A key goal is to detect what are the important events in time that delimits how one functional brain network defined by known patterns of correlated brain activity transitions into a “new” network. Such characterization can also lead to more accurate conceptual realization of brain states, thereby, defined in terms of time-resolved correlations. Nonetheless, identifying the canonical temporal window over which dynamic functional connectivity is operational is currently based on an ad-hoc selection of sliding windows that can certainly lead to spurious results. Here, we introduce a data-driven unsupervised approach to characterize the high dimensional dynamic functional connectivity into dynamics of lower dimensional patterns. The whole-brain dynamic functional connectivity states bearing functional significance for task or rest can be explored through the temporal correlations, both short and long range. The present study investigates the stability of such short- and long-range temporal correlations to explore the dynamic network mechanisms across resting state, movie viewing and sensorimotor action tasks requiring varied degrees of attention. As an outcome of applying our methods to the fMRI data of a healthy ageing cohort we could quantify whole-brain temporal dynamics which indicates naturalistic movie watching task is closer to resting state than the sensorimotor task. Our analysis also revealed an overall trend of highest short range temporal network stability in the sensorimotor task, followed by naturalistic movie watching task and resting state that remains similar in both young and old adults. However, the stability of neurocognitive networks in the resting state in young adults is higher than their older counterparts. Thus, healthy ageing related differences in quantification of network stability along task and rest provides a blueprint of how our approach can be used for cohort studies of mental health and neurological disorders.

Identifying Individuals by fNIRS-Based Brain Functional Network Fingerprints

“…The mainstream identification characteristics, including face, fingerprint, and so on, are easily counterfeited, unstable in time, and involve privacy implications (Prabhakar et al, 2003;Jain et al, 2004Jain et al, , 2006. Advanced studies indicate that brain functional network (BFN) estimated by the temporal correlation between pairs of brain regions has the advantages of anti-imitation and stability in timing (Hilger et al, 2017(Hilger et al, , 2020Wang et al, 2019b;Sastry et al, 2021). More importantly, the BFN-based "fingerprint" provides a potential way of exploring the brain at the individual level.…”