Network structure of cerebral cortex shapes functional connectivity on multiple time scales

Abstract

Neuronal dynamics unfolding within the cerebral cortex exhibit complex spatial and temporal patterns even in the absence of external input. Here we use a computational approach in an attempt to relate these features of spontaneous cortical dynamics to the underlying anatomical connectivity. Simulating nonlinear neuronal dynamics on a network that captures the large-scale interregional connections of macaque neocortex, and applying information theoretic measures to identify functional networks, we find structure-function relations at multiple temporal scales. Functional networks recovered from long windows of neural activity (minutes) largely overlap with the underlying structural network. As a result, hubs in these long-run functional networks correspond to structural hubs. In contrast, significant fluctuations in functional topology are observed across the sequence of networks recovered from consecutive shorter (seconds) time windows. The functional centrality of individual nodes varies across time as interregional couplings shift. Furthermore, the transient couplings between brain regions are coordinated in a manner that reveals the existence of two anticorrelated clusters. These clusters are linked by prefrontal and parietal regions that are hub nodes in the underlying structural network. At an even faster time scale (hundreds of milliseconds) we detect individual episodes of interregional phase-locking and find that slow variations in the statistics of these transient episodes, contingent on the underlying anatomical structure, produce the transfer entropy functional connectivity and simulated blood oxygenation level-dependent correlation patterns observed on slower time scales.functional MRI ͉ graph theory ͉ neuroanatomy ͉ synchrony T he anatomical connections between regions of the cerebral cortex form a structural network upon which neural activity unfolds. Cortical regions dynamically couple to one another forming functional networks associated with perception, cognition, and action (1-4), as well as during spontaneous activity in the default or resting state (5-12). Functional networks extracted from higher-frequency dynamics (Ϸ10 Hz) undergo rapid reconfiguration, e.g., in perceptual binding (13) or sensorimotor coordination (14). Functional networks extracted from lowerfrequency (Ͻ0.1 Hz) spontaneous cortical dynamics are organized into anticorrelated clusters (8, 10), and the transient activation of these clusters has been linked to attentional processes (10, 12). Recent analyses suggest that structural and functional brain networks share ''small-world'' topologies and hub nodes (15)(16)(17)(18)(19)(20) and that structural and functional clusters may closely correspond (21, 22). Nevertheless, it remains unclear how functional networks at different time scales relate to one another and to their common structural substrate.Here we studied mappings between structural and functional networks using a computational approach. A structural network of segregated regions and interregional pathways was obtained...

Measuring temporal, spectral and spatial changes in electrophysiological brain network connectivity

“…(Friston, 1997(Friston, , 2000), computational (e.g. (Breakspear et al, 2003, Honey et al, 2007, Ghosh et al, 2008, Deco et al, 2009 (Brookes et al, 2012b). Indeed this finding has been mirrored by other MEG studies (Hipp et al, 2012), and is in general agreement with findings from simultaneous electroencephalography (EEG) / fMRI.…”

“…(Friston, 1997(Friston, , 2000), computational (e.g. (Breakspear et al, 2003, Honey et al, 2007, Ghosh et al, 2008, Deco et al, 2009) and empirical (e.g. (Breakspear et al, 2004, Bassett et al, 2006) evidence suggests that complex and highly temporally variable neuronal dynamics underlie the coupling observed between spatially separate brain regions.…”

Structurally-informed Bayesian functional connectivity analysis

“…The idea is to infer structural connectivity from DWI data and use it as an additional constraint in our Bayesian model. The validity of this approach is supported by several recent experimental studies which found a substantial overlap between structural and functional networks both inside specific cortical areas (Koch et al, 2002) and on a whole brain scale (Hagmann et al, 2008;Honey et al, 2009Honey et al, , 2007Damoiseaux and Greicius, 2009;Greicius et al, 2009;Cabral et al, 2012). Related approaches have been used before in the context of functional and effective connectivity analysis (Stephan et al, 2009;Deligianni et al, 2011;Ng et al, 2012).…”

MR connectomics: Principles and challenges

“…Their anatomical locations, as well as the extent to which the matrix covers the surface of macaque cortex are shown in the panels at the top of the figure (lateral and medial views, respectively). Blue and green represents cortices belonging to the two major clusters identified by spectral methods (Honey et al, 2007). (B) Proportions of the total DSI fiber mass that are coinciding with "known present," "unknown," and "known absent" Cocomac pathways.…”

“…Towards this objective, a number of computational studies have attempted to combine interregional connectivity data with intra-regional circuit and cellular properties to produce large-scale models of cortical (Ghosh et al, 2008;Honey et al, 2007Honey et al, , 2009Knock et al, 2009) and corticothalamic function (Izhikevich and Edelman, 2008). The value of modeling studies is twofold.…”