Network Analysis of Time Series: Novel Approaches to Network Neuroscience DOI Creative Commons
Thomas F. Varley, Olaf Sporns

Frontiers in Neuroscience, Journal Year: 2022, Volume and Issue: 15

Published: Feb. 11, 2022

In the last two decades, there has been an explosion of interest in modeling brain as a network, where nodes correspond variously to regions or neurons, and edges structural statistical dependencies between them. This kind network construction, which preserves spatial, structural, information while collapsing across time, become broadly known “network neuroscience.” this work, we provide alternative application science neural data: network-based analysis non-linear time series review applications these methods data. Instead preserving spatial does reverse: it collapses information, instead temporally extended dynamics, typically corresponding evolution through some phase/state-space. allows researchers infer a, possibly low-dimensional, “intrinsic manifold” from empirical We will discuss three constructing networks nonlinear series, how interpret them context recurrence networks, visibility ordinal partition networks. By capturing continuous, dynamics form discrete show techniques science, theory can extract meaningful distinct what is normally accessible standard neuroscience approaches.

Language: Английский

Mixtures of large-scale dynamic functional brain network modes DOI Creative Commons
Chetan Gohil,

Evan Roberts,

Ryan C. Timms

et al.

NeuroImage, Journal Year: 2022, Volume and Issue: 263, P. 119595 - 119595

Published: Aug. 27, 2022

Accurate temporal modelling of functional brain networks is essential in the quest for understanding how such facilitate cognition. Researchers are beginning to adopt time-varying analyses electrophysiological data that capture highly dynamic processes on order milliseconds. Typically, these approaches, as clustering connectivity profiles and Hidden Markov Modelling (HMM), assume mutual exclusivity over time. Whilst a powerful constraint, this assumption may be compromising ability approaches describe effectively. Here, we propose new generative model linear mixture spatially distributed statistical "modes". The evolution governed by recurrent neural network, which enables generate with rich structure. We use Bayesian framework known amortised variational inference learn parameters from observed data. call approach DyNeMo (for Dynamic Network Modes), show using simulations it outperforms HMM when violated. In resting-state MEG, reveals modes activate fast time scales 100–150 ms, similar state lifetimes found an HMM. task MEG data, finds plausible, task-dependent evoked responses without any knowledge timings. Overall, provides decompositions approximate remapping HMM's while showing improvements overall explanatory power. However, magnitude suggests can reasonable practice. Nonetheless, flexible implementing assessing future developments.

Language: Английский

Citations

27
Edge-centric analysis of time-varying functional brain networks with applications in autism spectrum disorder DOI Creative Commons
Farnaz Zamani Esfahlani, Lisa Byrge, Jacob Tanner

et al.

NeuroImage, Journal Year: 2022, Volume and Issue: 263, P. 119591 - 119591

Published: Aug. 27, 2022

The interaction between brain regions changes over time, which can be characterized using time-varying functional connectivity (tvFC). common approach to estimate tvFC uses sliding windows and offers limited temporal resolution. An alternative method is use the recently proposed edge-centric approach, enables tracking of moment-to-moment in co-fluctuation patterns pairs regions. Here, we first examined dynamic features edge time series compared them those window (sw-tvFC). Then, used compare subjects with autism spectrum disorder (ASD) healthy controls (CN). Our results indicate that relative sw-tvFC, captured rapid bursty network-level fluctuations synchronize across during movie-watching. from second part study suggested magnitude peak amplitude collective co-fluctuations (estimated as root sum square (RSS) series) similar CN ASD. However, trough-to-trough duration RSS signal greater ASD, CN. Furthermore, an edge-wise comparison high-amplitude showed within-network edges exhibited findings suggest by provide details about disruption dynamics could potentially developing new biomarkers mental disorders.

Language: Английский

Citations

27
Edge-centric analysis of stroke patients: An alternative approach for biomarkers of lesion recovery DOI Creative Commons
Sebastián Idesis, Joshua Faskowitz, Richard F. Betzel

et al.

NeuroImage Clinical, Journal Year: 2022, Volume and Issue: 35, P. 103055 - 103055

Published: Jan. 1, 2022

Most neuroimaging studies of post-stroke recovery rely on analyses derived from standard node-centric functional connectivity to map the distributed effects in stroke patients. Here, given importance nonlocal and diffuse damage, we use an edge-centric approach order provide alternative description this disorder. These techniques allow for rendering metrics such as normalized entropy, which describes diversity edge communities at each node. Moreover, enables identification high amplitude co-fluctuations fMRI time series. We found that entropy is associated with lesion severity continually increases across patients' recovery. Furthermore, not only relate but are also level The current study first application a clinical population longitudinal dataset demonstrates how different perspective data analysis can further characterize topographic modulations brain dynamics.

Language: Английский

Citations

26
Dirac synchronization is rhythmic and explosive DOI Creative Commons

Lucille Calmon,

Juan G. Restrepo, Joaquı́n J. Torres

et al.

Communications Physics, Journal Year: 2022, Volume and Issue: 5(1)

Published: Oct. 17, 2022

Topological signals defined on nodes, links and higher dimensional simplices define the dynamical state of a network or simplicial complex. As such, topological are attracting increasing attention in theory, systems, signal processing machine learning. nodes typically studied dynamics, while much less explored. Here we investigate Dirac synchronization, describing locally coupled network, treated using operator. The dynamics is affected by phase lag depending nearby vice versa. We show that synchronization fully connected explosive with hysteresis loop characterized discontinuous forward transition continuous backward transition. analytical investigation diagram provides theoretical understanding this synchronization. model also displays an exotic coherent synchronized phase, called rhythmic non-stationary order parameters which can shed light mechanisms for emergence brain rhythms.

Language: Английский

Citations

26
Network Analysis of Time Series: Novel Approaches to Network Neuroscience DOI Creative Commons
Thomas F. Varley, Olaf Sporns

Frontiers in Neuroscience, Journal Year: 2022, Volume and Issue: 15

Published: Feb. 11, 2022

In the last two decades, there has been an explosion of interest in modeling brain as a network, where nodes correspond variously to regions or neurons, and edges structural statistical dependencies between them. This kind network construction, which preserves spatial, structural, information while collapsing across time, become broadly known “network neuroscience.” this work, we provide alternative application science neural data: network-based analysis non-linear time series review applications these methods data. Instead preserving spatial does reverse: it collapses information, instead temporally extended dynamics, typically corresponding evolution through some phase/state-space. allows researchers infer a, possibly low-dimensional, “intrinsic manifold” from empirical We will discuss three constructing networks nonlinear series, how interpret them context recurrence networks, visibility ordinal partition networks. By capturing continuous, dynamics form discrete show techniques science, theory can extract meaningful distinct what is normally accessible standard neuroscience approaches.

Language: Английский

Citations

24