Topological Filtering of Dynamic Functional Brain Networks Unfolds Informative Chronnectomics: A Novel Data-Driven Thresholding Scheme Based on Orthogonal Minimal Spanning Trees (OMSTs).

Stavros I Dimitriadis,Christos Salis,David E Linden,Ioannis Tarnanas

doi:10.3389/fninf.2017.00028

Abstract

The human brain is a large-scale system of functionally connected brain regions. This system can be modeled as a network, or graph, by dividing the brain into a set of regions, or “nodes,” and quantifying the strength of the connections between nodes, or “edges,” as the temporal correlation in their patterns of activity. Network analysis, a part of graph theory, provides a set of summary statistics that can be used to describe complex brain networks in a meaningful way. The large-scale organization of the brain has features of complex networks that can be quantified using network measures from graph theory. The adaptation of both bivariate (mutual information) and multivariate (Granger causality) connectivity estimators to quantify the synchronization between multichannel recordings yields a fully connected, weighted, (a)symmetric functional connectivity graph (FCG), representing the associations among all brain areas. The aforementioned procedure leads to an extremely dense network of tens up to a few hundreds of weights. Therefore, this FCG must be filtered out so that the “true” connectivity pattern can emerge. Here, we compared a large number of well-known topological thresholding techniques with the novel proposed data-driven scheme based on orthogonal minimal spanning trees (OMSTs). OMSTs filter brain connectivity networks based on the optimization between the global efficiency of the network and the cost preserving its wiring. We demonstrated the proposed method in a large EEG database (N = 101 subjects) with eyes-open (EO) and eyes-closed (EC) tasks by adopting a time-varying approach with the main goal to extract features that can totally distinguish each subject from the rest of the set. Additionally, the reliability of the proposed scheme was estimated in a second case study of fMRI resting-state activity with multiple scans. Our results demonstrated clearly that the proposed thresholding scheme outperformed a large list of thresholding schemes based on the recognition accuracy of each subject compared to the rest of the cohort (EEG). Additionally, the reliability of the network metrics based on the fMRI static networks was improved based on the proposed topological filtering scheme. Overall, the proposed algorithm could be used across neuroimaging and multimodal studies as a common computationally efficient standardized tool for a great number of neuroscientists and physicists working on numerous of projects.

Highlights

The brain is an inherently dynamic system which both restingstate networks (Hansen et al, 2015) and networks connected to executive cognition requires dynamic reconfiguration of highly evolving networks of brain areas that interact via transient complexity patterns (Bassett et al, 2011; Braun et al, 2015a)
Thresholding Scheme Alters Brain Fingerprinting Based on nNMTSGE To demonstrate how the adaptation of a thresholding scheme can alter the brain fingerprinting based on nNMTSGE, we illustrated the nNMTSGE from a single-subject in δ band at eyes-closed condition from Fz sensor (Figure 7)
We showed that a data-driven thresholding scheme based on orthogonal minimal spanning trees (OMSTs) implies a unique description of each individual brain activity based on the notion of nodal network metrics times series and namely the global efficiency

Summary

Introduction

The brain is an inherently dynamic system which both restingstate networks (Hansen et al, 2015) and networks connected to executive cognition requires dynamic reconfiguration of highly evolving networks of brain areas that interact via transient complexity patterns (Bassett et al, 2011; Braun et al, 2015a). The dynamic approach of brain connectivity is explored by adopting a time-window over experimental time and constructs quasi-static connectivity graphs from the temporal signal segments enclosed within this time-window where various network metrics can be estimated for each window leading to network metric time series (NMTS) (Dimitriadis et al, 2010a, 2012a, 2013a,b, 2015a,c; Sporns, 2011, 2014). Both static and dynamic graphs can be analyzed either as binary or weighted networks after first applying a thresholding criterion. The overrepresentation of functional connections between brain areas added difficulties to the extraction of significant topological information (Serrano et al, 2009)

Methods

Results

Discussion

Conclusion