Efficient Computation of Functional Brain Networks: toward Real-Time Functional Connectivity

Abstract

Functional Connectivity has demonstrated to be a key concept for unraveling how the brain balances functional segregation and integration properties while processing information. This work presents a set of open-source tools that significantly increase computational efficiency of some well-known connectivity indices and Graph-Theory measures. PLV, PLI, ImC, and wPLI as Phase Synchronization measures, Mutual Information as an information theory based measure, and Generalized Synchronization indices are computed much more efficiently than prior open-source available implementations. Furthermore, network theory related measures like Strength, Shortest Path Length, Clustering Coefficient, and Betweenness Centrality are also implemented showing computational times up to thousands of times faster than most well-known implementations. Altogether, this work significantly expands what can be computed in feasible times, even enabling whole-head real-time network analysis of brain function.

Functional Connectivity Disruption in Subjective Cognitive Decline and Mild Cognitive Impairment: A Common Pattern of Alterations

“…We checked the significance of each individual link for each group separately to ensure results interpretability using the analytical method depicted in (Wilkie, 1983 ; García-Prieto et al, 2017 ). All the links were significant after FDR correction ( Q = 0.05) in the three diagnostic groups (all p < 0.028 for HC, all p < 0.033 for SCD and all p < 0.033 for MCI).…”

The blessing of Dimensionality: Feature Selection outperforms functional connectivity-based feature transformation to classify ADHD subjects from EEG patterns of phase synchronisation

“…Thus, rather that emphasising the absolute values of the accuracies obtained, we stress that they are the changes in this index (i.e., its relative values) after applying different approaches to select the segments and reduce the dimensionality of the feature vector, which represent most interesting outcome of our paper. Furthermore, by sharing all our data and making the code for connectivity analysis publicly available [ 42 , 69 ], in line with recent efforts from our own research [ 12 ], we hope that other labs can apply the proposed classification model as build from our EEGs to their own data. This would be the best check for the validity of the proposed approach by estimating the accuracy of the model in external test sets, or alternatively would allow refining the model by enlarging the sample size.…”

Warped phase coherence: An empirical synchronization measure combining phase and amplitude information

“…Let us consider the angle differences for the initial and reshuffled signals as per Eqs. (8) and (10), and p(∆θ ′ ) both tend towards Dirac delta functions in |c| → ∞, they do so along different trajectories: their respective distributions determine the form of w(c). Namely, in the case of identical amplitude but unrelated phases, as expected one has w(0) = 0, above which the measured coherence increases until w(1.1) = 0.16 then decreases again.…”

“…However, they may be poorly sensitive in situations of partial synchronization where amplitude fluctuations remain largely uncorrelated and may require longer time-series for reliable estimation. 7,8 Accordingly, the phase locking value and its variants are often the methods of choice for the design of connectivity-based brain-computer and brain-machine interfaces, also in virtue of their rapid convergence and robustness to non-stationarity. 9,10 A measure primarily indexing phase synchronization but capable of including a variable amount of amplitude information could have considerable practical utility.…”