Effect of channel density, inverse solutions and connectivity measures on EEG resting-state networks reconstruction: A simulation study.

Along with the study of brain activity evoked by external stimuli, the past two decades witnessed an increased interest in characterizing the spontaneous brain activity occurring during resting conditions. The identification of connectivity patterns in this so-called "resting-state" has been the subject of a great number of electrophysiology-based studies, using the Electro/Magneto-Encephalography (EEG/MEG) source connectivity method. However, no consensus has been reached yet regarding a unified (if possible) analysis pipeline, and several involved parameters and methods require cautious tuning. This is particularly challenging when different analytical choices induce significant discrepancies in results and drawn conclusions, thereby hindering the reproducibility of neuroimaging research. Hence, our objective in this study was to shed light on the effect of analytical variability on outcome consistency by evaluating the implications of parameters involved in the EEG source connectivity analysis on the accuracy of resting-state networks (RSNs) reconstruction. We simulated, using neural mass models, EEG data corresponding to two RSNs, namely the default mode network (DMN) and dorsal attentional network (DAN). We investigated the impact of five channel densities (19, 32, 64, 128, 256), three inverse solutions (weighted minimum norm estimate (wMNE), exact low-resolution brain electromagnetic tomography (eLORETA), and linearly constrained minimum variance (LCMV) beamforming) and four functional connectivity measures (phase-locking value (PLV), phase-lag index (PLI), and amplitude envelope correlation (AEC) with and without source leakage correction), on the correspondence between reconstructed and reference networks. We showed that, with different analytical choices related to the number of electrodes, source reconstruction algorithm, and functional connectivity measure, high variability is present in the results. More specifically, our results show that a higher number of EEG channels significantly increased the accuracy of the reconstructed networks. Additionally, our results showed significant variability in the performance of the tested inverse solutions and connectivity measures. Such methodological variability and absence of analysis standardization represent a critical issue for neuroimaging studies that should be prioritized. We believe that this work could be useful for the field of electrophysiology connectomics, by increasing awareness regarding the challenge of variability in methodological approaches and its implications on reported results.

Spatio-temporal dynamics of large-scale electrophysiological networks during cognitive action control in healthy controls and Parkinson's disease patients.

Among the cognitive symptoms that are associated with Parkinson's disease (PD), alterations in cognitive action control (CAC) are commonly reported in patients. CAC enables the suppression of an automatic action, in favor of a goal-directed one. The implementation of CAC is time-resolved and arguably associated with dynamic changes in functional brain networks. However, the electrophysiological functional networks involved, their dynamic changes, and how these changes are affected by PD, still remain unknown. In this study, to address this gap of knowledge, 10 PD patients and 10 healthy controls (HC) underwent a Simon task while high-density electroencephalography (HD-EEG) was recorded. Source-level dynamic connectivity matrices were estimated using the phase-locking value in the beta (12-25 Hz) and gamma (30-45 Hz) frequency bands. Temporal independent component analyses were used as a dimension reduction tool to isolate the task-related brain network states. Typical microstate metrics were quantified to investigate the presence of these states at the subject-level. Our results first confirmed that PD patients experienced difficulties in inhibiting automatic responses during the task. At the group-level, we found three functional network states in the beta band that involved fronto-temporal, temporo-cingulate and fronto-frontal connections with typical CAC-related prefrontal and cingulate nodes (e.g., inferior frontal cortex). The presence of these networks did not differ between PD patients and HC when analyzing microstates metrics, and no robust correlations with behavior were found. In the gamma band, five networks were found, including one fronto-temporal network that was identical to the one found in the beta band. These networks also included CAC-related nodes previously identified in different neuroimaging modalities. Similarly to the beta networks, no subject-level differences were found between PD patients and HC. Interestingly, in both frequency bands, the dominant network at the subject-level was never the one that was the most durably modulated by the task. Altogether, this study identified the dynamic functional brain networks observed during CAC, but did not highlight PD-related changes in these networks that might explain behavioral changes. Although other new methods might be needed to investigate the presence of task-related networks at the subject-level, this study still highlights that task-based dynamic functional connectivity is a promising approach in understanding the cognitive dysfunctions observed in PD and beyond.

Mean-Field Modeling of Brain-Scale Dynamics for the Evaluation of EEG Source-Space Networks.

Understanding the dynamics of brain-scale functional networks at rest and during cognitive tasks is the subject of intense research efforts to unveil fundamental principles of brain functions. To estimate these large-scale brain networks, the emergent method called "electroencephalography (EEG) source connectivity" has generated increasing interest in the network neuroscience community, due to its ability to identify cortical brain networks with satisfactory spatio-temporal resolution, while reducing mixing and volume conduction effects. However, no consensus has been reached yet regarding a unified EEG source connectivity pipeline, and several methodological issues have to be carefully accounted to avoid pitfalls. Thus, a validation toolbox that provides flexible "ground truth" models is needed for an objective methods/parameters evaluation and, thereby an optimization of the EEG source connectivity pipeline. In this paper, we show how a recently developed large-scale model of brain-scale activity, named COALIA, can provide to some extent such ground truth by providing realistic simulations of source-level and scalp-level activity. Using a bottom-up approach, the model bridges cortical micro-circuitry and large-scale network dynamics. Here, we provide an example of the potential use of COALIA to analyze, in the context of epileptiform activity, the effect of three key factors involved in the "EEG source connectivity" pipeline: (i) EEG sensors density, (ii) algorithm used to solve the inverse problem, and (iii) functional connectivity measure. Results showed that a high electrode density (at least 64 channels) is required to accurately estimate cortical networks. Regarding the inverse solution/connectivity measure combination, the best performance at high electrode density was obtained using the weighted minimum norm estimate (wMNE) combined with the weighted phase lag index (wPLI). Although those results are specific to the considered aforementioned context (epileptiform activity), we believe that this model-based approach can be successfully applied to other experimental questions/contexts. We aim at presenting a proof-of-concept of the interest of COALIA in the network neuroscience field, and its potential use in optimizing the EEG source-space network estimation pipeline.

Dynamics of task-related electrophysiological networks: a benchmarking study.

Motor, sensory and cognitive functions rely on dynamic reshaping of functional brain networks. Tracking these rapid changes is crucial to understand information processing in the brain, but challenging due to the great variety of dimensionality reduction methods used at the network-level and the limited evaluation studies. Using Magnetoencephalography (MEG) combined with Source Separation (SS) methods, we present an integrated framework to track fast dynamics of electrophysiological brain networks. We evaluate nine SS methods applied to three independent MEG databases (N=95) during motor and memory tasks. We report differences between these methods at the group and subject level. We seek to help researchers in choosing objectively the appropriate SS method when tracking fast reconfiguration of functional brain networks, due to its enormous benefits in cognitive and clinical neuroscience.

Identification of Interictal Epileptic Networks from Dense-EEG.

Epilepsy is a network disease. The epileptic network usually involves spatially distributed brain regions. In this context, noninvasive M/EEG source connectivity is an emerging technique to identify functional brain networks at cortical level from noninvasive recordings. In this paper, we analyze the effect of the two key factors involved in EEG source connectivity processing: (i) the algorithm used in the solution of the EEG inverse problem and (ii) the method used in the estimation of the functional connectivity. We evaluate four inverse solutions algorithms (dSPM, wMNE, sLORETA and cMEM) and four connectivity measures (r 2, h 2, PLV, and MI) on data simulated from a combined biophysical/physiological model to generate realistic interictal epileptic spikes reflected in scalp EEG. We use a new network-based similarity index to compare between the network identified by each of the inverse/connectivity combination and the original network generated in the model. The method will be also applied on real data recorded from one epileptic patient who underwent a full presurgical evaluation for drug-resistant focal epilepsy. In simulated data, results revealed that the selection of the inverse/connectivity combination has a significant impact on the identified networks. Results suggested that nonlinear methods (nonlinear correlation coefficient, phase synchronization and mutual information) for measuring the connectivity are more efficient than the linear one (the cross correlation coefficient). The wMNE inverse solution showed higher performance than dSPM, cMEM and sLORETA. In real data, the combination (wMNE/PLV) led to a very good matching between the interictal epileptic network identified from noninvasive EEG recordings and the network obtained from connectivity analysis of intracerebral EEG recordings. These results suggest that source connectivity method, when appropriately configured, is able to extract highly relevant diagnostic information about networks involved in interictal epileptic spikes from non-invasive dense-EEG data.

Dynamic rewiring of electrophysiological brain networks during learning.

Human learning is an active and complex process. However, the brain mechanisms underlying human skill learning and the effect of learning on the communication between brain regions, at different frequency bands, are still largely unknown. Here, we tracked changes in large-scale electrophysiological networks over a 6-week training period during which participants practiced a series of motor sequences during 30 home training sessions. Our findings showed that brain networks become more flexible with learning in all the frequency bands from theta to gamma ranges. We found consistent increase of flexibility in the prefrontal and limbic areas in the theta and alpha band, and over somatomotor and visual areas in the alpha band. Specific to the beta rhythm, we revealed that higher flexibility of prefrontal regions during the early stage of learning strongly correlated with better performance measured during home training sessions. Our findings provide novel evidence that prolonged motor skill practice results in higher, frequency-specific, temporal variability in brain network structure.

The development of an automated machine learning pipeline for the detection of Alzheimer's Disease.

Although Alzheimer's disease is the most prevalent form of dementia, there are no treatments capable of slowing disease progression. A lack of reliable disease endpoints and/or biomarkers contributes in part to the absence of effective therapies. Using machine learning to analyze EEG offers a possible solution to overcome many of the limitations of current diagnostic modalities. Here we develop a logistic regression model with an accuracy of 81% that addresses many of the shortcomings of previous works. To our knowledge, no other study has been able to solve the following problems simultaneously: (1) a lack of automation and unbiased removal of artifacts, (2) a dependence on a high level of expertise in data pre-processing and ML for non-automated processes, (3) the need for very large sample sizes and accurate EEG source localization using high density systems, (4) and a reliance on black box ML approaches such as deep neural nets with unexplainable feature selection. This study presents a proof-of-concept for an automated and scalable technology that could potentially be used to diagnose AD in clinical settings as an adjunct to conventional neuropsychological testing, thus enhancing efficiency, reproducibility, and practicality of AD diagnosis.

Assessing HD-EEG functional connectivity states using a human brain computational model.

Objective.Electro/Magnetoencephalography (EEG/MEG) source-space network analysis is increasingly recognized as a powerful tool for tracking fast electrophysiological brain dynamics. However, an objective and quantitative evaluation of pipeline steps is challenging due to the lack of realistic 'controlled' data. Here, our aim is two-folded: (a) provide a quantitative assessment of the advantages and limitations of the analyzed techniques and (b) introduce (and share) a complete framework that can be used to optimize the entire pipeline of EEG/MEG source connectivity.Approach.We used a human brain computational model containing both physiologically based cellular GABAergic and Glutamatergic circuits coupled through Diffusion Tensor Imaging, to generate high-density EEG recordings. We designed a scenario of successive gamma-band oscillations in distinct cortical areas to emulate a virtual picture-naming task. We identified fast time-varying network states and quantified the performance of the key steps involved in the pipeline: (a) inverse models to reconstruct cortical-level sources, (b) functional connectivity measures to compute statistical interdependency between regional signals, and (c) dimensionality reduction methods to derive dominant brain network states (BNS).Main results.Using a systematic evaluation of the different decomposition techniques, results show significant variability among tested algorithms in terms of spatial and temporal accuracy. We outlined the spatial precision, the temporal sensitivity, and the global accuracy of the extracted BNS relative to each method. Our findings suggest a good performance of weighted minimum norm estimate/ Phase Locking Value combination to elucidate the appropriate functional networks and ICA techniques to derive relevant dynamic BNS.Significance.We suggest using such brain models to go further in the evaluation of the different steps and parameters involved in the EEG/MEG source-space network analysis. This can reduce the empirical selection of inverse model, connectivity measure, and dimensionality reduction method as some of the methods can have a considerable impact on the results and interpretation.

Methods Used to Estimate EEG Source-Space Networks: A Comparative Simulation-Based Study.

Along with the study of the brain activity evoked by external stimuli, an important advance in current neuroscience involves understanding the spontaneous brain activity that occurs during resting conditions. Interestingly, the identification of the connectivity patterns in "resting-state" has been the subject of a great number of electrophysiology-based studies. In this context, the Electroencephalography (EEG) source connectivity method enables estimating resting-state cortical networks from scalp-EEG recordings. However, there is still no consensus over a unified pipeline adapted in all cases (e.g., type of task, a priori on studied networks) and numerous methodological questions remain unanswered. In order to address this problem, we simulated, using neural mass models, EEG data corresponding to the default mode network (DMN), the most widely studied resting-state network, and tested the effect of different channel densities, two inverse solutions and two functional connectivity measures on the correspondence between the reconstructed networks and the reference networks. Results showed that increasing the number of electrodes enhances the accuracy of the network reconstruction, and that eLORETA/PLV led to better accuracy than other inverse solution/connectivity measure combinations in terms of the correlation between reconstructed and reference connectivity matrices. This work has a wide range of implications in the field of electrophysiology connectomics, and is a step towards a convergence and standardization of approaches in this emerging field.

An electroencephalography connectome predictive model of major depressive disorder severity.

Emerging evidence showed that major depressive disorder (MDD) is associated with disruptions of brain structural and functional networks, rather than impairment of isolated brain region. Thus, connectome-based models capable of predicting the depression severity at the individual level can be clinically useful. Here, we applied a machine-learning approach to predict the severity of depression using resting-state networks derived from source-reconstructed Electroencephalography (EEG) signals. Using regression models and three independent EEG datasets (N = 328), we tested whether resting state functional connectivity could predict individual depression score. On the first dataset, results showed that individuals scores could be reasonably predicted (r = 0.6, p = 4 × 10-18) using intrinsic functional connectivity in the EEG alpha band (8-13 Hz). In particular, the brain regions which contributed the most to the predictive network belong to the default mode network. We further tested the predictive potential of the established model by conducting two external validations on (N1 = 53, N2 = 154). Results showed statistically significant correlations between the predicted and the measured depression scale scores (r1 = 0.52, r2 = 0.44, p < 0.001). These findings lay the foundation for developing a generalizable and scientifically interpretable EEG network-based markers that can ultimately support clinicians in a biologically-based characterization of MDD.

Brain Network Dynamics Correlate with Personality Traits.

Identifying the neural substrates underlying the personality traits is a topic of great interest. On the other hand, it is now established that the brain is a dynamic networked system that can be studied by using functional connectivity techniques. However, much of the current understanding of personality-related differences in functional connectivity has been obtained through the stationary analysis, which does not capture the complex dynamical properties of brain networks. In this study, we aimed at evaluating the feasibility of using dynamic network measures to predict personality traits. Using the electro-encephalography (EEG)/magneto-encephalography (MEG) source connectivity method combined with a sliding window approach, dynamic functional brain networks were reconstructed from two datasets: (1) resting-state EEG data acquired from 56 subjects; (2) resting-state MEG data provided from the Human Connectome Project. Then, several dynamic functional connectivity metrics were evaluated. Similar observations were obtained by the two modalities (EEG and MEG) according to the neuroticism, which showed a negative correlation with the dynamic variability of resting-state brain networks. In particular, a significant relationship between this personality trait and the dynamic variability of the temporal lobe regions was observed. Results also revealed that extraversion and openness are positively correlated with the dynamics of the brain networks. These findings highlight the importance of tracking the dynamics of functional brain networks to improve our understanding about the neural substrates of personality.

Detecting modular brain states in rest and task.

The human brain is a dynamic networked system that continually reconfigures its functional connectivity patterns over time. Thus, developing approaches able to adequately detect fast brain dynamics is critical. Of particular interest are the methods that analyze the modular structure of brain networks, that is, the presence of clusters of regions that are densely interconnected. In this paper, we propose a novel framework to identify fast modular states that dynamically fluctuate over time during rest and task. We started by demonstrating the feasibility and relevance of this framework using simulated data. Compared with other methods, our algorithm was able to identify the simulated networks with high temporal and spatial accuracies. We further applied the proposed framework on MEG data recorded during a finger movement task, identifying modular states linking somatosensory and primary motor regions. The algorithm was also performed on dense-EEG data recorded during a picture naming task, revealing the subsecond transition between several modular states that relate to visual processing, semantic processing, and language. Next, we tested our method on a dataset of resting-state dense-EEG signals recorded from 124 patients with Parkinson's disease. Results disclosed brain modular states that differentiate cognitively intact patients, patients with moderate cognitive deficits, and patients with severe cognitive deficits. Our new approach complements classical methods, offering a new way to track the brain modular states, in healthy subjects and patients, on an adequate task-specific timescale.

Functional Brain Connectivity as a New Feature for P300 Speller.

The brain is a large-scale complex network often referred to as the "connectome". Cognitive functions and information processing are mainly based on the interactions between distant brain regions. However, most of the 'feature extraction' methods used in the context of Brain Computer Interface (BCI) ignored the possible functional relationships between different signals recorded from distinct brain areas. In this paper, the functional connectivity quantified by the phase locking value (PLV) was introduced to characterize the evoked responses (ERPs) obtained in the case of target and non-targets visual stimuli. We also tested the possibility of using the functional connectivity in the context of 'P300 speller'. The proposed approach was compared to the well-known methods proposed in the state of the art of "P300 Speller", mainly the peak picking, the area, time/frequency based features, the xDAWN spatial filtering and the stepwise linear discriminant analysis (SWLDA). The electroencephalographic (EEG) signals recorded from ten subjects were analyzed offline. The results indicated that phase synchrony offers relevant information for the classification in a P300 speller. High synchronization between the brain regions was clearly observed during target trials, although no significant synchronization was detected for a non-target trial. The results showed also that phase synchrony provides higher performance than some existing methods for letter classification in a P300 speller principally when large number of trials is available. Finally, we tested the possible combination of both approaches (classical features and phase synchrony). Our findings showed an overall improvement of the performance of the P300-speller when using Peak picking, the area and frequency based features. Similar performances were obtained compared to xDAWN and SWLDA when using large number of trials.