Mean-field modeling of brain-scale dynamics for the evaluation of EEG source-space networks

Abstract

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 good spatio-temporal resolution, while reducing mixing and volume conduction effects. However, the method is still immature and several methodological issues should be carefully accounted for to avoid pitfalls. Therefore, optimizing the EEG source connectivity pipelines is required, which involves the evaluation of several parameters. One key issue to address those evaluation aspects is the availability of a ‘ground truth’. 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 (epileptiform activity) 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 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 show that a high electrode density (at least 64 channels) is needed 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). The COALIA model and the simulations used in this paper are freely available and made accessible for the community. We believe that this model-based approach will help researchers to address some current and future cognitive and clinical neuroscience questions, and ultimately transform EEG brain network imaging into a mature technology.

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

“…In this study, the wMNE/PLV combination exhibits the best performance leading to the highest spatial similarity value between reconstructed and reference networks among all inverse model/functional connectivity combinations tested. Interestingly, these results are in line with previous comparative studies showing consistency and robust results for wMNE/PLV combination in the context of EEG source space connectivity using real data from picture naming task (Hassan et al, 2014) and simulated data from epileptogenic-modeled networks (Allouch et al, 2020;Hassan et al, 2017). The strength of PLV results may reflect the potential mechanisms of zero-lag synchronization of neural activity already discussed in previous works (Gollo et al, 2014;Roelfsema et al, 1997).…”

“…Finally, in this work, we set the number of EEG channels (257 channels) since it established accurate source localization results relative to lower sensor densities (Allouch et al, 2020; Song et al, 2015). It can be however interesting to follow the work of such studies and examine the effect of sensor spatial resolution by decreasing successively the number of electrodes from high-density (257 channels) to low-density (19 channels) EEG signals.…”

“…Technically, EEG simulated signals were computed via the forward model as previously shown by recent works (Allouch et al, 2020; Bensaid et al, 2019). In this paper, we aim to investigate to what extent we can effectively track time-varying brain networks that dominate neuronal activity using HD-EEG scalp signals, specifically on a short timescale that is commensurate with cognitive tasks.…”

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

“…Another critical influencing factor in the EEG source connectivity pipeline is the algorithm chosen to solve the inverse problem, which is ill-posed due to its non-uniqueness and the instability of its solution (see (Grech et al 2008) for a review). Several studies quantifying the performance of different inverse methods, in simulated and experimental EEG/MEG data, concluded that the choice of the inverse method significantly influences source estimation results (Anzolin et al 2019; Mahjoory et al 2017; Hedrich et al 2017; Bradley et al 2016; Grova et al 2006; Halder et al 2019; Tait et al 2021; Allouch et al 2022). However, no consistent conclusions have been made regarding one method that would stand apart from the others in terms of performance, which can also be related to the analyzed conditions.…”

“…A wide range of measures are used in the field, and each differs in the aspect of the data that is being investigated (amplitude-vs phase-based measures / directional vs non-directional connectivity, prone/robust to source leakage), (see (Friston 2011; Pereda, Quiroga, and Bhattacharya 2005; Cao et al 2022) for a review), resulting in a significant variability of performance and interpretations (Colclough et al 2016; H. E. Wang et al 2014; Wendling et al 2009; Hassan, Merlet, et al 2017; Allouch et al 2022).…”