Inferring multi-scale neural mechanisms with brain network modelling

Abstract

The neurophysiological processes underlying non-invasive brain activity measurements are incompletely understood. Here, we developed a connectome-based brain network model that integrates individual structural and functional data with neural population dynamics to support multi-scale neurophysiological inference. Simulated populations were linked by structural connectivity and, as a novelty, driven by electroencephalography (EEG) source activity. Simulations not only predicted subjects' individual resting-state functional magnetic resonance imaging (fMRI) time series and spatial network topologies over 20 minutes of activity, but more importantly, they also revealed precise neurophysiological mechanisms that underlie and link six empirical observations from different scales and modalities: (1) resting-state fMRI oscillations, (2) functional connectivity networks, (3) excitation-inhibition balance, (4, 5) inverse relationships between α-rhythms, spike-firing and fMRI on short and long time scales, and (6) fMRI power-law scaling. These findings underscore the potential of this new modelling framework for general inference and integration of neurophysiological knowledge to complement empirical studies.

Modeling Brain Dynamics in Brain Tumor Patients Using the Virtual Brain

“…After tuning the model parameters, the link-wise Pearson correlation between the upper triangular part of the simulated and empirical functional connectivity matrices was on average 0.33 across subjects (SD = 0.09). These magnitudes of similarity are similar to those reported in other studies that have simulated subject-specific brain dynamics at the large-scale level (see for example Deco et al, 2013; Schirner et al, 2018). The obtained correlations between simulated and empirical functional connectivity in turn correlated with the correlation coefficients between the subject’s structural and empirical functional connectivity (Pearson’s r (SC-FC emp , FC emp -FC sim ) = 0.68, 95% CI 0.45–0.82, p < 0.0001).…”

“…In this study, we simulated large-scale brain dynamics in brain tumor patients and control subjects using the Reduced Wong–Wang model (Deco et al, 2014) as implemented in a highly optimized C version of TVB’s simulation core (Schirner et al, 2018). After tuning the model parameters, the link-wise Pearson correlation between the upper triangular part of the simulated and empirical functional connectivity matrices was on average 0.33 across subjects (SD = 0.09).…”

“…1 for visual overview of the workflow). In particular, local dynamics in each of the 68 cortical brain regions were simulated using the Reduced Wong–Wang model (Deco et al, 2014), implemented as highly optimized C code that allows for efficient parameter exploration (Schirner et al, 2018). In particular, each cortical region of the Desikan–Killiany brain atlas (Desikan et al, 2006) was modeled as a local network composed of interconnected excitatory and inhibitory neural mass models coupled by excitatory (NMDA) and inhibitory (GABA) connections (we refer the interested reader to Deco et al, [2014] for a more detailed description of this model).…”

“…Diffusion MRI data processed to an SC matrix as input for TVB, and resting-state functional MRI data processed to an empirical functional connectivity (FCemp) matrix to evaluate model fit (red, positive connection weights; blue, negative connection weights). Personalized virtual brain model: Local dynamics in each of the 68 Freesurfer cortical brain regions were simulated using the Reduced Wong–Wang model (Deco et al, 2014), implemented as highly optimized C code that allows for efficient parameter exploration (Schirner et al, 2018). Excitatory neural mass models of all 68 brain regions were subsequently coupled according to the individual subject’s empirical structural connectome and weighted by a global scaling factor ( G ) to simulate large-scale brain dynamics.…”

“…Different approaches can be used to simulate activity of nodes (i.e., brain areas) in large-scale brain network models varying from detailed spiking neuron models (Deco and Jirsa, 2012), over so-called neural mass or mean-field models that describe the collective activity of cell populations (Deco and Jirsa, 2012), down to abstract models such as the Ising model (Deco et al, 2012; Haimovici et al, 2013; Marinazzo et al, 2014; we refer the interested reader to Deco et al [2008] for an in-depth explanation of the general principle behind these approaches and the mechanism allowing a wide class of models to bridge temporal and spatial scales). In a clinical context, biologically interpretable dynamical models are of special interest, as their activity and parameters can allow inference of internal states and processes of the large-scale model, which cannot be measured with noninvasive neuroimaging techniques (Falcon et al, 2015; Schirner et al, 2018). Hence, they may provide an entry point for understanding brain disorders at a causal mechanistic level, which might lead to novel, more effective therapeutic interventions (Deco and Kringelbach, 2014; Proix et al, 2017).…”