Dynamic Causal Modeling on the Identification of Interacting Networks in the Brain: A Systematic Review

Abstract

Dynamic causal modeling (DCM) has long been used to characterize effective connectivity within networks of distributed neuronal responses. Previous reviews have highlighted the understanding of the conceptual basis behind DCM and its variants from different aspects. However, no detailed summary or classification research on the task-related effective connectivity of various brain regions has been made formally available so far, and there is also a lack of application analysis of DCM for hemodynamic and electrophysiological measurements. This review aims to analyze the effective connectivity of different brain regions using DCM for different measurement data. We found that, in general, most studies focused on the networks between different cortical regions, and the research on the networks between other deep subcortical nuclei or between them and the cerebral cortex are receiving increasing attention, but far from the same scale. Our analysis also reveals a clear bias towards some task types. Based on these results, we identify and discuss several promising research directions that may help the community to attain a clear understanding of the brain network interactions under different tasks.

Topographic variation in neurotransmitter receptor densities explains differences in intracranial EEG spectra

“…The source-specific, four-population neural mass models (Figure 1E) generated power spectral densities (PSD) (1-60Hz, 1Hz steps) with very good fit for most iEEG recordings with few exceptions (2.4% (42/1770) electrode contacts showed a noticeable error; frequency fit: MSE median = 1.04*10 -6 and mode = 1.29*10 -11 ) (Figure 1F). In addition, the modelled PSDs match previously reported profiles (54)(55)(56) and showed characteristic lobe-specific features such as an alpha peak in the occipital lobe (Figure 1C), but also indicated high intra-region and inter-subject variability (Figure 1D). The canonical microcircuit model with four neural populations, which approximate cortical layer-specific activity, was used to model iEEG in the frequency domain.…”

“…A possible computational approach to integrate both iEEG and neuroreceptor data and to thereby provide neurobiological grounding, is to incorporate the receptor features as prior information to model fitting of electrophysiological recordings. Bayesian approaches such as dynamic causal modelling (DCM) (50,51) (50)(51)(52)(53)(54)(55)(56) allow for this integration of empirically derived priors (57,58), which enables us to evaluate different hypotheses about how neuroreceptor composition contributes to regional oscillatory signatures (iEEG).…”

Target localization intervention and prognosis evaluation for an individual with mild cognitive impairment

“…used the normal form of a Hopf bifurcation to simulate the state changes of a single brain node and analyzed the changes in resting‐state functional connectivity (FC) in the whole brain during AD progression. The dynamic causal model (DCM), a directed model, can describe differences in neuronal signals over time, providing a powerful and intuitive analytical tool for functional network simulation based on structural networks [9,10] …”

“…The dynamic causal model (DCM), a directed model, can describe differences in neuronal signals over time, providing a powerful and intuitive analytical tool for functional network simulation based on structural networks. [9,10] Most existing control schemes use state reproduction to adjust the system state by directly adjusting the parameter values in the model. [11] The control parameters cannot be automatically adjusted in real-time according to the status of the nodes and the system.…”