Dynamic causal modelling revisited

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Systems Neuroscience

Background:

• Dynamic causal modeling (DCM) traditionally uses Taylor approximations for neuronal dynamics.
• Existing DCM methods have limitations in fully integrating laminar-specific neuronal activity with hemodynamic responses.

Purpose of the Study:

• To introduce a novel dynamic causal model for fMRI timeseries using a neural mass model of the canonical microcircuit.
• To enable the fusion of hemodynamic and electrophysiological data within a generative model.
• To facilitate Bayesian model comparison of hypotheses regarding synaptic effects and neuronal activity.

Main Methods:

• Replaced the standard Taylor approximation in DCM with a neural mass model of the canonical microcircuit.
• Developed a generative model for laminar-specific responses.
• Applied the model to the attention to visual motion dataset.

Main Results:

• The proposed model can generate both hemodynamic and electrophysiological measurements.
• It allows for the investigation of questions regarding the origins of hemodynamic responses (afferent vs. intrinsic activity).
• Provides a framework for assessing the role of inhibitory interneurons in neurovascular coupling.

Conclusions:

• The novel neural mass model-based DCM offers a more biologically plausible approach to analyzing fMRI data.
• This framework allows for addressing fundamental questions in fMRI concerning neuronal activity and its relationship with measured signals.
• It paves the way for more sophisticated analyses of brain function by integrating diverse neurophysiological data.