Dynamic Causal Modeling for fMRI With Wilson-Cowan-Based Neuronal Equations

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Systems Neuroscience

Background:

• Dynamic Causal Modeling (DCM) infers directed brain connectivity from fMRI data.
• Standard DCM variants use bilinear equations, limiting interpretation of local neural dynamics.
• Interpreting macroscopic models in terms of microscopic neural behavior remains a challenge.

Purpose of the Study:

• To introduce a modified DCM for fMRI using non-linear Wilson-Cowan equations.
• To enhance the interpretability of DCM results regarding local neural dynamics.
• To validate the improved model evidence of the non-linear DCM variant.

Main Methods:

• Replaced the bilinear differential equation in DCM with a non-linear Wilson-Cowan equation.
• Employed Bayesian Model Comparison (BMC) to evaluate model evidence.
• Tested the modified DCM on empirical data (facial expression imitation, visual motion attention) and synthetic data.

Main Results:

• The non-linear Wilson-Cowan DCM variant demonstrated superior model evidence compared to the standard bilinear model.
• The modified model successfully analyzed data from 42 healthy participants performing a mirror neuron system (MNS) task.
• The sigmoid transfer function allowed direct inference from macroscopic to microscopic models.

Conclusions:

• The modified DCM with Wilson-Cowan equations provides a more accurate and interpretable model of brain connectivity.
• This approach bridges the gap between macroscopic imaging data and microscopic neural dynamics.
• The enhanced DCM variant shows significant potential for advancing neuroimaging analysis.