Dynamical causal modelling for M/EEG: spatial and temporal symmetry constraints

Area of Science:

• Neuroscience
• Computational Neuroscience
• Biophysics

Background:

• Dynamic Causal Modelling (DCM) analyzes brain activity using spatiotemporal generative models.
• Magneto- and electroencephalography (M/EEG) data are often analyzed with equivalent current dipole (ECD) models.
• Classical ECD models use hard symmetry constraints for homologous sources, assuming symmetric activation.

Purpose of the Study:

• To describe the application of spatial and temporal constraints within DCM for M/EEG.
• To introduce and evaluate 'soft' symmetry constraints using informed Bayesian priors in DCM.
• To enable testing for symmetry in temporal or spatial components independently.

Main Methods:

• Utilized dynamic causal modelling (DCM) with spatiotemporal generative models for M/EEG data.
• Implemented informed Bayesian priors for 'soft' symmetry constraints, allowing data to modulate constraint influence.
• Employed Bayesian model comparison to evaluate symmetric and non-symmetric model variants.

Main Results:

• Demonstrated the use of soft symmetry constraints in DCM for M/EEG data.
• Showed that symmetry constraints can be applied to temporal (neural-mass) or spatial (lead-field) components independently.
• Identified that soft symmetry priors are recommended for evoked responses to bilateral sensory input.

Conclusions:

• Soft symmetry priors in DCM offer a flexible approach to modelling brain activity compared to hard constraints.
• DCM with soft symmetry priors allows for data-driven assessment of symmetry in neural sources.
• This methodology enhances the analysis of M/EEG data, particularly for bilateral sensory processing.