The impact of MEG source reconstruction method on source-space connectivity estimation: A comparison between...

Area of Science:

• Neuroscience
• Biophysics
• Signal Processing

Background:

• Magnetoencephalography (MEG) is crucial for studying brain connectivity.
• Source-level connectivity analysis in MEG is challenged by the choice of inverse methods.
• The impact of inverse method selection on cortico-cortical coupling analysis is often overlooked.

Purpose of the Study:

• To investigate the effect of different inverse methods on detecting source coherence in MEG data.
• To compare the performance of L2-Minimum-Norm Estimate (MNE), LCMV beamforming, and DICS beamforming for connectivity analysis.
• To determine how source characteristics influence the optimal inverse method for MEG connectivity.

Main Methods:

• Simulated MEG data with thousands of randomly located source pairs were generated.
• Parameters manipulated included source correlation, size, and spatial configuration.
• Sensor-level MEG data were simulated at varying SNRs, and source-level coherence was calculated using MNE, LCMV, and DICS.

Main Results:

• Beamformers (LCMV, DICS) outperformed MNE for point-like sources.
• MNE showed superior performance for extended cortical patches with high intra-patch coherence.
• Beamformer performance improved for extended sources with partial intra-patch coherence.

Conclusions:

• The selection of an inverse method critically influences MEG source-space coherence results.
• Optimal inverse method choice depends on the spatial and synchronization properties of interacting cortical sources.
• Findings guide method selection and improve interpretation of MEG connectivity data.