Source activity correlation effects on LCMV beamformers in a realistic measurement environment

Area of Science:

• Neuroscience
• Biophysics
• Signal Processing

Background:

• Source localization in MEG/EEG studies is crucial for understanding brain connectivity.
• Beamformers are common tools for source localization but can exhibit bias due to correlated source time courses.
• The real-world impact of this bias on MEG/EEG data has remained debated.

Purpose of the Study:

• To investigate the effect of correlated source activity on beamformer localization performance using a realistic MEG phantom.
• To quantify the impact of correlation degree and source distance on localization accuracy.
• To address the long-standing debate on the practical implications of correlated sources in real MEG data.

Main Methods:

• Utilized a phantom in a realistic MEG environment to simulate simultaneously active neural sources.
• Varied the correlation degree and spatial distance between simulated current dipoles.
• Applied a linear constrained minimum variance (LCMV) beamformer to analyze the oscillating signals.

Main Results:

• High correlation significantly impacts localization for dipoles within 1-5 cm, causing them to merge.
• For dipoles separated by 3 cm, increasing correlation leads to attenuated power amplitudes and blurred source estimations.
• The study provides empirical evidence for the detrimental effects of source correlation on MEG source localization.

Conclusions:

• Correlated neural activity poses a significant challenge for accurate source localization in MEG.
• The spatial proximity of sources exacerbates the negative impact of correlation on beamformer performance.
• These findings highlight the need for advanced methods to mitigate correlation bias in MEG/EEG connectivity analyses.