Is sensor space analysis good enough? Spatial patterns as a tool for assessing spatial mixing of EEG/MEG rhythms

Area of Science:

• Neuroscience
• Biophysics

Background:

• Analyzing electroencephalography (EEG) and magnetoencephalography (MEG) data in sensor space is common but faces challenges due to volume conduction.
• Volume conduction causes signal mixing, where activity from different brain regions overlaps at individual sensors, complicating data interpretation.

Purpose of the Study:

• To illustrate the implications of volume conduction and signal mixing in EEG/MEG sensor space analysis.
• To demonstrate how spatial mixing affects the interpretation of brain rhythms, particularly alpha rhythms, and impacts common analytical approaches.

Main Methods:

• Simulations using a realistic 3D head model and lead field calculations.
• Analysis of a large resting-state EEG dataset using a sensor complexity measure.
• Illustrating signal mixing effects on alpha rhythms and their generators.

Main Results:

• Spatial mixing differentially affects electrode signals, with central electrodes showing more heterogeneous rhythms than posterior ones.
• Dominant occipital alpha rhythms can obscure other brain activities like sensorimotor mu-rhythms and temporal alpha rhythms.
• Strong occipital activity can significantly influence frontal channels, potentially compromising sensor space analyses.

Conclusions:

• Volume conduction poses significant challenges for EEG/MEG sensor space analysis, potentially leading to misinterpretations of brain activity.
• Researchers should carefully consider the impact of signal mixing when choosing sensor space for their investigations, especially for power, connectivity, and neurofeedback applications.