Monte Carlo simulation studies of EEG and MEG localization accuracy

Area of Science:

• Neuroscience
• Biophysics
• Medical Imaging

Background:

• Electroencephalography (EEG) and magnetoencephalography (MEG) are crucial non-invasive techniques for mapping brain activity.
• Source localization accuracy is influenced by inverse methods, source models, data characteristics, and head volume conductor (forward) models.

Purpose of the Study:

• To theoretically compare the source localization accuracy of EEG alone, MEG alone, and combined EEG/MEG datasets.
• To evaluate the impact of forward model errors by removing their influence using Monte Carlo simulations.
• To assess a novel noise sensitivity normalized inverse operator and the utility of fMRI constraints.

Main Methods:

• Utilized Monte Carlo simulations to isolate the effects of forward model errors on source localization.
• Employed a linear estimation inverse approach with a distributed source model and a realistic forward head model.
• Evaluated accuracy using crosstalk and point spread metrics, and examined a noise sensitivity normalized inverse operator.

Main Results:

• EEG source localization demonstrated higher accuracy than MEG for an equivalent number of sensors.
• Combined EEG/MEG datasets yielded the highest localization accuracy.
• The noise sensitivity normalized inverse operator enhanced spatial resolution compared to the standard operator.
• Incorporating a priori fMRI constraints consistently reduced both crosstalk and point spread errors.

Conclusions:

• Despite MEG's higher magnetic sensitivity, EEG offers superior accuracy in certain source localization scenarios.
• Combining EEG and MEG data provides the most accurate brain activity localization.
• Noise sensitivity normalization and fMRI constraints are valuable tools for improving spatial resolution and accuracy in source localization.