Improving Localization Accuracy of Neural Sources by Pre-processing: Demonstration With Infant MEG Data

Area of Science:

• Biophysics
• Neuroscience
• Medical Imaging

Background:

• Infant magnetoencephalography (MEG) studies are technically demanding.
• Existing MEG data acquisition methods face challenges in clinical and research settings.

Purpose of the Study:

• To address specific challenges in infant MEG data acquisition and pre-processing.
• To provide solutions applicable to various difficult MEG scenarios, including clinical applications.

Main Methods:

• Detailed discussion of pre-processing steps: movement compensation, artifact suppression (internal/external magnetic interference, physiological signals like cardiac artifacts).
• Emphasis on noise covariance structure in inverse modeling and proper matrix treatment.
• Assessment of pre-processing outcomes by comparing signal representations before and after processing.

Main Results:

• Identified that statistical metrics can erroneously indicate reliable source reconstruction despite severe head movement distortions.
• Demonstrated the impact of pre-processing on signal quality and source localization accuracy.
• Highlighted the critical role of noise covariance in accurate inverse modeling.

Conclusions:

• Proper signal processing is essential for reliable infant MEG studies.
• Standard statistical metrics may be misleading in the presence of significant artifacts like head movements.
• The presented methods offer solutions for improving MEG data quality in challenging scenarios.