Noise Optimization of Basic Signal Component Extraction for Cryogenic and On-Scalp Magnetoencephalography (MEG)

Area of Science:

• Neuroscience
• Biophysics
• Biomedical Engineering

Background:

• Magnetoencephalography (MEG) provides non-invasive, high spatio-temporal precision measurement of neural activity, crucial for cognitive neuroscience.
• On-scalp MEG technologies like OPM-MEG enhance sensitivity to neuronal magnetic fields, improving source localization.
• Current MEG preprocessing methods, while essential for isolating neural signals, risk discarding valuable data or introducing artifacts.

Purpose of the Study:

• To introduce and evaluate a novel preprocessing method for MEG data.
• To address limitations in existing signal space separation (SSS) techniques.
• To enhance the accuracy and robustness of neuronal source localization in MEG and OPM-MEG.

Main Methods:

• Development and application of a novel SSS method incorporating Foster's inverse, a weighted matrix inversion protocol.
• Utilizing sensor noise and artifact information to reconstruct neuronal activity.
• Validation through simulations, phantom head recordings, and subject recordings with two OPM-MEG systems.

Main Results:

• Foster's inverse with SSS demonstrates a more robust and stable reconstruction of neuronal magnetic fields compared to existing methods.
• The proposed method effectively mitigates the impact of sensor noise and artifacts on signal reconstruction.
• Improved signal fidelity was observed across simulations, phantom data, and human subject recordings.

Conclusions:

• Foster's inverse with SSS presents a powerful advancement in MEG data preprocessing.
• This technique offers significant improvements in noise reduction and source localization accuracy for both conventional MEG and OPM-MEG.
• The method is poised to benefit cognitive neuroscience research utilizing advanced MEG technologies.