Refined signal space separation methods for on-scalp MEG systems

Area of Science:

• Biophysics
• Neuroimaging
• Signal Processing

Background:

• Biomagnetic measurements are crucial for understanding brain activity.
• Signal Space Separation (SSS) is a key data processing technique for magnetoencephalography (MEG).
• The advent of on-scalp optically-pumped magnetometers (OPMs) necessitates refined SSS methods for improved spatial resolution.

Purpose of the Study:

• To develop and evaluate novel multi-origin signal space separation (mSSS) techniques for on-scalp MEG systems.
• To adapt SSS methods, including vector spheroidal harmonics, for improved brain-space coverage with OPMs.
• To assess the performance of mSSS against standard SSS using simulated and real MEG data.

Main Methods:

• Investigated two-origin mSSS and spheroidal constructions.
• Utilized simulated data with internal dipole sources and external interference across various MEG systems (Kernel Flux OPM, MEGIN SQUID, QuSpin OPM).
• Processed real auditory evoked data, empty-room data, and audiovisual data from OPM systems.

Main Results:

• mSSS demonstrated comparable or superior stability to SSS across all tested sensor geometries.
• mSSS effectively reconstructed internal signals while suppressing external interference, outperforming SSS with on-scalp systems.
• Results showed mSSS provides a lower noise floor and better performance, especially for low-channel-count OPM systems.

Conclusions:

• mSSS is a robust and straightforward modification to SSS, suitable for novel on-scalp MEG sensor systems.
• Updated data analysis techniques like mSSS are essential as on-scalp MEG systems become more prevalent.
• The proposed mSSS method enhances the reliability and spatial resolution of biomagnetic measurements using OPMs.