Enhanced Neural Decoding with Optically Pumped Magnetometer MEG Using Multivariate Pattern Analysis

Area of Science:

• Neuroscience
• Biophysics
• Biomedical Engineering

Background:

• Magnetoencephalography (MEG) decodes neural activity using multivariate pattern analysis (MVPA).
• Optically pumped magnetometers (OPMs) offer improved spatial resolution over traditional superconducting quantum interference devices (SQUIDs) due to closer scalp proximity.

Purpose of the Study:

• To directly compare the decoding performance of OPM-MEG and SQUID-MEG within subjects.
• To assess the impact of spatial sampling and frequency content on decoding accuracy.
• To evaluate the potential of OPM-MEG for enhanced neural decoding applications.

Main Methods:

• Within-subject comparison of OPM-MEG and SQUID-MEG using time-resolved MVPA.
• Decoding neural responses to visual images and words.
• Controlled sensor counts and spatial-frequency analysis using spherical harmonic expansion.

Main Results:

• OPM-MEG demonstrated higher decoding accuracy than SQUID-MEG in sensor-matched comparisons.
• The advantage of OPM-MEG was most pronounced in word decoding, outperforming even full-array SQUID-MEG.
• Higher-order spatial components significantly benefited OPM-MEG decoding performance.

Conclusions:

• OPM-MEG provides enhanced recoverability of fine-grained neural representations for MVPA.
• This advancement has significant implications for cognitive neuroscience and brain-computer interface development.
• OPM-MEG represents a promising technological leap for high-resolution neural decoding.