Robust interpolation of EEG/MEG sensor time-series via electromagnetic source imaging

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalography (EEG) and magnetoencephalography (MEG) are crucial non-invasive neuroimaging techniques.
• Limitations include low spatial resolution, incomplete brain coverage, and sensor noise, which distort brain activity reconstructions.
• Addressing these limitations is vital for accurate clinical and cognitive neuroscience research.

Purpose of the Study:

• To propose a robust electromagnetic source imaging framework for interpolating poor quality or missing EEG/MEG measurements.
• To enhance the spatial resolution and data quality of EEG and MEG.
• To overcome sensor array limitations and improve the estimation of brain dynamics.

Main Methods:

• A two-step framework: 1. Robust inverse algorithm to estimate brain source activity using good sensor data. 2. Interpolation of poor quality/missing sensor data using reconstructed sources and leadfield matrices.
• Evaluation through simulations and real EEG/MEG datasets.
• Comparison against neighborhood and spherical spline interpolation benchmarks.

Main Results:

• The proposed framework demonstrates superior performance compared to benchmark methods in simulations and real-world data.
• It shows robustness against highly correlated brain activity and low signal-to-noise ratio (SNR) data.
• Accurate estimation of cortical dynamics was achieved, outperforming existing methods.

Conclusions:

• The developed framework provides a rigorous platform for enhancing EEG and MEG data quality.
• It effectively addresses challenges posed by partial sensor coverage (e.g., in optically pumped magnetometer arrays) and noisy sensors.
• Implementation can expand the applications of EEG and MEG by improving data reliability and analytical capabilities.