Graph Signal Processing as a tool for mitigating the impact of spatial blurring in EEG-based neuroelectrical imaging.

Area of Science:

• Neuroscience
• Signal Processing
• Biomedical Engineering

Background:

• Spatial blurring in EEG scalp maps is influenced by electrode placement and volume conduction.
• Graph Signal Processing (GSP) offers methods to decompose EEG signals into network harmonics.
• Crosstalk among electrodes can be modeled as a linear combination of slow-varying network harmonics.

Purpose of the Study:

• To investigate the impact of different GSP filtering techniques on EEG scalp map spatial resolution.
• To determine if graph filtering can mitigate spatial blurring caused by crosstalk in EEG.
• To assess the contribution of network harmonics to spatial blurring in EEG scalp maps.

Main Methods:

• Applied various graph filtering procedures to EEG signals from 15 healthy subjects performing motor tasks.
• Analyzed grand average scalp maps derived from filtered EEG data.
• Utilized GSP to decompose EEG signals into network harmonics.

Main Results:

• High-pass graph filtering of EEG signals demonstrably improved scalp localization.
• Slow-varying network harmonics were found to lack spatial localization, contributing to scalp map blurring.
• Specific graph filtering procedures effectively mitigated spatial blurring.

Conclusions:

• GSP provides a practical approach to enhance the spatial resolution of EEG scalp maps.
• High-pass filtering is effective in improving scalp localization in EEG.
• Understanding network harmonic contributions is crucial for reducing spatial blurring in EEG analysis.