Simultaneous ocular and muscle artifact removal from EEG data by exploiting diverse statistics

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalography (EEG) recordings are susceptible to ocular and muscle artifacts.
• Current artifact removal methods often employ blind source separation (BSS) techniques using either second-order statistics (SOS) or higher-order statistics (HOS) separately.
• Existing methods rely on assumptions of statistical independence or distinct autocorrelation, which may not fully hold true for real-world artifacts.

Purpose of the Study:

• To introduce and evaluate Independent Vector Analysis (IVA) as a novel BSS technique for simultaneously addressing ocular and muscle artifacts in EEG.
• To demonstrate the capability of IVA to jointly utilize both HOS and SOS for artifact removal.
• To compare the performance of IVA against traditional separate SOS and HOS methods.

Main Methods:

• Implementation of Independent Vector Analysis (IVA), a BSS technique.
• Joint application of higher-order statistics (HOS) and second-order statistics (SOS) within the IVA framework.
• Validation through numerical simulations and analysis of real EEG recordings.

Main Results:

• IVA demonstrated superior performance in isolating both ocular and muscle artifacts compared to traditional methods.
• The effectiveness of IVA was particularly notable in raw EEG data with low signal-to-noise ratio.
• IVA successfully integrated separate SOS and HOS processing steps into a single, unified procedure.

Conclusions:

• Independent Vector Analysis (IVA) provides a unified and effective approach for removing ocular and muscle artifacts from EEG data.
• IVA overcomes the limitations of methods relying solely on HOS or SOS by jointly exploiting both statistical measures.
• The proposed method offers significant advantages for artifact correction, especially in challenging low signal-to-noise ratio EEG recordings.