HEAR to remove pops and drifts: the high-variance electrode artifact removal (HEAR) algorithm

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• High-quality electroencephalographic (EEG) signals are crucial for functional neuroimaging and brain-computer interfacing (BCI).
• Transient artifacts, such as electrode pop and drift (PD), often contaminate EEG data, stemming from impedance variations at the electrode-scalp interface due to ion concentration changes.
• Existing artifact correction methods struggle to effectively remove these transient, high-variance artifacts while preserving signal integrity.

Purpose of the Study:

• To introduce the high-variance electrode artifact removal (HEAR) algorithm and its online variant (oHEAR) for mitigating transient PD artifacts in EEG signals.
• To evaluate the performance of (o)HEAR against state-of-the-art artifact correction techniques using both simulated and real-world EEG data.
• To demonstrate the utility of (o)HEAR in improving data quality for BCI applications and neuroimaging studies.

Main Methods:

• Development and implementation of the HEAR algorithm, designed to identify and remove high-variance transient artifacts.
• Validation of HEAR and oHEAR using simulated EEG datasets with introduced PD artifacts.
• Application of (o)HEAR to real-world EEG data, including recordings from a center-out reaching task and BCI training scenarios.

Main Results:

• HEAR and oHEAR demonstrated superior performance in removing simulated PD artifacts compared to existing methods, achieving approximately 25 dB attenuation.
• The algorithm successfully maintained a high signal-to-noise ratio (SNR) during artifact-free periods.
• For real-world EEG data, (o)HEAR halved the proportion of outlier trials and preserved the waveform of movement-related cortical potentials.
• In BCI training, oHEAR improved feedback reliability by mitigating the impact of PD artifacts.

Conclusions:

• The HEAR algorithm provides an effective, open-source solution for removing transient PD artifacts from EEG signals.
• oHEAR enhances the reliability and quality of EEG data, particularly benefiting BCI applications by improving user feedback.
• The developed algorithm represents a significant advancement in artifact correction for electrophysiological data, enabling more robust neuroimaging and BCI research.