Effects of Different Re-referencing Methods on Spontaneously Generated Ear-EEG.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Ear-based electroencephalography (ear-EEG) offers unobtrusive, ambulatory brain activity monitoring.
• Miniaturized ear-EEG devices often require reference electrodes near recording sites, potentially distorting brain signals.
• Re-referencing is crucial to mitigate signal interference in ear-EEG recordings.

Purpose of the Study:

• To systematically investigate the impact of various re-referencing methods on ear-EEG signals.
• To evaluate how different re-referencing techniques affect alpha power and event-related (de)synchronization (ERD/ERS).
• To assess the influence of re-referencing on signal-to-noise ratio (SNR) and mental task classification using ear-EEG.

Main Methods:

• Compared five re-referencing methods (all-mean, contralateral-mean, ipsilateral-mean, contralateral-bipolar, ipsilateral-bipolar) on two ear-EEG datasets.
• Analyzed alpha power changes during eyes-closed (EC) and eyes-open (EO) tasks.
• Evaluated ERD/ERS during mental arithmetic (MA) and mental singing (MS) tasks.
• Estimated SNRs and assessed classification performance for MA vs. MS tasks.

Main Results:

• Alpha power and ERD/ERS patterns were generally consistent across re-referencing methods.
• The contralateral-mean method yielded statistically higher SNRs compared to other methods.
• Classification performance for mental tasks was significantly better with the contralateral-mean method.

Conclusions:

• The contralateral-mean re-referencing method is effective for improving ear-EEG signal quality and reliability.
• This method enhances the performance of unobtrusive, ambulatory ear-EEG applications, particularly for endogenous paradigms.
• Contralateral mean information optimizes spontaneous ear-EEG analysis.