Applying dynamic data collection to improve dry electrode system performance for a P300-based brain-computer...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Dry electrodes offer faster electroencephalography (EEG) setup compared to traditional wet electrodes.
• However, dry electrodes can produce noisier recordings due to suboptimal skin contact.
• This noise can potentially degrade the performance of brain-computer interfaces (BCIs).

Purpose of the Study:

• To compare the performance of wet versus dry electrodes in a P300 speller BCI system.
• To investigate the effectiveness of a dynamic data collection algorithm in compensating for the lower signal-to-noise ratio (SNR) of dry electrodes.
• To identify signal characteristics that differ between wet and dry electrode recordings.

Main Methods:

• Performance comparison of wet and dry electrodes using the P300 speller system.
• Inclusion of both healthy participants and individuals with communication disabilities (ALS, PLS).
• Application and evaluation of a data-driven dynamic stopping algorithm to optimize data collection.

Main Results:

• Dry electrodes resulted in significantly lower accuracy compared to wet electrodes in healthy participants.
• The dynamic stopping algorithm improved dry electrode performance but did not fully compensate for the reduced SNR.
• Dry electrode recordings showed consistently higher power in delta (0.1-4 Hz) and theta (4-8 Hz) frequency bands, suggesting artifact issues.

Conclusions:

• Despite faster setup, dry electrodes currently exhibit poorer online BCI performance than wet electrodes for both healthy and impaired users.
• Dynamic stopping algorithms offer partial mitigation for the lower SNR of dry electrodes.
• Further signal processing techniques are likely required to fully realize the potential of dry electrodes in BCI applications.