Related Experiment Video
Updated: Oct 7, 2025
06:09
P300-Based Brain-Computer Interface Speller Performance Estimation with Classifier-Based Latency Estimation
Published on: September 8, 2023
695
Automated Artifact Rejection Algorithms Harm P3 Speller Brain-Computer Interface Performance
David E Thompson1, Md Rakibul Mowla1, Katie J Dhuyvetter1
1Brain and Body Sensing (BBS) Lab, Mike Wiegers Department of Electrical and Computer Engineering, Kansas State University, Manhattan, KS, USA.
Brain Computer Interfaces (Abingdon, England)
|January 6, 2022
Summary
Automated artifact removal methods for electroencephalogram (EEG) brain-computer interfaces (BCIs) often degrade performance. Even the best methods significantly reduced P3 Speller BCI accuracy, highlighting challenges in EEG signal processing for assistive technologies.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Brain-Computer Interfaces (BCIs) offer communication and control for individuals with severe paralysis.
- Non-invasive electroencephalogram (EEG)-based BCIs are susceptible to significant noise from artifacts like electro-oculogram (EOG).
- Artifacts can be substantially larger than the neural signals of interest, complicating data interpretation.
Purpose of the Study:
- To compare the performance of ten different automated artifact removal methods for EEG.
- To evaluate the impact of these methods on P3 Speller Brain-Computer Interface (BCI) performance.
Main Methods:
- Implementation and testing of ten distinct automated artifact removal algorithms.
- Application of these algorithms to EEG recordings used for a P3 Speller BCI.
- Quantitative assessment of BCI accuracy before and after artifact removal.
Main Results:
- All tested automated artifact removal methods significantly reduced P3 Speller BCI performance.
- Methods were more likely to decrease BCI accuracy than to improve it.
- SOBI, JADER, and EFICA were the least detrimental, yet still caused an average ~10% drop in BCI accuracy.
Conclusions:
- Current automated artifact removal techniques may not be suitable for enhancing EEG-based BCI performance.
- The empirical reduction in BCI accuracy suggests potential mechanistic issues with these signal processing methods.
- Further research is needed to develop artifact removal strategies that preserve or improve BCI efficacy.
