onEEGwaveLAD: A fully automated online EEG wavelet-based learning adaptive denoiser for artefacts identification and
Luca Longo1,2, Richard B Reilly1
1Trinity Centre for Biomedical Engineering, Trinity College Dublin, Dublin, Ireland.
Plos One
|January 28, 2025
Summary
A new automated method, onEEGwaveLAD, effectively removes artefacts from electroencephalography (EEG) signals in real-time. This technique enhances brain activity analysis for Brain-Computer Interfaces without needing multiple channels or human supervision.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Signal Processing
Background:
- Electroencephalography (EEG) offers high temporal resolution for brain activity monitoring.
- EEG signals are frequently contaminated by non-stationary artefacts, hindering accurate analysis.
- Existing denoising methods often require multi-channel data, reference signals, and offline processing.
Purpose of the Study:
- To introduce a novel, fully automated, online EEG denoising pipeline named onEEGwaveLAD.
- To address the need for real-time, unsupervised artefact reduction in EEG for practical applications like Brain-Computer Interfaces.
- To demonstrate the framework's adaptability for various artefact types, starting with blink artefact reduction.
Main Methods:
- Development of onEEGwaveLAD, a wavelet-based Learning Adaptive Denoiser pipeline.
- Instantiation of the framework for blink artefact detection and reduction.
- Evaluation of the denoising performance using signal-to-noise ratio analysis across 30 participants.
Main Results:
- The onEEGwaveLAD pipeline successfully identified and reduced blink artefacts in EEG signals.
- The method demonstrated effectiveness in enhancing signal quality for online processing.
- Quantitative analysis confirmed improvements in the signal-to-noise ratio.
Conclusions:
- onEEGwaveLAD provides a robust, automated, and online solution for EEG artefact reduction.
- The developed framework facilitates real-time EEG analysis, crucial for advancing Brain-Computer Interfaces.
- This research paves the way for more reliable and accessible EEG applications in diverse settings.
