Instantiating the onEEGwaveLAD Framework for Real-Time Muscle Artefact Identification and Mitigation in EEG Signals
Luca Longo1,2, Richard Reilly2
1The Artificial Intelligence and Cognitive Load Research Lab, Centre of Explainable Artificial Intelligence, Technological University Dublin, D07 EWV4 Dublin, Ireland.
Sensors (Basel, Switzerland)
|August 28, 2025
Summary
This study introduces an automated online denoiser to remove muscle artefacts from electroencephalography (EEG) signals using wavelet transforms and machine learning. The novel system effectively cleans EEG data in real-time, improving signal quality for brain activity research.
Area of Science:
- Neuroscience
- Signal Processing
- Machine Learning
Background:
- Electroencephalography (EEG) is vital for brain activity research but often corrupted by artefacts.
- Existing artefact removal methods are typically offline and require human intervention.
- Automated, real-time artefact mitigation is crucial for advancing EEG applications.
Purpose of the Study:
- To develop a novel, fully automated online denoiser for identifying and mitigating artefacts in EEG signals.
- To create an adaptive framework for artefact-specific and context-dependent parameter instantiation.
- To instantiate the framework for real-time muscle artefact identification and removal.
Main Methods:
- Utilized Discrete Wavelet Transformation (DWT) for time-frequency analysis of EEG data.
- Employed the Isolation Forest algorithm for anomaly detection in a sliding buffer.
- Implemented a denoising strategy operating on DWT coefficients before returning to the time domain.
Main Results:
- The proposed framework successfully identified myogenic muscle movements in EEG signals.
- The system demonstrated effectiveness in transforming artefact-contaminated EEG into cleaner signals.
- Challenges related to the wavelet transform's cone of influence and real-time processing tradeoffs were highlighted.
Conclusions:
- The developed online wavelet-based adaptive denoiser shows promise for real-time EEG artefact mitigation.
- The framework offers a flexible approach adaptable to specific artefact types and contexts.
- Further research is needed to address wavelet transformation limitations and optimize real-time performance.
