A Wavelet- and Machine Learning-Based Framework for the Automatic Detection of Artefacts in Electromyography REM...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Sleep Medicine

Background:

• Rapid eye movement sleep Without Atonia (RWA) is crucial for diagnosing REM Sleep Behavior Disorder (RBD).
• Current RWA scoring relies on visual inspection of electromyography (EMG) signals, which is time-consuming and susceptible to inter-rater variability due to artefacts.
• A standardized method for artefact removal in EMG recordings during REM sleep is lacking.

Purpose of the Study:

• To develop and validate a Machine Learning (ML) based framework for identifying artefacts in EMG signals during REM sleep.
• To facilitate automated RWA scoring by reducing manual labour and improving reliability.
• To propose a Matched-Wavelet approach for characterizing EMG signal morphology.

Main Methods:

• A Matched-Wavelet approach was used to characterize EMG signal morphology.
• A Machine Learning (ML) framework was developed to identify artefacts in EMG recordings during REM sleep.
• The framework was evaluated for its effectiveness in detecting artefacts from background and phasic activity.

Main Results:

• The best ML models achieved F1 scores of 79.2% for detecting background artefacts and 86.3% for detecting phasic activity artefacts.
• The proposed method demonstrates the feasibility of automatic artefact removal with low computational cost.
• The framework significantly improves the reliability of RWA assessments.

Conclusions:

• The developed ML framework offers a robust tool for artefact assessment in EMG recordings.
• Automated artefact removal enhances the reliability and diagnostic accuracy of RWA scoring.
• This approach contributes to more efficient and consistent diagnosis of REM Sleep Behavior Disorder (RBD).