Noise-assisted multivariate empirical mode decomposition for multichannel EMG signals

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Ensemble Empirical Mode Decomposition (EEMD) is widely used for single-channel Electromyography (EMG) signal processing.
• Limited research exists on temporal and spatial characteristics across multiple muscle groups using multichannel EMG signals.

Purpose of the Study:

• To compare Empirical Mode Decomposition (EMD)-based approaches for multichannel EMG signal processing.
• To evaluate Noise-Assisted MEMD (NA-MEMD) for its effectiveness in analyzing spatial and temporal characteristics of muscle activity.

Main Methods:

• Utilized experimental data from 11 healthy male subjects performing leg extension, standing flexion, and walking exercises.
• Applied and compared three EMD-based methods: EEMD, Multivariate EMD (MEMD), and NA-MEMD.
• Assessed methods based on the number of Intrinsic Mode Functions (IMFs), mode-alignment, and mode-mixing.

Main Results:

• Both MEMD and NA-MEMD methods consistently produced an equal number of IMFs.
• NA-MEMD demonstrated optimal performance in mode-alignment and mode-mixing compared to MEMD and EEMD.
• MEMD showed slightly better performance than EEMD in terms of mode-alignment and mode-mixing.

Conclusions:

• Proposes NA-MEMD as an effective approach for multichannel EMG signal processing.
• NA-MEMD enables simultaneous analysis of IMFs across different frequency bands.
• Highlights the clinical implications for understanding neuromuscular coordination in daily functional activities.