Influence of smoothing window length on electromyogram amplitude estimates

Area of Science:

• Biomedical Engineering
• Neuroscience
• Rehabilitation Science

Background:

• Surface electromyogram (EMG) signal processing is crucial for understanding muscle activity.
• Estimating EMG amplitude accurately requires careful consideration of signal processing parameters.
• The influence of smoothing window length on EMG amplitude estimates' signal-to-noise ratio (SNR) has not been systematically studied.

Purpose of the Study:

• To experimentally investigate how smoothing window length affects the SNR of EMG amplitude estimates.
• To compare the performance of different EMG processing schemes under varying window lengths.
• To provide guidance for selecting optimal smoothing window durations in surface EMG analysis.

Main Methods:

• Surface EMG signals were recorded during isometric contractions of biceps or triceps muscles (10%-75% MVC).
• EMG amplitude estimates were calculated using eight different processing schemes with smoothing window durations from 2.45 ms to 500 ms.
• Signal-to-noise ratio (SNR) was computed for each estimate, defining noise as deviations from the mean.

Main Results:

• SNR increased with smoothing window length in a square root relationship for all processing schemes.
• Whitened, multiple-channel EMG processing yielded higher average SNRs (3.2-37.3) compared to unwhitened, single-channel processing (1.4-16.2).
• Observed SNR performance did not reach theoretical predictions, indicating room for processor improvement.

Conclusions:

• The findings support the use of longer smoothing windows to improve EMG amplitude SNR.
• Results offer practical recommendations for selecting smoothing window lengths in conventional surface EMG studies.
• This research can inform the development of advanced EMG processors with adaptive window length tuning for time-varying signals.