Simplified Optimal Estimation of Time-Varying Electromyogram Standard Deviation (EMGσ).

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Electromyogram (EMG) signal processing is crucial for applications like EMG to torque modeling.
• Traditional EMG whitening methods require subject-specific calibration, limiting their practical use.
• Additive noise in EMG signals can affect the accuracy of downstream applications.

Purpose of the Study:

• To develop a universal whitening filter for EMG signals that eliminates the need for subject-specific calibration.
• To evaluate the effectiveness of the universal whitening filter in improving EMG-torque modeling.
• To assess the impact of root difference of squares (RDS) processing on attenuating additive noise in EMG signals.

Main Methods:

• A universal whitening filter was created by averaging subject-specific filters from a dataset of 64 subjects.
• The processed EMG signals (with and without whitening) were used to model surface EMG to torque about the elbow.
• Root difference of squares (RDS) post-processing was applied to assess its noise attenuation capabilities.

Main Results:

• The universal whitening filter achieved comparable EMG-torque modeling benefits to traditional methods, improving performance by approximately 14% during dynamic contractions.
• Both traditional and universal whitening statistically improved EMG-torque modeling compared to unwhitened signals.
• RDS processing significantly reduced additive noise in EMG channels from 2-4% to below 1%, regardless of whether whitening was applied.

Conclusions:

• A universal whitening filter effectively processes EMG signals without cumbersome calibration, offering significant benefits for EMG-torque modeling.
• The combination of universal whitening filters and RDS processing shows promise for enhancing real-time applications, such as prosthesis control, by improving signal quality and reducing noise.