Novel formulation of a double threshold algorithm for the estimation of muscle activation intervals designed for...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Neuroscience

Background:

• Surface electromyography (sEMG) signal analysis is crucial for understanding muscle activity.
• Existing double threshold algorithms for sEMG onset-offset detection can degrade in performance with fluctuating signal-to-noise ratios (SNR).
• Adaptive algorithms are needed to maintain detection accuracy under varying SNR conditions.

Purpose of the Study:

• To develop an improved, adaptive formulation of the double threshold algorithm for sEMG onset-offset detection.
• To enhance the algorithm's robustness to changes in SNR during signal analysis.
• To validate the novel formulation's performance on both simulated and real sEMG data.

Main Methods:

• An adaptive double threshold algorithm was developed, updating detection parameters based on on-line SNR estimation.
• The novel algorithm was tested against the standard fixed-threshold approach using simulated sEMG data with constant and time-varying SNR.
• Performance was further evaluated on real sEMG data recorded during isometric contractions at various force levels.

Main Results:

• The novel adaptive algorithm demonstrated superior performance compared to the standard approach under time-varying SNR conditions.
• For constant SNR > 8 dB, the standard algorithm showed bias and standard deviation < 10 and 15 ms, with >95% detection.
• Under time-varying SNR (10-25 dB), the standard approach's detection rate dropped to 50%, while the novel method maintained higher accuracy.
• Real-world data showed significant improvements: standard implementation (StD=134 ms, FP=22%) vs. novel (StD=42 ms, FP=2%) on average.

Conclusions:

• The developed adaptive double threshold algorithm effectively addresses the limitations of fixed-threshold methods in sEMG analysis.
• The algorithm's ability to adapt to changing SNR improves detection accuracy in both static and dynamic muscle activation scenarios.
• This enhanced detection capability holds promise for real-time biofeedback applications utilizing myoelectric information.