Detection of stretch reflex onset based on empirical mode decomposition and modified sample entropy

Area of Science:

• Rehabilitation Medicine
• Biomedical Engineering
• Signal Processing

Background:

• Accurate spasticity assessment is crucial for effective rehabilitation.
• Detecting stretch reflex onset is key for evaluating spasticity.
• Surface electromyogram (sEMG) signals in spasticity present challenges like false peaks and unstable data length, hindering accurate onset detection.

Purpose of the Study:

• To propose a novel method for detecting stretch reflex onset in sEMG signals.
• To address the limitations of traditional sEMG analysis in spasticity assessment.
• To enhance the accuracy of rehabilitation evaluation through improved onset detection.

Main Methods:

• Utilizing empirical mode decomposition (EMD) for denoising sEMG signals.
• Applying modified sample entropy (MSE) for recognizing stretch reflex onset.
• Comparing the proposed EMD-MSE method with existing algorithms like wavelet thresholding.

Main Results:

• EMD demonstrates superior denoising performance for sEMG signals compared to wavelet thresholding.
• The proposed method increased the stretch reflex onset recognition rate by 56% without added noise.
• Optimal MSE parameters achieved up to 100% recognition rate, with an average of 93%.

Conclusions:

• EMD effectively removes noise and baseline activity from sEMG signals preceding stretch reflex onset.
• The combined EMD and MSE approach significantly improves stretch reflex onset identification accuracy.
• This method enables precise determination of stretch reflex onset for objective spasticity evaluation.