Heterogeneity Counts More than Power for HD-sEMG-Based Joint Force Estimation.

Area of Science:

• Biomedical Engineering
• Neuroscience
• Signal Processing

Background:

• Accurate joint force estimation is crucial for biomechanics and rehabilitation.
• High-density surface electromyography (HD-sEMG) offers rich spatial information for muscle activity.
• Existing methods for force estimation from HD-sEMG can be limited by input signal processing.

Purpose of the Study:

• To develop and validate a novel input signal extraction method for joint force estimation using HD-sEMG.
• To leverage signal heterogeneity information for improved force prediction accuracy.
• To compare the performance of the proposed method against conventional power-based approaches.

Main Methods:

• HD-sEMG and joint force data were collected during isometric elbow flexion.
• Principal Component Analysis (PCA) was used to decompose HD-sEMG signals.
• Otsu's method and Moore-Neighbor tracing identified signal heterogeneity, guiding principal component selection.
• A polynomial fitting model was employed for force estimation using the processed HD-sEMG signal.

Main Results:

• The heterogeneity-based input signal significantly reduced force estimation error compared to the power-based signal.
• The selected principal component captured maximum signal heterogeneity, indicating key neuromuscular activation.
• The method demonstrated enhanced extraction of neuromuscular control information.

Conclusions:

• The proposed heterogeneity-based input signal extraction method improves joint force estimation accuracy from HD-sEMG.
• This approach offers a more informative signal for force modeling.
• Future work will involve testing the method on additional muscles and force tasks.