Progressive FastICA Peel-Off and Convolution Kernel Compensation Demonstrate High Agreement for High Density Surface...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electromyogram (EMG) decomposition is crucial for understanding motor unit plasticity.
• High-density surface EMG requires advanced signal processing techniques for accurate decomposition.
• Convolution Kernel Compensation (CKC) and Progressive FastICA Peel-off (PFP) are recent methods for EMG decomposition.

Purpose of the Study:

• To compare the performance of CKC and PFP methods for high-density surface EMG decomposition.
• To assess the accuracy and yield of motor unit identification using both CKC and PFP.
• To investigate the potential of combining CKC and PFP for enhanced EMG decomposition.

Main Methods:

• Independent application of CKC and PFP algorithms to 64-channel surface EMG data.
• EMG signals were recorded from the first dorsal interosseous (FDI) muscle of 9 neurologically intact subjects.
• Analysis of 91 trials to identify and compare motor units detected by each method.

Main Results:

• A total of 1477 motor units were identified across both methods, with 969 common motor units.
• An average of 10.6 ± 4.3 common motor units per trial were identified.
• High agreement (97.85 ± 1.85%) in discharge instants for common motor units was observed, supporting decomposition accuracy.

Conclusions:

• Both CKC and PFP demonstrate high accuracy in decomposing high-density surface EMG signals.
• The high agreement in common motor units validates the decomposition capabilities of both methods.
• Combining CKC and PFP approaches may offer increased motor unit decomposition yield.