Independent component analysis based algorithms for high-density electromyogram decomposition: Systematic evaluation...

Area of Science:

• Neuromuscular Control
• Biomedical Signal Processing
• Computational Neuroscience

Background:

• Motor unit activity is crucial for understanding neuromuscular control.
• High-density electromyogram (HD EMG) recordings offer detailed insights.
• Accurate decomposition of HD EMG signals is essential for clinical and theoretical analysis.

Purpose of the Study:

• To systematically evaluate the performance of three independent component analysis (ICA)-based EMG decomposition algorithms: Infomax, FastICA, and RobustICA.
• To compare algorithm accuracy, motor unit yield, and computation time under various simulated HD EMG signal conditions.
• To provide guidance for selecting appropriate decomposition algorithms based on specific research needs.

Main Methods:

• Simulated HD EMG signals were generated across different muscle contraction levels and signal qualities.
• Performance metrics included decomposition accuracy, motor unit discharge timing precision, and yield.
• Three ICA algorithms (Infomax, FastICA, RobustICA) were applied to the simulated data.

Main Results:

• All three algorithms achieved high accuracy (85%-100%) in motor unit discharge timing.
• RobustICA demonstrated consistent high accuracy, particularly with low signal quality and varying contractions, but lower yield at high contractions.
• FastICA showed lower accuracy but detected the most motor units at high contraction levels; computation times were similar for FastICA and RobustICA.

Conclusions:

• Algorithm performance varies based on signal quality and contraction levels, impacting accuracy and motor unit yield.
• RobustICA is suitable for applications prioritizing accuracy, while FastICA excels in detecting a higher number of motor units.
• Findings guide the selection of ICA-based EMG decomposition algorithms for specific neuromuscular research applications.