Rigorous performance assessment of the algorithms for resolving motor unit action potential superpositions

Area of Science:

• Biomedical Engineering
• Neuroscience
• Signal Processing

Background:

• Intramuscular electromyography (EMG) signal decomposition is crucial for extracting motor unit action potential (MUAP) waveforms and firing times.
• Existing algorithms like Peel-Off (PO), Branch and Bound (BB), Genetic Algorithm (GA), and Particle Swarm Optimization (PSO) aim to resolve superimposed MUAPs.

Purpose of the Study:

• To quantitatively compare the accuracy and running time of PO, BB, GA, and PSO algorithms for MUAP decomposition.
• To identify the most effective algorithm for resolving superimposed MUAPs under varying signal conditions.

Main Methods:

• Simulated 1000 superimposed MUAP instances using two template sets (varying energies and similarity) with added noise.
• Resolved superpositions using Peel-Off (PO, POI), Branch and Bound (BB), Genetic Algorithm (GA), and Particle Swarm Optimization (PSO).
• Utilized generalized estimating equations for significance testing and rank products for overall performance ranking.

Main Results:

• Overall ranking favored BB, PSO, and GA over PO and POI.
• Set 1 (wide energy range) ranking: PSO, BB, GA, PO, POI.
• Set 2 (high similarity) ranking: BB, PSO, GA, PO, POI.
• BB algorithm, while generally fast, exhibited slow performance in specific scenarios, limiting real-time applicability.

Conclusions:

• BB and PSO algorithms offer superior performance for MUAP decomposition compared to PO and GA.
• Algorithm choice depends on specific application requirements, particularly regarding real-time processing needs.
• Further optimization of BB may enhance its suitability for real-time EMG decomposition.