Resolving superimposed MUAPs using particle swarm optimization

This paper presents an algorithm to resolve superimposed action potentials encountered during the decomposition of electromyographic signals. The algorithm uses particle swarm optimization with a variety of features including randomization, crossover, and multiple swarms. In a simulation study involving realistic superpositions of two to five motor-unit action potentials, the algorithm had an accuracy of 98%. Index Terms--Alignment, decomposition, electromyography, particle swarm optimization, superposition.