An iterative algorithm for the separation of action potential sequences.

Various iterative algorithms for separation of superimposed event sequences were designed, and their efficiency examined through simulation studies. It has been found that a double parallel-serial algorithm improved markedly the efficiency of the filtration process. For the same probability of detection of the event series to be extracted, the probability of 'impurity' events dropped by roughly 0.10, as compared to the non-iterative fixed window width algorithm. Conclusions are drawn regarding separation of motor unit action potential sequences from the total electromyogram.