The Effect of the number of EMG signals on the efficiency of muscle synergy extraction

One common hypothesis in motor control is that each human movement is created by the combination of a small number of muscle synergies. The muscle synergy analysis is a muscle activation decomposition technique used in electromyogram signals. In this study, simulated muscle activation signals are produced by the combination of wavelets, similar to the real signals. The algorithm of non-negative matrix factorization is used in extracting the muscle synergy and reconstructing signals. We investigated the effect of number of muscle activation signals on the efficiency of muscle synergy extraction. All processes are performed on signals with and without noise, considering the signal to noise ratio between 0 and 20 dB.According to the results, the reconstruction error in signals without noise was very small and was increased by noise addition. In addition, in both cases of signals with and without noise with increased number of signals, mean of RMSE values tend to decrease. However by selecting a percentage of the best results, mean and standard deviation of RMSE values are reduced significantly, and mean of RMSE values with increasing number of signals trend to increase. If there is less possibility of repetition due to online processing limitations, to reduce error, it is necessary to increase the number of muscle signals. Otherwise, to reduce error, instead of selecting a large number of muscles can be used less muscles to record signals and can be increased the number of iterations and then can be selected a percentage of the best results.