Skip to content

Developed Work and Current State (IV)

Developed Work and Current State (IV)

Detection and Classification of MUAPs in simulated multichannel EMG signals

Different methods have been evaluated in order to detect, extract and classify Motor Unit Action Potentials (MUAPs) from individual motor units in surface EMG signals. In order to evaluate the performance of the methods, artificial signals composed by different MUAP waveforms with various sampling rates and SNR have been simulated.

The developed algorithms are based in the information provided by 3 single differential channels recorded on a linear electrode array aligned with the propagation direction of the potentials. The detection of MUAPs in the EMG signal is based on wavelet techniques, while the classification of different groups corresponding to different MU is based on clustering techniques like Gaussian mixtures and k-means. Features like amplitude, conduction velocity and duration of the potential are used in the classification process, taking into account the refractory period of the MUs.