Application of bispectrum based signal reconstruction to sEMG signal

The surface electromyogram (sEMG) conveys information about the physiological properties of muscles. Unlike the power spectrum, the bispectrum can suppress noise when characterizing non-Gaussian random signals. In this paper we esablish a bispectrum based method to estimate a motor unit action potential from a simulated sEMG signal, improving on an earlier approach which combined bispectrum and power spectrum.