Estimation of filtering properties of living tissue for inverse filtering of surface EMG signals

We present a method for the estimation of the filtering properties of living tissue from the experimental output data stimulated by inputting sinusoidal electric signals through intramuscular electrodes in the lower leg muscle. This method is based on approximating the tissue filter by a rational transfer function. The estimation procedure consists of three parts: estimation of frequency response from sinusoidal input and output data using a correlation method, recursive weighted least square (RWLS) estimation to synthesize a continuous transfer function from the frequency response data and discretization of the transfer function. The estimated "tissue filter" is then used in recovering the intramuscular signal from the surface signal. An example of inverse filtering real EMG waveforms using the estimated tissue filter is given.