Extraction of the envelope from surface EMG signals

The envelope of a surface myoelectric signal has been historically related to the force exerted by muscles during contraction. In fact, during isometric contractions, signal amplitude has been shown to be linearly related to the force. This relationship is no longer valid when myoelectric data are recorded during body movement. In routine work, the envelope of the signal is extracted by means of a technique based on a full-wave rectifier followed by an integrator (smoothing filter) giving rise to the so-called Integrated ElectroMyography (IEMG). This technique presents some drawbacks that are mainly related to the subjective choice of parameters and to the loss of adaptivity to signal characteristics, which limit its use in dynamic protocols and affects the comparison of results obtained by different experimenters. New approaches are therefore needed. This paper presents a method which aims to improve the quality of the estimation and the standardization of the results while, at the same time, being suitable for signals recorded both in static and dynamic conditions. The new approach is based on an adaptive iterative procedure which automatically sets and dynamically changes (according to signal characteristics) the length of the smoothing filter used for the estimation. The estimation error is far lower than that given by classical estimators and approaches the theoretical lower bound. Moreover, the automatic choice of filter length guarantees good repeatability of the results and standardization in the processing approaches. This technique can therefore help in analyzing myoelectric signals recorded during dynamic protocols, and in studying the physiological mechanism driving muscular force during movement, and also the evolution of muscular fatigue.