Acquisition and Decomposition of the EMG Signal

In this chapter we refer to the decomposition of the myoelectric (ME) signal as the procedure by which the ME signal is separated into its constituents motor units action potential trains (MUAPTs). This concept is illustrated in Figure 1. The development of a system to accomplish such a decomposition will be beneficial to both researchers interested in understanding motor unit properties and behaviour, and clinicians interested in assessing and monitoring the state of a muscle. In the clinical environment, measurements of some characteristics of the motor unit action potential (MUAP) waveform (for example shape and amplitude) are currently used to assess the severity of a neuromuscular disease or in some cases to assist in making a diagnosis. Thus, the decomposition of the ME signal is useful in two ways. First, a partial decomposition must be implicitly performed by the clinical investigator to insure that what is actually observed is a MUAP and not a superposition of two or more MUAPs or some other ephemeral artifact. Second averaging the MUAP waveforms present in the same MUAPT will produce a low noise representation of the MUAP and hence provide a more faithful representation of the events occuring within the muscle. Any decomposition scheme devised for such application, (i. e. to extract only MUAP shape and amplitude) will have weak constraints on its performance. A useful technique should allow detection of some (but not necessarily all) firing of a single unit in a particular record. Simultaneous observation of more than one MUAP, although useful, is not necessary because several MUAPs can be derived from different records with only one MUAP detected per record.