Research Spotlight Archive
Title: Decomposition of Highly Unpredictable Real-Life EMG Signals
Funding: National Institutes of Health
Background: The EMG signal is composed of the action potentials from groups of muscle fibers organized into functional units called motor units. This signal can be detected with sensors placed on the surface of the skin or with needle or wire sensors introduced into the muscle tissue. When only two or three motor units in the vicinity of the sensors are active, it is usually possible to visually identify most of the individual motor unit action potentials, because the incidence of superposition amongst the individual motor unit action potentials is relatively low. However, when the EMG signal contains the activity of four or more motor units the individual action potentials become, in large part, indistinguishable to the naked eye because the incidence of superposition among two or more motor unit action potentials becomes numerous and the shapes of the motor unit action potentials may approach in similarity.
Description: In many cases, it is desirable to study and/or employ the information contained in the timing of the discharges of individual motor units. For example, in investigations for furthering the understanding of how motor units are controlled by the CNS in generating force; or for assessing the degree of dysfunction in upper motoneuron diseases such as cerebral palsy, Parkinson’s disease, amyotrophic lateral sclerosis (ALS), and other disorders. This may be achieved by “decomposing” the EMG signal obtained from a sensor placed on the skin surface near the muscle. The concept is depicted in the figure below. A decomposed EMG signal provides all the information available in the EMG signal. The timing information provides a complete description of the inter-pulse interval, firing rate and synchronization characteristics. The morphology of the shapes of the motor unit action potentials provides information concerning the anatomy and health of the muscle fibers.
Results: We have recently developed algorithms that can decompose real-life surface EMG signals from muscle contractions ranging up to maximum voluntary force levels into 20-50 motor unit action potential trains per contraction with accuracy levels > 90%. This stands in marked contrast to the performance of all other known surface EMG signal decomposition algorithms. Those algorithms have been reported to accurately decompose only a few motor units per muscle contraction, and even that performance level is only reached when real-life muscle contractions are limited to very low force levels.
Publications: S. H. Nawab, S. S. Chang, and C. J. De Luca, “High-Yield Decomposition of Surface EMG Signals,” Clinical Neurophysiology, vol. 121, pp. 1602-1615, October 2010.
S. H. Nawab, R. P. Wotiz, and C. J. De Luca, “Decomposition of Indwelling EMG Signals,” Journal of Applied Physiology, vol. 105, pp. 700-710, August 2008.