Implications of probabilistic activation on estimating the number of motor units in a muscle

Methods aimed at estimating the numbers of motor units [MUs] in muscles have gained increased prominence over the course of the last several years. Currently, all of the these methods are based on a sampling approach in which the properties of only a handful of MUs are examined. It is thus critical that 1) the MU sample be representative of the entire MU population and 2) the signals assumed to represent individual motor unit action potentials [MUAPs] are indeed valid MUAPs. The authors are particularly interested in the latter. Specifically, in the majority of methods, the MU sample is obtained by stimulating the nerve with carefully graded stimulus pulses. In doing so, it is assumed that each successive increase in the recorded compound muscle potential corresponds to the successive activation of an individual MU. This assumption is valid only in those cases in which each MU can be activated independently from the remaining non-active MUs. In practice, only the first few MUs can be activated in this manner. Successive MUs then tend to be activated in groups. Hence, the resultant muscle potentials can represent various combinations of active and/or inactive MUs, and thus successive incremental increases in the observed compound muscle potential will not necessarily correspond to the additional activation of new MUs. As a result, the average MUAP size (i.e. amplitude or area) calculated using these methods tends to be under-estimated, which in turn leads to the over-estimation of the number of MUs in the muscle. Here the authors examine the magnitude of this error, which increases with the number of MUs undergoing probabilistic activation, and investigate possible means of alleviating it