An experimental and model based investigation of the potential and limitations of surface EMG spectral analysis for assessment of motor unit recruitment strategy

Characteristic frequencies of the surface EMG power spectrum have been used in the past as indicative of motor unit (MU) recruitment, since they are rather insensitive to changes of MU firing rates and thus they should remain constant when only rate coding is used to modulate muscle force. However, this speculation has not been yet validated by simulated and experimental data. In this paper, a model of surface EMG signal generation and detection is used to simulate EMG signals detected during linearly increasing force contractions. Different MU control strategies (corresponding to different ways for force generation by recruitment and rate coding) are simulated. A number of simulations are performed to study the effect of a random distribution of MUs in the muscle's cross-section upon the surface EMG. The results are compared with those obtained analyzing the EMG signals detected experimentally during linearly increasing force contractions of the biceps brachii muscle in 10 subjects. Results show that the volume conductor properties may act as confounding factors which may mask any relationship between characteristic spectral frequencies and conduction velocity as a size principle parameter. It is concluded that more advanced signal processing techniques which aim at the analysis of single MU activity are required for the surface EMG based assessment of the central nervous system control strategy.