Fractal dimension of surface EMG and its determinants

The fractal dimension (FD) has been proposed as a useful measure for the characterization of EMG interference patterns. This paper studies the relationship between the FD and the properties of EMG signals, such as recruitment number, firing rate and motor unit action potentials (MUAPs). The study is based on simulated surface EMG signals from a volume conductor computation. The FD is calculated by a box counting algorithm. It is shown that recruitment is the main determinant of the FD. A sensitive region and a saturation region are proposed to distinguish the two parts of the overall FD-recruitment relationship. The FD increases steeply as the recruitment number increases in the sensitive region, and it reaches a plateau in the saturation region. The FD-recruitment relationship is found to be strongly affected by the firing rate and the MUAP duration, but not by MUAP amplitude variations. The effects of the firing rate mainly lie in the sensitive region. However, the influence of duration is manifested in the saturation region. By understanding the effects of the physiological properties of the surface EMG, we hope that it can benefit the application of fractal analysis to EMG, in such areas as neuromuscular disorder diagnosis, muscle fatigue monitoring, etc