Analysis of dual-channel surface electromyogram using second-order and higher-order spectral features

Electromyography is gaining prominence in wearable electronic devices with improved design of surface electrodes and increased demand in various fields such as health monitoring, prosthetic control and assistive technology. Surface electromyogram signals have been analyzed using conventional time-domain, frequency-domain, and time-frequency domain features as well as high-order spectral analysis since they have been shown to be non-Gaussian random processes arising from non-linear system. In this paper, two novel high-order spectral features have been proposed for analysis of dual-channel surface electromyogram signals for different hand activities and rest position. The characteristics of the surface electromyogram signals revealed by the proposed features are shown to be different from the corresponding features extracted from second-order spectra.