Reliable and accurate information extraction from surface electromyographic signals

The electrical activity generated in contracting muscles is measured using electromyographic (EMG) signals. By placing an array of electrodes on the surface of the skin, surface EMG (sEMG) signals are recorded non-invasively. The sEMG signal is a stochastic signal whose amplitude is generally between 1 and 10 mV, with the most dominant spectral power between 50 and 150 Hz. The sEMG signal is often corrupted by various types of noise, such as movement artifacts, power-line interference and activity from the other muscles. Moreover, the electrode placement could affect the recorded signals. They make analysing and classifying sEMG signals difficult. sEMG has applications in rehabilitation, sport science, kinesiology, ergonomics, muscle architecture identification, neurological disease diagnosis, prosthesis control and human–machine interfaces. In this chapter, the sEMG signal processing methods used in such applications are discussed, and critical issues are considered. Finally, a basic prosthesis control example is provided for interested readers.This work was supported by the Ministry of Economy and Competitiveness (MINECO), Spain, under contract DPI2017-83989-R and Ministry of Science and Innovation (MICINN), Spain, under contract PRE2018-085387. CIBER-BBN is an initiative of the Instituto de Salud Carlos III, Spain. JFA is a Serra Hunter Fellow. This work was also supported in part by the University of Isfahan.Peer ReviewedPostprint (published version