The detection of muscle activity using surface electromyography is one of the fundamental steps in many applications, such as motor control, posture and gait analysis, and myoelectric control of prosthetic devices.
Research has widely reported various computerized detection techniques of muscle activity onset. Among them, the most common used parameters are associated with the electromyography signal amplitude.
However, most electromyography onset detection methods up to now only give electromyography burst presence information in the time domain. The more detailed information in each frequency component is absent.
Recently, researchers from the Rehabilitation Institute of Chicago, the Institute of Acoustics of the Chinese Academy of Sciences and the Northwestern University (Chicago) work together to have presented a novel approach. This approach is for muscle activity measurement in the time-frequency domain by using electromyography burst presence probability.
The electromyography burst presence is represented by electromyography burst presence probability in each subband of the electromyography signal.
The constrained sequential hidden Markov model is employed to model the log-power sequence. The reason for using it is that the hidden Markov model is intrinsically advantageous in modeling the temporal correlation of electromyography burst/non-burst presence.
The electromyography burst presence probability is eventually yielded by hidden Markov model based on the criterion of maximum likelihood.
The performance of the proposed method is examined using both simulated and experimental surface electromyography signals.
The results show that the electromyography burst presence probability can effectively detect bursts of electromyography by suppressing the interference of frequency components of the non-burst electromyography. Besides, the electromyography burst presence probability method is resilient to noise.
This method is assessed by simulated and real surface electromyography signals. The real surface electromyography signals are recorded from the brachioradialis of one intact subject (31, male) with a Refa electromyography system (TMS International B.V., Netherlands).
The experimental results show that the electromyography onset can be visually determined from the electromyography burst spectral structure depicted by the electromyography burst presence probability than the spectrograms of the electromyography signal.
Funding for the research come part from the National Institutes of Health of the U.S. Department of Health and Human Services under Grant 2R24HD050821, part from the National Natural Science Foundation of China under Grant 61271426, and part from the Strategic Priority Research Program of the Chinese Academy of Sciences (No. XDA06030100).
References:
LIU Jie, YING Dongwen, William Zev Rymer. EMG Burst Presence Probability: A Joint Time-frequency Representation of Muscle Activity and Its Application to Onset Detection. Journal of Biomechanics (Vol.48, No. 6, 13 April 2015, pp. 1193–1197). doi:10.1016/j.jbiomech.2015.02.017