GOAL: The purpose of this study was to design a method for extracting the volitional EMG from recorded surface electromyography (EMG), contaminated by functional electrical stimulation (FES) artifact. METHODS: Considering that the FES artifact emerges periodically with rather large amplitude in nonstationary EMG, we designed an adaptive-matched filter (AMF) via genetic algorithm (GA) optimization. Both the simulated and real data from seven subjects were processed, using the GA-AMF filter and comb filter, respectively. To test the filtering effect on the EMG, contaminated with FES artifact of different current intensities, the contaminated EMG was simulated by combining the simulation artifact and clean EMG with various FES artifacts to clean EMG ratios. RESULTS: The results show that, in simulation test, compared to the EMG filtered by comb filter, the simulated EMG (p < 0.05), filtered by using GA-AMF, had significantly higher correlation coefficient, higher signal to noise ratio, and lower normalized root mean square error, whereas the real EMG (p < 0.05), filtered by using GA-AMF had higher power reduction than that filtered by using comb filter. The results indicate that GA-AMF can effectively remove FES artifact from the EMG of the stimulated muscle and its adjacent muscle, and the GA-AMF filter performed better than did the comb filter. CONCLUSION: All these results demonstrate that the GA-AMF filter is capable of extracting volitional EMG from the stimulated muscle and adjacent muscles. SIGNIFICANCE: GA-AMF could provide technical support for improving EMG feedback control of FES rehabilitation system.
GOAL: The purpose of this study was to design a method for extracting the volitional EMG from recorded surface electromyography (EMG), contaminated by functional electrical stimulation (FES) artifact. METHODS: Considering that the FES artifact emerges periodically with rather large amplitude in nonstationary EMG, we designed an adaptive-matched filter (AMF) via genetic algorithm (GA) optimization. Both the simulated and real data from seven subjects were processed, using the GA-AMF filter and comb filter, respectively. To test the filtering effect on the EMG, contaminated with FES artifact of different current intensities, the contaminated EMG was simulated by combining the simulation artifact and clean EMG with various FES artifacts to clean EMG ratios. RESULTS: The results show that, in simulation test, compared to the EMG filtered by comb filter, the simulated EMG (p < 0.05), filtered by using GA-AMF, had significantly higher correlation coefficient, higher signal to noise ratio, and lower normalized root mean square error, whereas the real EMG (p < 0.05), filtered by using GA-AMF had higher power reduction than that filtered by using comb filter. The results indicate that GA-AMF can effectively remove FES artifact from the EMG of the stimulated muscle and its adjacent muscle, and the GA-AMF filter performed better than did the comb filter. CONCLUSION: All these results demonstrate that the GA-AMF filter is capable of extracting volitional EMG from the stimulated muscle and adjacent muscles. SIGNIFICANCE: GA-AMF could provide technical support for improving EMG feedback control of FES rehabilitation system.
Authors: Kevin Joseph; Soheil Mottaghi; Olaf Christ; Thomas J Feuerstein; Ulrich G Hofmann Journal: Front Neurosci Date: 2018-04-30 Impact factor: 4.677
Authors: Minjae Kim; Yaejin Moon; Jasmine Hunt; Kelly A McKenzie; Adam Horin; Matt McGuire; Keehoon Kim; Levi J Hargrove; Arun Jayaraman Journal: Front Hum Neurosci Date: 2021-06-03 Impact factor: 3.169