An Adaptive Spatial Filtering Method for Multi-Channel EMG Artifact Removal During Functional Electrical Stimulation with Time-Variant Parameters.
Xiaoling Chen,Yuntao Jiao,Dong Zhang,Ying Wang,Xinyu Wang,Yuqi Zang,Zhenhu Liang,Ping Xie
DOI: https://doi.org/10.1109/tnsre.2023.3311819
IF: 4.9
2023-01-01
IEEE Transactions on Neural Systems and Rehabilitation Engineering
Abstract:Removing the stimulation artifacts evoked by the functional electrical stimulation (FES) in electromyogram (EMG) signals is a challenge. Previous researches on stimulation artifact removal have focused on FES modulation with time-constant parameters, which has limitations when there are time-variant parameters. Therefore, considering the synchronism of muscle activation induced by FES and the asynchronism of muscle activation induced by proprioceptive nerves, we proposed a novel adaptive spatial filtering method called G-S-G. It entails fusing the Gram-Schmidt orthogonalization (G-S) and Grubbs criterion (G) algorithms to remove the FES-evoked stimulation artifacts in multi-channel EMG signals. To verify this method, we constructed a series of simulation data by fusing the FES signal with time-variant parameters and the voluntary EMG (vEMG) signal, and applied the G-S-G method to remove any FES artifacts from the simulation data. After that, we calculated the root mean square (RMS) value for both preprocessed simulation data and the vEMG data, and then compared them. The simulation results showed that the G-S-G method was robust and effective at removing FES artifacts in simulated EMG signals, and the correlation coefficient between the preprocessed EMG data and the recorded vEMG data yielded a good performance, up to 0.87. Furthermore, we applied the proposed method to the experimental EMG data with FES-evoked stimulation artifact, and also achieved good performance with both the time-constant and time-variant parameters. This study provides a new and accessible approach to resolving the problem of removing FES-evoked stimulation artifacts.