Towards resolving the co-existing impacts of multiple dynamic factors on the performance of EMG-pattern recognition based prostheses

• Published in: Computer methods and programs in biomedicine Vol. 184; p. 105278
• Main Authors: Asogbon, Mojisola Grace, Samuel, Oluwarotimi Williams, Geng, Yanjuan, Oluwagbemi, Olugbenga, Ning, Ji, Chen, Shixiong, Ganesh, Naik, Feng, Pang, Li, Guanglin
• Format: Journal Article
• Language: English
• Published: Ireland Elsevier B.V 01.02.2020
• Subjects: Adult; Artificial Limbs; Electromyogram (EMG); Electromyography - methods; Female; Humans; Male; Maximum Voluntary Contraction (MVC); Movement - physiology; Muscle contraction force variation; Pattern recognition; Pattern Recognition, Automated - methods; Principal Component Analysis; Subject mobility; Upper-limb prostheses; Young Adult; Electromyogram (EMG); Muscle contraction force variation; Subject mobility; Upper-limb prostheses; Pattern recognition; Maximum Voluntary Contraction (MVC)
• DOI: 10.1016/j.cmpb.2019.105278
• ISSN: 0169-2607; 1872-7565

•This study systematically investigated the co-existing impact of multiple dynamic factors on the performance of EMG pattern recognition system (EMG-PR).•An invariant time-domain descriptor was proposed to resolve such co-existing impacts with its performance validated.•The proposed method significantly mitigated combined impact of such factors on the performance of the EMG-PR system.•The outcomes of the study would be potential for improving the clinical robustness of multifunctional myoelectric prostheses. Mobility of subject (MoS) and muscle contraction force variation (MCFV) have been shown to individually degrade the performance of multiple degrees of freedom electromyogram (EMG) pattern recognition (PR) based prostheses control systems. Though these factors (MoS-MCFV) co-exist simultaneously in the practical use of the prosthesis, their combined impact on PR-based system has rarely been studied especially in the context of amputees who are the target users of the device. To address this problem, this study systematically investigated the co-existing impact of MoS-MCFV on the performance of PR-based movement intent classifier, using EMG recordings acquired from eight participants who performed multiple classes of targeted limb movements across static and non-static scenarios with three distinct muscle contraction force levels. Then, a robust feature extraction method that is invariant to the combined effect of MoS-MCFV, namely, invariant time-domain descriptor (invTDD), was proposed to optimally characterize the multi-class EMG signal patterns in the presence of both factors. Experimental results consistently showed that the proposed invTDD method could significantly mitigate the co-existing impact of MoS-MCFV on PR-based movement-intent classifier with error reduction in the range of 7.50%~17.97% (p<0.05), compared to the commonly applied methods. Further evaluation using 2-dimentional principal component analysis (PCA) technique, revealed that the proposed invTDD method has obvious class-separability in the PCA feature space, with a significantly lower standard error (0.91%) compared to the existing methods. This study offers compelling insight on how to develop accurately robust multiple degrees of freedom control scheme for multifunctional prostheses that would be clinically viable. Also, the study may spur positive advancement in other application areas of medical robotics that adopts myoelectric control schemes such as the electric wheelchair and human-computer-interaction systems.