1. The influence of common component on myoelectric pattern recognition
- Author
-
Xu Zhang, Bo Yao, Yingchun Zhang, Yun Peng, Ping Zhou, and Jiangbo Pu
- Subjects
Male ,Medicine (General) ,sequential feed-forward selection method ,030204 cardiovascular system & hematology ,Biochemistry ,Signal ,Models, Biological ,TMSi system ,03 medical and health sciences ,0302 clinical medicine ,R5-920 ,Interference (communication) ,Component (UML) ,Medicine ,Humans ,Medical systems ,signal shielding technology ,business.industry ,Electromyography ,Muscles ,Biochemistry (medical) ,Reproducibility of Results ,Pattern recognition ,Cell Biology ,General Medicine ,Neuromuscular Diseases ,Special Issue: Application of emerging and advanced rehabilitation engineering techniques in stroke recovery ,common component ,myoelectrical pattern recognition ,030220 oncology & carcinogenesis ,Pattern recognition (psychology) ,Electromagnetic shielding ,Female ,Artificial intelligence ,business ,Algorithms - Abstract
Objective Using the Twente Medical Systems international B.V. (TMSi) electromyography (EMG) system, active signal shielding was applied to clean signals and create data without interference and cable movement artifacts. TMSi, used in high-density surface EMG pattern recognition, controls myoelectric rehabilitation robots, yet few have studied how active signal shielding influences pattern recognition. This study aimed to investigate how active signal shielding used within the TMSi influenced motion pattern recognition. Methods Surface EMG of dominant side forearm and hand muscles was studied in eight healthy participants. The common component’s influence was accessed by the classification performance of wrist and hand functional movements. Results The classification performance of EMG signals with the common component was obviously lower than signals without the common component using one to five electrodes. Conversely, a higher motion classification performance was achieved using signals with the common component using over 12 electrodes. Optimal channel distribution was examined based on the sequential feed-forward selection method, showing that the common component affected the optimal channel location. Conclusions Active signal shielding in the TMSi improved classification accuracy in motion pattern recognition when over 12 electrodes were used. The optimal channel distribution was related to the common component when using the TMSi.
- Published
- 2020