Controlling an electromyography-based power-assist device for the wrist using electroencephalography cortical currents. (17th January 2017)

• Record Type: Journal Article
• Authors: Kawase, Toshihiro; Yoshimura, Natsue; Kambara, Hiroyuki; Koike, Yasuharu
• Is Part Of: Advanced robotics. Volume 31:Number 1/2(2017)
• Page Start: 88
• Page End: 96
• Publication Date: 2017-01-17
• Subjects: Brain–machine interface -- current source estimation -- robot control -- EEG; Robotics -- Periodicals; Robotics -- Japan -- Periodicals; Robotics; Japan; Periodicals; 629.89205
• DOI: 10.1080/01691864.2016.1215935
• Languages: English
• ISSNs: 0169-1864

Abstract

Graphical Abstract: Abstract : A power-assist device controlled with brain activity would offer improved quality of life for people with motor impairments. Such a device requires extracting the necessary information from brain activity, and if possible, in a non-invasive manner. Here, we reconstructed muscle activity for wrist flexion and extension from electroencephalography (EEG) cortical current source (CCS) signals as well as standard EEG sensor signals. We then used the reconstructed muscle activity to control an electromyography (EMG)-based power assist device. We evaluated performance by comparing equilibrium point, stiffness, and torque estimated from EEG with those calculated from EMG. We found that EEG-CCS signals provided higher estimation accuracy than EEG sensor signals. Moreover, we performed offline control of the EMG-based device using reconstructed muscle activity signals. Our results suggest that muscle activity can be reconstructed from EEG-CCS signals with sufficient accuracy to control an EMG-based power-assist device. We demonstrate that brain–machine interfaces can be used to control EMG-based power-assist devices, mutually broadening their applicability in rehabilitation.