Intervention

Non-dominant hand function training was performed by the MI-BCI system (BCI-Hand with 24 EEG channels, Rehab Medical Technology Co., Ltd., Shenzhen, China), which consists of an EEG cap, a computer terminal (i.e., the control interface), an external manipulator, and a 23-inch computer monitor (Figure 2A). Subjects performed motor imagery by watching video cues about hand function (Figure 2B). The EEG cap was based on the International 10–20 System as a reference (Figure 2C), with 24-electrode conduction channels (including 22 recording electrodes and 2 reference electrodes) setting over the frontal and parietal regions. The EEG data were collected using the EEG amplifier with unipolar Ag/Ag-cl electrode channels, digitally sampled at 256 Hz with a 22-bit resolution for voltage ranges of ±130 mV. The real-time EEG signals collected were amplified by computer terminal according to the central processing control algorithm, and the mu ERD score (score from 0 to 100) during motor imagery was calculated. The external robotic arm would be driven when the score reaches 60 points. The non-dominant hand could perform the action while providing real-time feedback (both sensory and visual) (Figure 2D).

Diagram of the motor imagery brain computer interface (MI-BCI) upper limb rehabilitation training system. (A) The MI-BCI training setting. (B) The screen providing visual clues for motor imagery. (C) The robotic arm for motion performing and feedback. (D) Electroencephalography (EEG) electrode placement.