Feasibility of a sensorimotor rhythm based mobile brain-computer interface

Abstract

Recently, mobile BCI has gained increasing attention. In the present study we investigated whether electroencephalogram (EEG) signals can be reliably measured when the subject is in walking condition. 5 subjects were recruited to perform motor tasks (making their fists with left or right hand) in two sessions, which included the control session (sitting in the armchair) and the experimental session (walking on the treadmill). Two sessions were conducted in the same environment. The mean classification accuracies are 70% (seated) and 66% (walking), so a bit lower classification accuracies when walking compared to that when sitting in the condition of performing motor execution for most subjects. Moreover, the motor and sensory cortices were activated and obvious ERD of contralateral area can be seen when performing motor execution no matter the subject was sitting or walking, and no obvious artifacts according to the presented time-frequency analysis. This leads us to conclude that it is possible to establish a mobile BCI system based on portable EEG devices.