Motor imagery BCI control based on dry EEG acquisition and short training period: Evaluation study on the control of an asynchronous 2D car game

Motor imagery-based brain–computer interfaces (MI-BCIs) using EEG acquisition have been extensively studied and successfully applied to control various applications. However, most of these studies are conducted in highly controlled laboratory environments and do not address challenges related to the use of these systems in out-of-lab scenarios. In this paper, we evaluate a MI-BCI system under two key conditions critical for designing MI-BCIs for out-of-lab use. The first is the type of EEG electrodes used for brain activity measurement. While previous studies have provided evidence supporting the use of dry EEG electrodes in exogenous BCI paradigms, their potential in endogenous paradigms such as motor imagery remains underexplored, due to the complexity of decoding internally generated neural activity without external stimulation. The second is the calibration time, as the long training periods typically required in such studies are impractical for everyday use in real-world scenarios. In this paper, we investigate the performance of a MI-BCI system under these two conditions. We implemented a MI-BCI to control a real-time 2D game where participants controlled a car. The data was acquired using dry EEG, and participants underwent a short training period to evaluate the performance of this BCI in pseudo-realistic scenario. The study involved 30 healthy subjects. Various metrics were used for evaluation. The results of the study were promising: 40% of subjects exhibited high-level motor imagery control, 30% had medium-level motor imagery control, 20% of subjects demonstrated low-level motor imagery control, and only 10% of subjects were not able to control the game using motor imagery. In addition, we also analyzed the subjects’ impressions and feelings during different runs of this experiment. In this paper, we showed the feasibility of using dry EEG acquisition and a short training duration to control an endogenous MI-BCI game. Moreover, the results and data acquired in this study will help to advance the development of MI-BCI based on EEG acquisition in out-of-lab scenarios.