Bi-Directional Imagined Hand Movement Classification Using Low Cost EEG-Based BCI

The notion of developing thought controlled devices (games, robots, cars etc.) is becoming increasingly popular with the introduction of low cost commercial headsets that record neuroelectric activity and the extensive research in the area of Brain Computer Interfaces (BCIs). In this paper, we study the feasibility of using a commercial low cost EEG amplifier which has only limited number of electrodes, to develop a motor control BCI system. The objective is to extract brain activity responsible for direction specific imagined and executed motor activity, which can be used to identify the motor task performed by the user using the simultaneously recorded EEG. An experiment is conducted to engage the user in bi-directional horizontal movement execution and imagination of the dominant hand. The analysis includes investigation of the time-frequency bins of the recorded EEG that provides maximum discrimination of directional movement. Further, the features are extracted using Filter Bank Common Spatial Pattern (FBCSP), followed by Fisher Linear Discriminant (FLD) for classification. The classification performance at various time instants of each trial are considered, and a control strategy was introduced at the classifier output to enhance performance. The performance in terms of average classification accuracy over five subjects is obtained as 81.3 % (movement execution) and 82.4 % (movement imagination). The results indicate the applicability of this EEG-BCI system to provide directional motor control to an interfaced device such as a robotic arm or a game element.