Automatic Detection of Cursor Movements from the EEG Signals via Deep Learning Approach

The classification of motor imagery (MI) tasks is one of the key objectives of EEG-based brain-computer interface (BCI) systems. To ensure successful classification performance to BCI systems, researchers endeavor to extract appropriate features. However, these challenges are based on the conventional method. In this study, EEG signals related to MI tasks are classified using the convolutional neural network (CNN), which does not need a separate feature extraction. EEG records, which are generally evaluated as one-dimensional in machine learning problems, were taken into consideration as the image representation by using a novel method. The datasets were taken from a healthy subject. The subject was asked to move a cursor up and down on a computer screen, while his cortical potentials were taken. The EEG signals recorded over the 3.5-second time interval were evaluated for both the whole time and sub time intervals. Thus, the most effective time interval that has distinguishing features for EEG recordings related to different cursor movements was tried to be determined as well. As a result, it has been shown that the proposed model based on deep learning approach can successfully classify EEG signals related to cursor movements.