BCI classification task

Seven healthy volunteers (3 male and 4 female) participated in the study, all were right-handed and between the ages of 23 to 30 (mean age 28). All participants gave written, informed consent to participate in the study. Participants were seated in a chair at a distance of 80 cm from an LCD screen with both hands resting on a table. They held a tennis ball in each hand and were told to remain relaxed and strive to minimize movement and eye blinks. When required to respond, they were to squeeze the tennis ball in their hand but try to avoid tensing their arms or shoulders. Each session (ME and MI—Figure 2) was repeated twice. The whole experiment thus consisted of four sessions. Every session lasted 30–40 minutes with 10 to 15 minutes breaks between sessions. The duration of the whole experiment, including setup, was kept below 3 hours to minimize fatigue. EEG data was recorded and sampled at 250 Hz using 63 active electrodes (BrainVision actiCHamp) placed according to the 10/20 montage. Bipolar electromyography (EMG) electrodes were placed on the Brachioradialis for both hands as a sanity check for any movement in MI session.Sessions 1 and 3 were designed to identify EEG signals related to ME. Participants were instructed to squeeze the tennis ball with their right or left hand while fixating on the cross displayed on the screen. They were encouraged to minimize all other movement and to only use the designated hand. One hundred trials were collected for each hand.Session 2 and 4 aimed to show that a decoding model based on actual ME, derived from the first session, could be used to decode EEG activity in the absence of execution. Participants were instructed to carry out MI of the repetitive hand movement instructed in session 1 while fixating on the cross displayed on the screen. One hundred trials were collected for both left and right imagination per each session. All other aspects of the task were identical to session 1. This session also allowed us to screen participants for the presence of motor-related EEG oscillations, and at least minimal voluntary control over these oscillations. Hence, overall, we collected 200 trials of ME and 200 trials of MI for each subject.<br>The dataset was saved to Matlab data file (X1.mat, X2.mat, ... X7.mat).<br>Each data file includes Xsr, Xsi ,ysr ,ysi.s : subject IDr : real or motor execution (ME)i : motor imagery (MI)<br>X [trial, sample, channels]y label: 1 and 3(left) , 2 and 4(right)<br>**ME is 4s and MI is 6s. In our dataset, we put one second before the ME and MI. In total, ME file has 5s (1s break+ 4s ME task) and MI file has 7s (1s break+ 6s ME task)**<br>