Effects of Movement Speed and Analysis Window Characteristics on Finger Movement Classifications: An EEG Study

In recent years, many studies have been conducted on brain-computer interfaces using electroencephalograms obtained during the presentation of motor imagery. In this study, we investigated the effects of movement speed and analysis window characteristics, which are typical parameters used in electroencephalography studies of finger movement (bending and stretching), on finger movement classification accuracy. Two movement speeds (fast and slow) and two types of analysis window (synchronized and not synchronized with finger flexion) were considered. In an additional experiment, we investigated whether classification accuracy changes with window size. Two-way analysis of variance of the accuracies of the finger movement classifications according to movement speed and synchronization was then performed. Significant differences in accuracy were found according to both finger movement speed and analysis window characteristics, and the average classification accuracy was higher when movement speed was faster and the analysis window was synchronized with finger flexion. These results indicate that optimal conditions exist for obtaining high finger movement classification accuracy, and instructions should be provided in advance of task execution.