Investigating Window Segmentation on Mental Fatigue Detection Using Single-Channel EEG

Mental fatigue condition can be a serious problem if it is not handled properly. It also has a correlation with acute or chronic illness. Many research has been done to detect mental fatigue condition using several methods. The Physiological method is proved as a robust indicator, one of which is electroencephalogram (EEG). EEG is the most widely used as a physiological indicator in the few decades. However, most of the research in mental fatigue detection based on EEG used long time segment and complex computation method. In this paper, a window segmentation was employed to investigate mental fatigue information that might contain in a specific segment. Power percentage feature was extracted from each segment. The detection of mental fatigue employs three classifiers, LDA, QDA, and SVM. According to our experiment, LDA yields the highest performance with 92.82 % of accuracy. This result obtained from 30s length window segment which contains only the first and the last segment of the EEG signal data points. This result showed that information of mental fatigue in EEG signal may be better detected in short time segment and can be found in specific window segment.