A New Method for Human Mental Fatigue Detection with Several EEG Channels
Previous researchers have applied EEG in human mental fatigue detection and concluded that the ratio of slow wave to fast wave EEG activities increased with increasing mental fatigue. But these methods failed to reflect functional interactions between different brain areas. The objective of this paper is to develop an indicator for detecting human mental fatigue with several electroencephalogram (EEG) channels considering functional interactions among the brain. To this end, we collected eighteen participants’ EEG data, and constructed the adjacency matrix on the basis of EEG electrode positions using mutual information (MI) to determine the functional connectivity between all pairwise EEG channels. The maximum eigenvalue of adjacency matrix was discussed during the mental fatigue process and considered as the reflections of the corresponding network features. The results indicated that significant statistical differences of the maximum eigenvalue were found only in alpha1 (8–10 Hz) band during task state. With increasing levels of mental fatigue, the maximum eigenvalue increased, indicating that the maximum eigenvalue can be used as an indicator for mental fatigue estimation taking brain function changes into account. For real-time engineering applications, we reduced the EEG channels by finding the central vertices in the brain functional network using weighted degree centrality. Finally, the channels of F3, F4, C3, C4, P3, P4, Fz, Cz, and Pz were obtained for further applications. We also provided an algorithm for automatic human mental fatigue evaluation with these 9 EEG channels. Our results have potential applications in several domains, including in flight, traffic and industrial human safety.