A morphological approach for mental fatigue assessment

Mental fatigue has various influences on daily tasks, so the study of these effects play an important role in assessment of drowsiness, decrease in attention and alertness. In this paper, a novel method for mental fatigue measurement, based on the electroencephalogram (EEG) spectrogram is proposed. This spectrogram is considered as an image. Then by using a morphological approach we assess the effects of mental fatigue on EEG signals. The intensified areas of the spectrogram are considered as direct consequence of the mental fatigue on EEG Alpha activity, which are extracted via Chan-Vese active contour. Furthermore, we estimate a trend in the spectrogram that is assumed the variation of Alpha band frequencies. This variation is considered as another possible outcome of the mental fatigue. Trend estimation is a procedure of data variation analysis over time, which can be done by Gaussian process regression. The results obtained from these two proposed methods show that an increase in the mental fatigue can be a valid reason for a growth in the contoured areas. Moreover, the fatigue can increase the frequencies of Alpha band during the time.