Mu-suppression detection in motor imagery electroencephalographic signals using the generalized extreme value distribution
This work addresses a domain-specific problem in EEG-based BCI systems, but it appears incremental as it builds on existing statistical models and classifiers.
The paper tackled the problem of detecting mu-suppression in EEG signals for brain-computer interfaces by proposing an algorithm based on the generalized extreme value distribution and a linear classifier, achieving very good classification accuracy for imagery, movement, and resting events.
This paper deals with the detection of mu-suppression from electroencephalographic (EEG) signals in brain-computer interface (BCI). For this purpose, an efficient algorithm is proposed based on a statistical model and a linear classifier. Precisely, the generalized extreme value distribution (GEV) is proposed to represent the power spectrum density of the EEG signal in the central motor cortex. The associated three parameters are estimated using the maximum likelihood method. Based on these parameters, a simple and efficient linear classifier was designed to classify three types of events: imagery, movement, and resting. Preliminary results show that the proposed statistical model can be used in order to detect precisely the mu-suppression and distinguish different EEG events, with very good classification accuracy.