Noncerebral waves detection from frontal brain electrical activity using the quantitative measure of nongaussianity

The aim of this paper was to separate the EEG recordings into cerebral and noncerebral waves and compare a statistical properties of chosen components using the coefficient of excess kurtosis. Noncerebral waves, particularly the ocular artifacts, should be properly identified, because some of them imitate the cerebral potentials. The eye opening and closure, blinks or eye flutter are similar to the frontal intermittent rhythmic delta activity (FRIDA), which has been associated with brain disorders e.g., encephalopathies. For separation of the EEG data the infomax rule with a new nonlinearity was used. The accuracy of separation of the EEG´s has been measured using the Performance Index.