Detection of ocular artifacts in EEG data using the Hurst exponent

The human brain activity is composed of cerebral waves described by values of frequency or amplitudes and many noncerebral potentials. The most common physiological potentials in EEG data are ocular artifacts, i.e. eye blinks. In clinical practice, detection of this noncerebral wave is very important, because of similarity to the brain components associated with some disorders e.g., encephalopathies. In the case of low voltage EEG data, an estimate of some ocular artifacts, in particular eye blinks, seems to be an impossibility. For that reason, the Hurst exponent to the analysis of the eye blink artifacts is proposed. For separation of the EEG signals, the infomax algorithm was used. In order to compare the results of detecting eye blinks, thirty healthy subjects (15 males, 15 females, age range: 20-60 years) and ten patients (5 males, 5 females, age range: 24-63 years) with suboccipital lesions of the cerebellum have been examined. All EEG data have been recorded without using EOG additive channels.