ABSTRACT
By separating the EEG data into dierent frequency components, it was reported that CD and Lmax can eectively dierentiate data obtained from healthy subjects and epileptic patents during the interictal versus ictal phase when the signals were sectioned into specic frequency subbands (Adeli et al. 2007). e use of a generic algorithm for the detection of seizures from the 8 channels of electrocorticogram from 14 subjects was proposed (Osorio et al. 2002). e analysis was performed o- line using a band-pass wavelet lter in the 8-42-Hz range, and the relative median power of the signals demonstrated that a subject-dependent seizure detection approach results in less than 0.16 false alarms per hour with a delay of detection of less than 5 seconds.
