Model Order Sensitivity of Burg Method for EEG Diagnosis

Abstract

In the present study, eyes open and eyes closed EEG records of healthy volunteersa and intracrinal EEG records of epilepsy patiants are analysed by the Burg Method. The results of power spectral densities shows that clinical long EEG data correspondes with high order autoregressive (AR) model. The higher frequency resolution is obtained as empirically selected model order increases. High order for AR modeling provides us to more sensitive diagnose in EEG analysis.