An Adaptive Automatic EEG Signal Segmentation Method Based on Generalized Likelihood Ratio

It is often needed to label electroencephalogram (EEG) signals by segments of similar characteristics that are particularly meaningful to clinicians and for assessment by neurophysiologists. Within each segment, the signals are considered statistically stationary, usually with similar characteristics such as amplitude and/or frequency. In order to detect the segments boundaries of a signal, we propose a method using time-varying autoregressive (TVAR) model, integral, and basic generalized likelihood ratio (GLR). Since autoregressive (AR) model for the GLR method is valid for only stationary signals, TVAR as a valuable and powerful tool for non-stationary signals is suggested. Moreover, to improve the performance of the basic GLR and increase its speed, we propose to use moving steps more than one sample for successive windows in the basic GLR method. By using synthetic and real EEG data, the proposed method is compared with the conventional ones, i.e. the GLR and wavelet GLR (WGLR). The simulation results indicate the absolute advantages of the proposed method.