A New Statistical Model for Timely-detection of Dynamical Change in EEG Signals

Abstract

Detecting dynamical change in EEG signals is essential and popular for wide range of clinic applications. To achieve this result, this paper presents a new statistical model for timely-detection of EEG signals. The new statistical model combines wavelet packet decomposition (WPD) with graph modeling. After a given original EEG signal is filtered by a sub-band pass filter. WPD is firstly adopted to the collected EEG signal and the decomposed component as inputs to construct the graph. Adaptive weighting fusion is used to calculate the anomaly scores to characterize the dynamical change of filtered EEG signal. A common hypothesis testing is finally employed to inspect whether an EEG change occurs or not during a monitoring period. The proposed method is applied to real EEG signals to validate its effectiveness. Experimental results show that this method has broad application prospects in practical usages.