Time - Frequency Analysis And Its Application In Eeg Analysis

Epilepsy is a kind of phenomenon caused by the abnormal discharge of brain nerve cells results in the brain functional disorder. The waves of Epileptic discharge include spikes, sharp waves, spike-and-slow waves, sharp-and-slow waves etc. The most important thing is to detect the existence of spikes in a clinical examination. It is a complicated and time consuming work to check EEG waves and detect the Epileptic waves visually. It also needs professionally trained specialists. The automatic spike detection in EEG is significant for both detecting the Epileptic waves and reducing the heavy work of doctors. Since EEG signal is non-stationary and with high background noise, traditional signal analysis methods, such as time domain and frequency domain analysis, become inaccurate. With the development of science and technology, especially a great progress of non-stationary signal theory, EEG signal can be analyzed with new methods. Time-frequency distribution is one of the great progresses. The objective of this thesis is to extract the spikes automatically with improved time-frequency distribution (TFD) methods.The contents of this thesis are as follows:First, the basic theory, the history and the development of time-frequency analysis are reviewed, as well as the application of existent time-frequency methods in EEG.Second, the basic theory of wavelet is studied, especially B-spline wavelet which does well in detecting break point. The common method to detect break point is to smooth thesignal with a smoothing function first. According to the 1st and the 2nd order differential coefficients, we can find signal's break points. When the wavelet function chosen has a relation with smoothing function, we can detect the break point based on the extremum of wavelet coefficients. This thesis believes that B-spline wavelet when J=5 is suitable for the application.Third, RID does well in reducing the troublesome interference terms. After applying RID into the designed reference signal and EEG signal, this thesis detects the spike by evaluating the correlation. The validity of the method is verified through experiments.Finally, my lab developed a new type video electroencephalograph and holter, supported by Liaoning Educational Committee and the "life+X" project of Dalian University of Technology. The goal of this project is to expand the functions of existent electroencephalograph and to improve its performance, according to the theory and the newest progress of digital signal processing technology. This thesis puts three time-frequency methods into new type video frequency electroencephalograph, whose spike detection rate reaches 90%.