基于小波变换和样本熵的脑电信号癫痫特征提取

EEG Epileptic Feature Extraction Based on Wavelet Transform and Sample Entropy

癫痫发作时的脑电信号(EEG)中含有大量的癫痫特征波信息,特征波形是确定癫痫是否发作的重要依据。小波变换可以对信号进行多分辨率分解,将信号分解成不同频率的含有特征信息的细节信号。癫痫发作时,脑电信号的复杂性会降低,样本熵可以表示信号系统的复杂程度。论文采用小波变换(wavelet transform)和样本熵(sample entropy)结合的分析方法,首先运用小波变换分别对正常状态、癫痫发作间期和发作期的脑电信号进行多分辨率分解,提取出含有特征信息的子频带,然后对子频带求样本熵。实验结果表明,此方法可以有效地提取出含有尖波(sharp wave)、棘波(spike wave)等癫痫特征信息。

EEG contains a large number of epileptic characteristic wave in epileptic seizures,the characteristic waveform is an important basis for determining whether epileptic seizures occur. The wavelet transform can decompose the signal with multi-resolution and decompose the signal into detailed signals with different frequency and characteristic information. The complexity of EEG signals will be reduced during epileptic seizures,sample entropy can indicate the complexity of the signal system. In this paper,a method combining wavelet transform with sample entropy is applied. First,wavelet transform is used to decompose the EEG signal in normal state,interictal interval and attack phase,then the sample entropy is calculated for the subband. Experimental results show that this method can effectively extract spike and spike characteristics.