基于多分辨分析和近似熵的脑电癫痫波自动检测

AUTOMATIC DETECTION OF EPILEPTIFORM ACTIVITY IN EEG BASED ON MULTI-RESOLUTION AND APPROXIMATE ENTROPY

脑电癫痫特征的自动提取在临床应用上具有重要意义。分析了小波多分辨分析和近似熵特征提取的特点,提出了8通道脑电信号癫痫波的检测方法。首先每个通道的信号利用小波变换进行5层分解,然后对分解的细节信号作近似熵计算,发现含有癫痫活动的脑电信号与正常脑电有显著的区别,最后利用Neyman-Pearson准则进行检验比较。实验结果表明,在一定误检率下,检测率最高的是在第一层,而且这种方法保证了检测系统具有较小的误检率和较高的检测率。

The automatic extraction of epileptiform activity feature in EEG is significant to clinical application. A new scheme is presented for detecting epileptiform activity in 8-channel EEG data. The scheme is based on the analysis of multi-resolution and characteristic extraction of approximate entropy（ApEn） of EEG signals. First, the EEG signals on each channel are decomposed up to five levels using discrete wavelet transform, and then detailed coefficients obtained are performed approximate entropy computation, distinct differences are found between the ApEn values of the epileptic and the normal EEG. At last, EEG signals are compared with Neyman-Pearson criteria. Experimental results show that the highest detection rate is at sub-band level 1 when the false detection rate is a certain one,and this scheme assures the detection system has higher detection rate with lower false detection rate.