基于改进的伪Wigner分布检测癫痫EEG棘波的方法

摘要

A new method for detecting epileptic spikes from EEG signals is presented in this paper. The time-frequency representations of the EEG signals are computed by the modified pseudo Wigner distribution method. The method can reduce or remove the cross terms efficiently and perform fast. With this method the epileptiform transient waves are partitioned from the background EEG by the defined “central frequency” at each interval in the time-frequency domain, and the transient components in the EEG are extracted by a mirror filter. The spikes are detected by the local maxima of the energy of the transient components. The results show that the new method works well in detecting spikes.%本文提出了一种新的检测癫痫EEG棘波的方法。采用改进的伪Wigner分布对脑电信号进行时频变换，这种时频分布不仅能够有效地消除交叉项，而且可以快速实现。在时频平面上，以每个时间点上的中心频率划分背景脑电和癫痫样瞬态特征，通过一个镜像滤波器提取瞬态分量，最后，利用瞬态分量的能量局部极值检测棘波。临床应用表明，这种方法能够快速有效地从背景脑电中提取癫痫样瞬态特征。