基于近似熵快速算法的静息态脑磁信号分析

摘要

In order to study the nonlinear dynamics of the schizophrenic patient's MEG signals in resting-state,this paper presents a method of feature extraction which combined the wavelet variation with the approximate entropy.The brain magnetic signals of 10 controls and 10 patients are decomposed to six levels by wavelet decomposition and wavelet coefficient is extracted corresponding to the θ rhythm and a rhythm of MEG signals.Then the distribution of approximate entropy between two kinds of people are calculated and compared.The experiment results show that the entropy of each brain region and channel of the MEG signals in schizophrenic patients were generally higher than controls under the same situation,especially frontal and central regions in α rhythm.This result provides a guideline for the study of EEG signal characteristics of the patients and establishes the appropriate classification diagnostic model.%为了探究静息态精神分裂症患者脑磁信号的非线性动力学特性,提出了一种将小波变换和近似熵相结合的特征提取方法.该方法首先通过小波变换,将10个正常人和10个精神分裂症患者的脑磁信号进行6层小波分解,提取对应于脑磁信号θ波段和α波段的小波系数,继而计算和比较两类人近似熵的分布情况.实验结果表明,相同情况下精神分裂症患者MEG信号的各脑区和各通道间的近似熵都普遍高于正常人,α波段的额叶和中央区域尤为突出.该结果为进一步研究患者MEG信号特征进而建立相应的分类诊断模型提供了思路.