基于多重多尺度熵的孤独症静息态脑电信号分析

Autistic Resting EEG Signals Analysis via Multiple Multi-scale Entropies

面向孤独症儿童脑功能状态评估问题,提出一种多重多尺度熵脑电特征提取算法.算法针对传统多尺度熵信息丢失问题,在移动均值粗粒化基础上,采用延搁取值法构建多个尺度上的多重脑电信号序列,再进一步计算各个尺度的样本熵.算法不仅克服了传统多尺度熵的信息丢失问题,还能充分挖掘脑电信号的细节信息,同时减小了尺度间的波动.基于该算法分析了16名孤独症儿童和16名正常儿童的19个通道的脑电信号.结果表明:正常儿童F7、F8、T4、P3通道的多重多尺度熵和复杂度均高于孤独症儿童,且存在显著性差异(P<0.05).表明前颞叶(F7、F8)可以作为孤独症儿童脑功能状态评估的敏感脑区,T4、P3可以作为辅助干预的敏感通道.

To focus on the assessment of brain functional status of autism spectrum disorders(ASD),an electroencephalogram(EEG) feature extraction algorithm of multiple multi-scale entropies is proposed in this paper.In order to solve the problem of losing EEG information by traditional multi-scales entropy(MSE),moving averaging(MA) coarse graining is done first in the multiple multi-scale entropy algorithm,then multiple scale EEG signals are built using delay value method,before the sample entropy of each scale is calculated.The algorithm not only overcomes the information loss problem of traditional multi-scale entropy,but also fully excavates details of the EEG and reduces fluctuations between the scales.Based on this algorithm,19 channels of EEG signals with 16 autistic children and 16 normal children are analyzed,and the result shows that multiple multi-scale entropies of normal children are higher those of children with autism in channels F7,F8,T4,P3 by a significant difference(P <0.05).It is suggested that the anterior temporal lobe(F7 and F8) should be used as a sensitive brain area for evaluating the brain function of autistic children,and T4 and P3 as sensitive channels for auxiliary and intervention.