摘要
癫痫是大脑神经细胞群超同步放电的一种常见慢性神经疾病,为了更好地对癫痫进行检测,提出了基于EEG的样本熵和深度神经网络(Deep Neural Network,DNN)的方法.将EEG信号用小波变换进行预处理后,以10 s为时间片求出样本熵,实验表明癫痫发作期间样本熵下降,经统计分析样本熵能够和数据集中已标注的标签基本一致,以样本熵下降处的点作为癫痫发作标签对数据集进行深度神经网络学习,能够达到99.5%的检测准确性.
Epilepsy is a common chronic neurological disease in which the neurons in the brain discharge synchronously.In order to detect epilepsy better,a method based on EEG sample entropy and deep neural networks(DNN)is proposed.After the EEG signal is preprocessed by wavelet transform,the sample entropy is obtained by taking 10s as time slice.The experimental results show that the entropy of samples decreases during epileptic attack.After statistical analysis,the entropy of samples can be basically consistent with the labeled labels in the data set.The points where the entropy of samples decreases are used as epileptic attack labels for deep neural networks learning of the data set,which can achieve 99.5%detection accuracy.
作者
卢小杰
唐真言
汪严生
潘媛媛
叶奔
LU Xiao-jie;TANG Zhen-yan;WANG Yan-sheng;PAN Yuan-yuan;YE Ben(School of Medicine Information,Wannan Medical College,Wuhu 241002,China;Research Center of Health Big Data Mining and Applications,Wannan Medical College,Wuhu 241002,China)
出处
《西安文理学院学报（自然科学版）》
2020年第2期54-57,共4页
Journal of Xi’an University(Natural Science Edition)
基金
皖南医学院大学生科研课题项目(WK2019S50):“基于卷积神经网络的癫痫发作预测方法研究”
皖南医学院大学生科研课题项目(WK2018S52):“基于深度学习的单通道脑电图自动睡眠阶段评分模型研究”
安徽高校人文社会科学研究项目(SK2018A0198):“大数据背景下Ⅱ型糖尿病危险人群风险评价与预警机制研究”
皖南医学院中青年科研基金项目(WK201920):“基于LSTM的肝脏系统功能障碍死亡率预测研究”。
关键词
癫痫发作
样本熵
DNN
小波变换
epileptic seizure
sample entropy
deep neural networks(DNN)
wavelet transform
