基于静息态脑电的心因性非癫痫发作的识别算法研究

发布时间：2018-05-13 00:02

本文选题：心因性非癫痫发作 + 癫痫　；参考：《电子科技大学》2014年硕士论文

【摘要】：癫痫(epilepsy,EP),是一种慢性反复发作性短暂功能失调神经性疾病,其发作时会伴随着神经元的异常放电。其主要的诊断依据是发作时脑电是否发生异常放电。心因性非癫痫发作(psychogenic nonepileptic seizures,PNES)的发作症状与癫痫发作时极为相似,只是不伴随异常的脑电放电,视频脑电图(video-EEG,vEEG)结合患者及目击者提供的病史,可作为PNES的诊断标准。即便如此,还是有很多PNES被误诊为癫痫发作,导致延误诊断时间和治疗。已经建立的PNES早期诊断方法需要详细的信息,而且这些信息的收集都非常耗时间。并且,在有些地方尚未开展vEEG,对于有些患者,也难以记录其发作事件过程。如果能发展另外一种早期诊断PNES的方法,这对临床医生将是很有帮助的。近些年,基于图论的分析方法已经被广泛用于对中枢神经系统的解剖和功能网络研究。本文是从静息态脑电的网络拓扑结构层面去寻找PNES和癫痫之间的网络变化,从而提取它们之间有差异的特征来对其进行识别。本文主要完成的内容及研究结果如下:1.采用了图论分析方法对PNES、癫痫和正常组的网络属性进行统计分析,统计结果表明,PNES和癫痫患者的网络属性与正常组相比发生了显著性变化,但PNES组和癫痫组之间并未发现显著性变化,从而PNES和癫痫之间基于网络属性的分类精度很低。2.在功能连接网络的基础上采用空间模式网络方法来提取网络的空间拓扑结构信息作为PNES与癫痫的识别特征,该方法的结果显示,PNES和局灶性癫痫的最高分类性能为:准确率92.00%,敏感性100.00%,特异性80.00%。3.分别将上述分析过程设计成了两个基于Matlab实现的GUI(Graphical User Interface,图形用户界面)模块:网络特征提取模块(包括网络属性特征提取和空间模式网络特征提取)和特征分类模块。本文研究发现,PNES和癫痫之间的功能网络拓扑结构存在着显著性差异,可是网络属性统计分析却并没有显著性地反应出这些差异。从而本文提出了从加权网络的空间拓扑结构中提取具有可识别性的空间信息的空间模式网络方法,来挖掘PNES和癫痫之间隐藏的异常空间结构信息,同时还为临床医生诊断PNES提供了一个参考方法。
[Abstract]:Epilepsy (EP) is a chronic recurrent transient transient dysfunctional neurologic disease, which is accompanied by abnormal discharge of neurons. The main diagnostic basis is the abnormal discharge of electroencephalogram (psychogenic nonepileptic seizures, PNES) and epileptic seizures at the time of seizures. Very similar, only without abnormal electroencephalogram, video-EEG (vEEG) combined with patient and eyewitness history, can be used as a diagnostic criterion for PNES. Even so, many PNES were misdiagnosed as epileptic seizures, resulting in delayed diagnosis and treatment. The early diagnostic methods of PNES need detailed information. And the collection of these information is very time-consuming. And, in some places, it is not possible to carry out vEEG, and for some patients it is difficult to record the process of the attack. It will be helpful for the clinicians if we can develop another method of early diagnosis of PNES. In recent years, the analysis based on graph theory has been widely used in the middle of the world. The anatomical and functional network of the armature nervous system. This paper is to find the network changes between PNES and epilepsy from the network topology of resting state EEG, and extract the differences between them to identify them. The main contents and research results of this paper are as follows: 1. using the graph theory analysis method to PNES, epilepsy. Statistical analysis of the network properties of epilepsy and normal groups showed that the network properties of PNES and epileptic patients had significant changes compared with the normal group, but there was no significant change between the PNES group and the epileptic group, and the classification accuracy of PNES and epilepsy based on network attributes was very low.2. in the basis of the functional connection network. The spatial pattern network method is used to extract the spatial topology information of the network as the recognition feature of PNES and epilepsy. The results show that the highest classification performance of PNES and focal epilepsy is: accuracy rate 92%, sensitivity 100%, and specific 80.00%.3. design two Matlab implementation based on the previous analysis process, respectively. GUI (Graphical User Interface, graphical user interface) module: network feature extraction module (including network attribute feature extraction and spatial pattern network feature extraction) and feature classification module. In this paper, it is found that there are significant differences in functional network topology between PNES and epilepsy, but the statistical analysis of network attributes is not In this paper, we propose a spatial pattern network method to extract the identifiable spatial information from the spatial topology of the weighted network, to excavate the abnormal spatial structure information hidden between PNES and epilepsy, and also provide a reference method for clinicians to diagnose PNES.

【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R742.1;TP301.6

【相似文献】