【作者】 孙波；

【导师】 尧德中；

【作者基本信息】 电子科技大学 ， 工程硕士（专业学位）， 2020， 硕士

【摘要】 特发性全面性癫痫(Idiopathic Generalized Epilepsy,IGE)是一种以全面性发作为发作类型的癫痫综合征,IGE患者常规的影像学检查常提示无明显癫痫病灶,脑电图放电频率通常为3-6HZ的慢波放电,癫痫放电会对患者的注意力、工作记忆以及执行能力造成不同程度的损伤。同步脑电磁共振研究(EEG-fMRI)可以无创地探测脑电事件与血液动力学的相关,将生理活动和功能定位相结合来探究癫痫放电对大脑功能的影响。为了考查IGE患者的EEG和fMRI两个模态脑网络及其网络共变,我们选取79名IGE患者和60名健康受试者采集同步EEG-fMRI的数据。静息态下,将同步记录的EEG数据反演到皮层并分成七个脑电网络,研究选取theta与alpha两个低频段,基于动态时间窗构建脑电全脑皮层网络的连接矩阵,使用K-means聚类方法对所有时间窗进行聚类分析,fMRI数据计算了标准化的低频振幅(amplitude of low-frequency fluctuation,ALFF),分析并对比了癫痫被试和健康被试mALFF图谱中响应脑电聚类状态连接模式的脑区。研究结果发现脑电的额顶网络(frontoparietal network,FPN)和默认网络(default mode network,DMN),额顶网络和感觉运动网络(sensorimotor network,SMN)在两个低频段都表现出较强的连接性。IGE被试两个模态间高度共变的区域集中在海马、额叶和顶叶等脑区,这些区域位于以往癫痫网络研究中的关键节点,提示癫痫放电活动影响了EEG和fMRI间的耦合性。基于工作记忆任务(working memory,WM)的研究中,我们选取研究对象中记录了任务态影像数据的32名GTCS亚型和31名健康受试者,通过癫痫和健康被试的fMRI任务态激活图谱选取感兴趣区域,利用生理心理交互分析(psychophysiological interaction,PPI)建立任务态有向功能连接网络。研究计算了脑电额叶theta频段功率,分析了PPI网络的差异以及脑电额叶theta功率与PPI网络节点信息流间的相关。研究发现工作记忆任务下GTCS患者与健康被试相比,DMN网络去激活程度降低,网络中楔前叶、额叶和后扣带回与枕叶、小脑以及梭状回的功能连接降低,GTCS患者工作记忆任务中额叶的信息输出和脑电theta频段的功率呈负相关,表明癫痫患者DMN功能异常,DMN与其他脑区的功能连接也存在改变,额叶进行WM任务时需要更多能量,这都提示癫痫放电对大脑正常注意状态的损伤。本文基于同步EEG-fMRI融合探究了IGE患者和健康受试者的大脑网络特征和两个模态的共变,结果表明IGE患者在静息态、任务态下都存在脑网络异常以及EEG和fMRI两个模态间的耦合异常,这项研究提供了有效的多模态共变分析方法以进一步理解癫痫疾病的大脑网络损伤。更多还原

【Abstract】 Idiopathic Generalized Epilepsy(IGE)is an epilepsy syndrome with generalized seizures.Conventional imaging methods applying in patients with IGE are difficult to detect epileptic lesions,the typical EEG characteristics of IGE is 3-4Hz generalized spike-wave discharges.The epilepsy discharges may cause different degree of damage to the patient in attention,working memory and execution ability.Simultaneous EEG-fMRI fusion can noninvasively detect the correlation between EEG events and hemodynamics,and combine physiological activities and functional localization to explore the effect of epilepsy discharges on brain function.To examine the EEG and fMRI brain networks and the co-variation of the two modal,we obtained synchronous EEG-fMRI resting state and working memory task state fMRI data from 79 subjects with IGE and 60 healthy subjects.In the resting state,the EEG signals were inverted to the cortex and divided into seven networks to construct connection matrices based on dynamic time windows.The study was performed in two low frequency bands,including theta and alpha.We applied K-means clustering method to perform cluster analysis on all the time windows to obtain the strong relationship between EEG pattern and fMRI maps.Our results showed that the connectivity in frontoparietal networks(FPN)and default mode network(DMN),frontoparietal networks and sensorimotor networks(SMN)were both strongly in two slow-bands.The regions with high co-variance between the two modalities of IGE are concentrated in the hippocampus,frontal lobe,parietal lobe,and thalamus which are located in key nodes in previous epileptic network studies,suggesting that epilepsy discharges activity may affect the coupling between EEG and fMRI.For the working memory task state,we obtained the task state fMRI data from 32 subjects with GTCS subtypes and 31 healthy subjects who have performed working memory tasks.To identify task-specific changes in the relationship between brain areas,the functional connection network under the working memory task(WM)was constructed with psychophysiological interaction(PPI)analysis.We selected regions of interest of PPI analysis based on the fMRI task activation maps of IGE and healthy subjects.As the functional network is directed,in addition to network-based analysis,we also analyzed the information flow of each node in the network.The frontal lobe theta is considered to be a symbol of WM.We select the electrode located in the frontal lobe to calculate the power of the theta band,and analyzed the correlation between theta power and the information flow direction of the PPI network nodes.The study found that compared with healthy subjects,GTCS patients had a lower degree of deactivation of the DMN during working memory tasks,and the functional connections of the precuneus,frontal lobe and posterior cingulate gyrus with the occipital lobe,cerebellum,and fusiform gyrus were reduced.The information flow-output of the frontal lobe in the working memory task of the patient is negatively correlated with the power of theta band,indicating that not only the internal function of the DMN is abnormal in patients with epilepsy,but also the functional connection with other brain areas is also altered,which suggested that epilepsy discharges damage the normal attention state.Our work has studied the brain network characteristics and co-variation of two modalities in IGE patients with epilepsy and healthy subjects based on synchronous EEG-fMRI fusion.The results show that IGE patients have network abnormity and coupling abnormity between EEG and fMRI both in the resting state and task state,this study provides an effective multi-modal co-variation analysis method to reveal the brain network damage of epilepsy disease and the characteristics of both EEG and fMRI in healthy human brain.更多还原