基于静息态功能磁共振成像的精神分裂症脑网络特征分类研究

如何从复杂的静息态功能核磁共振成像(rs-fMRI)数据中提取高鉴别性特征,是提升精神分裂症识别精度的关键。本文使用一种加权稀疏脑网络构建方法,采用肯德尔相关系数(KCC)从脑网络中提取连接特征,并基于线性支持向量机对57例精神分裂症患者与64例健康受试者进行分类研究,最终得到了较高的分类精度(81.82%)。本文研究结果表明,相较于传统的皮尔逊相关和基于稀疏表示的脑网络构建方法,以及常用的双样本t检验(t-test)和最小绝对收缩与选择算子(Lasso)特征选择方法,本文提出的算法可以更有效地提取出能够区分精神分裂症患者与健康人群的脑功能网络连接特征,进而提升分类精度;同时本研究中所提取的鉴别性连接特征或可作为潜在的临床生物学标志物,用以辅助精神分裂症的诊断。     

基于脑电的精神疾病自动分类研究进展

临床上精神疾病的诊断大多依赖于精神科医生的主观评价，缺少客观有效的生物学指标。脑电信号分析与机器学习方法相结合，在精神疾病辅助诊断领域的应用发展迅速，采用传统机器学习算法和深度学习算法，对脑电信号特征进行学习，从而实现精神疾病的分类研究。文章介绍了脑电信号的基本概念、处理流程及其常用特征，笔者总结了脑电信号在抑郁症、双相情感障碍、精神分裂症等精神疾病自动分类的研究进展，最后展望了机器学习方法在该领域的发展趋势。采用传统的机器学习方法仍然是用于自动分类主流的工具。但深度学习内部复杂的体系结构及训练过程阻碍了对其内部的理解，难以解释其在医学领域的作用，因此深度学习也是脑电研究发展方向之一。此外，单独对脑电图进行分析难以囊括患者所有的特征，需与其他模态的生理参数结合进行多生理参数融合分析，使得疾病诊断更加智能化。     

基于异方差混合转移分布模型和支持向量机的脑电分类研究

为解决线性分析和单一非线性动力学指标方法无法准确描述脑电信号的问题,本研究提出基于异方差混合转移分布模型脑电特征提取方法。首先对采集到的脑电信号依据条件期望最大化(ECM)算法建立异方差混合转移分布模型,求得模型条件方差序列的均值及方差作为脑电信号的特征,将得到的脑电信号特征采用支持向量机进行分类。通过对6个人的正常脑电信号和带有眼电伪迹脑电信号进行分类仿真实验,其结果表明该方法能很好地拟合出脑电信号,且分类精确度能达到99.166 7%,说明此方法可有效提取脑电特征并准确识别出眼电伪迹。     

意念力游戏训练改善脑认知功能状态研究

本文采用意念力游戏训练对轻度认知功能障碍人群进行干预治疗,辅助改善其脑认知功能状态。研究共采集了40名受试者训练前及两次训练后的脑电数据,并分析了脑电信号的连续复杂度特征,评估脑认知功能状态,探讨意念力游戏训练对脑认知功能状态的改善效果。结果显示,经过两次训练后,受试者脑电信号连续复杂度有所升高(0.012 44±0.000 29,P 0.05),曲线波动幅度逐渐减小,表明随训练次数增加,脑电信号连续复杂度显著提升,脑认知功能有明显改善且状态稳定。本文的研究结果可以表明,意念力游戏训练可以改善脑认知功能状态,该结果或可为今后辅助干预脑认知功能障碍提供支持与帮助。     

基于多核学习方法的前瞻性记忆脑电识别

目的:分别提取急性缺血性卒中患者中认知功能正常和轻度障碍的病人进行前瞻性记忆实验时的脑电信号特征,利用多核学习方法对特征进行分类,获得较好的识别效果,验证选用数据特征以及分类方法应用于脑电信号分析时的有效性。方法:首先设计了一种前瞻性记忆实验,然后将认知功能正常和轻度障碍的急性缺血性卒中病人完成前瞻性记忆实验时的脑电信号进行采集、预处理并提取出样本熵、近似熵和AR系数三种数据特征,最后将所提取的数据特征送入多核学习分类器进行分类识别,通过迭代学习,获得较高的分类精度。结果:选用样本熵、近似熵和AR系数作为脑电信号特征进行分类时,均获得了较高的分类精度,最高可达90.9%。其中,使用样本熵特征作数据特征时,更易获得较高的分类精度。结论:前瞻性记忆实验可用于认知功能障碍识别研究,选用样本熵、近似熵、AR系数和多核学习方法作为数据特征和分类器可以取得较好的分类结果,为认知功能障碍的定量识别提供了有效的方法。     

小波熵在脑电信号分析中的应用

小波熵是一个衡量非线性信号多尺度动力学行为有序、无序程度的量化指标,其可提供信号非线性动力学过程复杂程度的信息.近年来,小波熵在脑电信号中的研究日益受到关注,国内外学者用小波熵研究脑电信号、诱发电位、事件相关电位等的复杂程度,进一步揭示了大脑电活动的动力学机制.其主要应用于大脑感知、认知活动的研究,癫痫脑电信号的动态观测,睡眠、网络成瘾、头外伤后脑神经的康复等几个方面.小波熵不仅可以显示受到刺激后脑电信号频率上同步化的动态演变过程,而且可以有效区分癫痫发作前状态和癫痫发作状态,从而加深了对脑动力学机制的理解,成为认知功能研究的一种新的方法,显示了在脑电信号分析中良好的应用前景.     

精神分裂症患者脑电信号多重分形的异常

精神分裂症患者的脑电信号 (EEG)在线性动力学分析方法中并未出现典型的异常现象 ,利用非线性动力学研究方法分析EEG的多重分形特性 ,计算精神分裂症患者的 16导EEG的奇异谱曲线 ,发现精神分裂症患者的各导EEG信号中的奇异强度分布范围Δα都比健康人有显著的增大 ,通过t检验 ,在Fp2、F3、C4、P3、O2、F7、T4、T5区的P值小于 0 .0 5 ,具有明显的变化 ,这种异常现象可以成为精神分裂症患者EEG信号诊断的参考依据。     

基于通道注意力和稀疏时频分解的运动想象分类EI北大核心CSCD

运动想象脑电信号是低信噪比的非平稳时间序列,单通道脑电分析方法难以有效刻画多通道信号之间的交互特征。本文提出了一种基于多通道注意力的深度学习网络模型,该模型对预处理后的数据进行稀疏时频分解,增强了脑电信号时频特征的差异性。然后利用注意力模块在时间和空间对数据进行注意力映射,让模型可以充分利用脑电信号不同通道的数据特征。最后利用改进的时间卷积网络进行特征融合并进行分类识别。利用BCI competition Ⅳ-2a数据集对所提算法进行验证,结果表明所提算法可有效提升运动想象脑电信号的分类正确率,9名受试者的平均识别率为83.03%,与现有方法相比,提高了脑电信号的分类精度。所提方法增强了不同运动想象脑电数据之间的差异特征,对提升分类器性能的研究具有重要意义。     

基于残差注意力神经网络模型的癫痫脑电信号分类

目的 针对癫痫脑电信号特征提取过程复杂、信息提取不充分及分类精度较低等问题，本文提出一种基于残差注意力神经网络模型（residual attention module neural network,RAM-Net）用于实现癫痫脑电信号的自动分类。方法 首先对脑电信号进行去噪和分段处理，使网络更有效提取细节特征；然后根据脑电信号在时频域幅值特点，将信号转换为二维时频图像作为模型输入；最后借鉴残差网络思想，在每个残差块中融合注意力机制，构建分类模型，在临床数据集上做验证。结果 该方法分类准确率为97.16%，精确率为97.00%，可实现癫痫发作、间期和正常状态的脑电信号三分类。结论 基于RAMNet的癫痫脑电信号分类方法将脑电信号转化为二维图像，降低了方法复杂度；融合注意力机制增强了网络的有效信息提取能力，可为癫痫临床辅助诊断提供一种新的分析思路和处理方法。     

脑电图微状态分析及应用研究进展EI北大核心

脑电信号具有高时间分辨率的特征,各类脑电信号分析方法近年来发展迅速。脑电微状态分析方法能够研究毫秒级范围内的大脑变化,同时也可呈现脑电信号在拓扑层面上的分布,从而反映全脑的不连续和非线性特征。经历三十多年的丰富和完善,脑电微状态分析已经渗透到脑科学相关的多个研究领域。本文总结了脑电微状态分析方法的基本原理,系统阐述了微状态特征参数改变、微状态与脑功能网络的关系以及微状态特征提取与分类在脑疾病和脑认知方面的主要应用进展,期望能够为该领域的研究人员提供一定的参考。     

EEG-based functional networks in schizophrenia

Schizophrenia is often considered as a dysconnection syndrome in which, abnormal interactions between large-scale functional brain networks result in cognitive and perceptual deficits. In this article we apply the graph theoretic measures to brain functional networks based on the resting EEGs of fourteen schizophrenic patients in comparison with those of fourteen matched control subjects. The networks were extracted from common-average-referenced EEG time-series through partial and unpartial cross-correlation methods. Unpartial correlation detects functional connectivity based on direct and/or indirect links, while partial correlation allows one to ignore indirect links. We quantified the network properties with the graph metrics, including mall-worldness, vulnerability, modularity, assortativity, and synchronizability. The schizophrenic patients showed method-specific and frequency-specific changes especially pronounced for modularity, assortativity, and synchronizability measures. However, the differences between schizophrenia patients and normal controls in terms of graph theory metrics were stronger for the unpartial correlation method.     

Chaos in schizophrenia associations, reality or metaphor?

There is evidence that schizophrenic associations display “chaotic”, random-like behavior and decreased predictability. The evidence suggests a hypothesis that the “chaotic” mental disorganization could be explained within the concept of nonlinear dynamics and complexity in the brain that may cause chaotic neural organization. Testing of the hypothesis in the present study was performed using nonlinear analysis of bilateral electrodermal activity (EDA) during resting state and an association test in 56 schizophrenic patients and 44 healthy participants. EDA is a suitable measure of brain and autonomic activity reflecting neurobiological changes in schizophrenia that may indicate changes in nonlinear neural dynamics related to associative process. The results show that quantitative indices of chaotic dynamics (the largest Lyapunov exponents) calculated from EDA signals recorded during rest and the association test are significantly higher in schizophrenia patients than in the control group and increase during the test in comparison to the resting state. The difference was confirmed by statistical methods and using surrogate data testing that rejected an explanation within the linear statistical framework. The results provide supportive evidence that pseudo-randomness of schizophrenic associations and less predictability could be linked to increased complexity of nonlinear neural dynamics, although certain limitations in data interpretation must be taken into account.     

The nonlinear dynamical analysis of the EEG in schizophrenia with temporal and spatial embedding dimension

We applied nonlinear dynamics theory to EEG analysis of schizophrenic patients and estimated the correlation dimension with both temporal embedding and spatial embedding. A higher D₂ was found when using a time-delay embedding method. Especially at F₇ and Fp₁, a significant increase showed. We concluded that more complex activity occurred in certain lobes of schizophrenic patients. Using the spatial embedding method, a relative lower global correlation dimension was obtained. This shows that there might be a diffuse slow wave activity through a schizophrene's global cerebrum. Finally, we discuss the study from three angles of clinical semiology, spectrum analysis and neuropsychology and draw some conclusions about the relationship between the nonlinear analysis of schizophrenia EEG and clinical research. It seems that the theory of a nonlinear dynamics system is a powerful tool for EEG research and may prove useful in complementing visual analysis of EEG accompanied with other study means for brain electrical activity.     

The impact of cognitive training on spontaneous gamma oscillations in schizophrenia

Schizophrenia patients exhibit less gamma‐frequency EEG/MEG activity (>30 Hz), a finding interpreted as evidence of poor temporal neural organization and functional network communication. Research has shown that neuroplasticity‐oriented training can improve task‐related oscillatory dynamics, indicating some reorganization capacity in schizophrenia. Demonstrating a generalization of such task training effects to spontaneous oscillations at rest would not only enrich understanding of this neuroplastic potential but inform the interpretation of spontaneous gamma oscillations in the service of normal cognitive function. In the present study, neuromagnetic resting‐state oscillatory brain activity and cognitive performance were assessed before and after training in 61 schizophrenia patients, who were randomly assigned to 4 weeks of neuroplasticity‐oriented targeted cognitive training or treatment as usual (TAU). Gamma power of 40–90 Hz increased after training, but not after TAU, in a frontoparietal network. Across two types of training, this increase was related to improved cognitive test performance. These results indicate that abnormal oscillatory dynamics in schizophrenia patients manifested in spontaneous gamma activity can be changed with neuroplasticity‐oriented training parallel to cognitive performance.     

Alpha reactivity in schizophrenia and in schizophrenic spectrum disorders: demographic, clinical and hemispheric assessment

Alpha EEG reactivity was assessed in a carefully diagnosed sample of 84 schizophrenic and schizophrenic spectrum disorder patients, both under resting conditions (eyes closed and eyes open) and during two spatial-geometric cognitive tasks. The influence of the subject's demographic (sex and age), clinical (diagnostic subtypes, disease course, CT scan characteristics) and neurophysiological (hemispheric recording and different cognitive tasks) characteristics on alpha peak reactivity was analyzed by means of multivariate analysis of variance. The results indicated a significant effect of type of illness on alpha EEG reactivity, patients with a diagnosis of undifferentiated and disorganized schizophrenia having the lowest alpha reactivity levels. None of the other variables considered had any contributing effect. The results are discussed in terms of orienting responses and hemispheric CNS organization in functional psychoses.     

EEG phase synchronization in patients with paranoid schizophrenia

Recent findings suggest that specific deficits in neural synchrony and binding may underlie cognitive disturbances in schizophrenia and that key aspects of schizophrenia pathology involve discoordination and disconnection of distributed processes in multiple cortical areas associated with cognitive deficits. In the present study we aimed to investigate the underlying cortical mechanism of disturbed frontal-temporal-central-parietal connectivity in schizophrenia by examination of the synchronization patterns using wavelet phase synchronization index and coherence between all defined couples of 8 EEG signals recorded at different cortical sites in its relationship to positive and negative symptoms of schizophrenia. 31 adult schizophrenic outpatients with diagnosis of paranoid schizophrenia (mean age 27.4) were assessed in the study. The obtained results present the first quantitative evidence indicating direct relationship between wavelet phase synchronization and coherence in pairs of EEG signals recorded from frontal, temporal, central and parietal brain areas and positive and negative symptoms of schizophrenia. The performed analysis demonstrates that the level of phase synchronization and coherence in some pairs of EEG signals is inversely related to positive symptoms, negative symptoms and general psychopathology in temporal scales (frequency ranges) given by wavelet frequencies (WFs) equal to or higher than 7.56 Hz, and positively related to negative symptoms in wavelet frequencies equal to or lower than 5.35 Hz. This finding suggests that higher and lower frequencies may play a specific role in binding and connectivity and may be related to decreased or increased synchrony with specific manifestation in cognitive deficits of schizophrenia.     

Epilepsy identification based on EEG signal using RQA method

PurposeEpilepsy is one of the most common neurological diseases and its cause is not unequivocal. Thus, additional methods and searches that may help to diagnose the disease are used in the clinical practice. In this study, we tested the possibility of using the Recurrence Quantification Analysis (RQA) method to identify epilepsy and present the analysis of EEG signals of healthy patients and epileptic patients by the RQA method.Materials/methodsThe recordings of signals belong to 13 patients, which were divided into 2 groups: Group A (5 epileptic patients) and Group B (8 healthy patients). In this study Fp1, Fp2, T3 and T4 electrodes were considered in the analysis using the RQA method.ResultsIt is difficult to explore the dynamics of signals by linear methods. In this study, another way of analyzing the dynamics of signals by the RQA method is presented. The RQA method revealed differences in the dynamics between the epileptic and normal signals, which seemed important in an organoleptic way. It was found that the dynamics of epileptic signals is more periodic than normal signals. To confirm the correctness of the statements issued for the RQA data the Principal Component Analysis mapping was applied. This method showed more clearly the differences in the dynamics of both signals.ConclusionsThe RQA method can be used to identify nonlinear biomedical signals such as EEG signals.     

精神分裂症和抑郁症患者静息态脑电功率谱熵的对照研究

研究精神分裂症患者(SC)和抑郁症患者(DP)的静息态脑电功率谱熵,深入比较该指标在两种疾病人群中的表现,探究这一指标对两种疾病大脑异常情况的反映。 选择性别、年龄相匹配的精神分裂症、抑郁症患者各100例(男:50,女:50),采集睁眼、闭眼两种状态下的静息态脑电数据;对脑电信号进行信号预处理,并通过独立成分分析实现伪迹校正;基于Welch变换进行功率谱分析,归一化后利用相对功率计算脑电的功率谱熵;采用t检验、方差分析等统计手段,对该指标做统计分析。 结果表明, 在任意相同状态下,精神分裂症组的组平均功率谱熵在每一导联上都低于抑郁症患者组(导联平均功率谱熵:闭眼状态下,SC:1.26; DP:1.32;睁眼状态下,SC:1.33; DP:1.37),且在多数导联上差异显著(P<0.05);对于两组被试,其闭眼状态下的功率谱熵均分别低于睁眼状态下的功率谱熵;对于功率谱熵从睁眼状态至闭眼状态的减少量,两组被试在Fp1、Fp2导联存在显著差异(P<0.05)(Fp1:SC为0.08,DP为0.02;Fp2:SC为0.09,DP为0.02);在睁眼状态下,精神分裂症与抑郁症患者大脑左右半球功率谱熵的不对称性存在差异,精神分裂症组表现出更广的不对称性(呈显著不对称性的电极对:SC有4对,包括F3-F4、O1-O2、F7-F8、T5-T6;DP有2对,包括P3-P4、F7-F8);闭眼状态下二者的不对称性均只显著体现在F7-F8、T5-T6导联(P<0.05)。 功率谱熵这一指标能够敏感直观地描述功率谱的分布情况与不规则程度,进而反映脑电信号的复杂程度以及个体大脑活跃程度。该指标能够作为一项有效参考,反映两种疾病人群在静息态下大脑活动的差异,并有望用于区分精神分裂症与抑郁症。     

自发脑电Alpha节律的非线性动力学特性研究

针对正常对照组和精神分裂症病人自发脑电．从三个角度获取了alpha节律波．并对其非线性进行了检验。结果表明，无论是对照组还是病人组的alpha节律都不具有典型的非线性特征，提示对脑电节律的研究不能盲目采用非线性动力学中的相关理论t要将传统临床研究和信号处理领域有机地结合起来，才能得出有实际意义的结论。     

A comparative study of magnetic resonance imaging on the gray matter and resting‐state function in prodromal and first‐episode schizophrenia

It is very difficult to predict the future development possibility of schizophrenia through the clinical symptoms of the high‐risk cases. Therefore, how to determine the possibility of developing into schizophrenia individuals before the onset of the diseases are particularly important. The study investigated cerebral gray matter volume differences and resting‐state functional connections among patients with psychosis risk syndrome (PRS), patients with first‐episode schizophrenic (FES), and healthy controls (HC), aiming to provide scientific clinical evidence for schizophrenia early identification and intervention. A total of 19 PRS patients, 18 FES patients, and 29 HC were recruited. Gray matter volume and amplitude of low‐frequency fluctuation (fALFF) during resting‐state functional studies were measured. Comparison of gray matter volumes showed that PRS and FES groups had common reduced gray matter volume in the right caudate. PRS and FES patients showed altered connectivity mainly in the semantic processing‐related brain areas. fALFF analysis found that PRF and FES patients had significant differences in fALFF values of the brain region mainly located in the subcortical network, visual recognition network, and auditory network. In addition, PRF individuals had a higher fALFF value and a lower fALFF value in the anterior wedge of the cerebral network than the HC group. Gray matter volume loss between related brain areas might appear prior to illness onset. Similar fALFF values occurred in PRS and FES groups indicated that multiple brain regions of neuronal activity abnormalities and unconventional neural network mechanism have been existed in PRS patients.