Analysis of EEG signals by implementing eigenvector methods/recurrent neural networks

The implementation of recurrent neural network (RNN) employing eigenvector methods is presented for classification of electroencephalogram (EEG) signals. In practical applications of pattern recognition, there are often diverse features extracted from raw data which needs recognizing. Because of the importance of making the right decision, the present work is carried out for searching better classification procedures for the EEG signals. Decision making was performed in two stages: feature extraction by eigenvector methods and classification using the classifiers trained on the extracted features. The aim of the study is classification of the EEG signals by the combination of eigenvector methods and the RNN. The present research demonstrated that the power levels of the power spectral density (PSD) estimates obtained by the eigenvector methods are the features which well represent the EEG signals and the RNN trained on these features achieved high classification accuracies.     

Expert systems for time-varying biomedical signals using eigenvector methods

In this paper, we present the expert systems for time-varying biomedical signals classification and determine their accuracies. The combined neural network (CNN), mixture of experts (ME), and modified mixture of experts (MME) were tested and benchmarked for their performance on the classification of the studied time-varying biomedical signals (ophthalmic arterial Doppler signals, internal carotid arterial Doppler signals and electroencephalogram signals). Decision making was performed in two stages: feature extraction by eigenvector methods and classification using the classifiers trained on the extracted features. The inputs of these expert systems composed of diverse or composite features were chosen according to the network structures. The present study was conducted with the purpose of answering the question of whether the expert system with diverse features (MME) or composite feature (CNN, ME) improve the capability of classification of the time-varying biomedical signals. The purpose was to determine an optimum classification scheme for the problem and also to infer clues about the extracted features. Our research demonstrated that the power levels of power spectral density (PSD) estimations obtained by the eigenvector methods are the valuable features which are representing the time-varying biomedical signals and the CNN, ME, and MME trained on these features achieved high classification accuracies.     

A modified mixture of experts network structure for ECG beats classification with diverse features

This paper illustrates the use of modified mixture of experts (MME) network structure to guide model selection for classification of electrocardiogram (ECG) beats with diverse features. The MME is a modular neural network architecture for supervised learning. Expectation-maximization (EM) algorithm was used for training the MME so that the learning process is decoupled in a manner that fits well with the modular structure. The wavelet coefficients and Lyapunov exponents of the ECG signals were computed and statistical features were calculated to depict their distribution. The statistical features, which were used for obtaining the diverse features of the ECG signals, were then input into the MME network structure for training and testing purposes. We explored the ability of designed and trained MME network structure, combined with wavelet preprocessing (computing wavelet coefficients) and nonlinear dynamics tools (computing Lyapunov exponents), to discriminate five types of ECG beats (normal beat, congestive heart failure beat, ventricular tachyarrhythmia beat, atrial fibrillation beat, partial epilepsy beat) obtained from the Physiobank database. The MME achieved accuracy rates which were higher than that of the mixture of experts (ME) and feedforward neural network models (multilayer perceptron neural network—MLPNN). The proposed MME approach can be useful in classifying long-term ECG signals for early detection of heart diseases/abnormalities.     

Combined neural network model employing wavelet coefficients for EEG signals classification

This paper illustrates the use of combined neural network model to guide model selection for classification of electroencephalogram (EEG) signals. The EEG signals were decomposed into time–frequency representations using discrete wavelet transform and statistical features were calculated to depict their distribution. The first-level networks were implemented for the EEG signals classification using the statistical features as inputs. To improve diagnostic accuracy, the second-level networks were trained using the outputs of the first-level networks as input data. Three types of EEG signals (EEG signals recorded from healthy volunteers with eyes open, epilepsy patients in the epileptogenic zone during a seizure-free interval, and epilepsy patients during epileptic seizures) were classified with the accuracy of 94.83% by the combined neural network. The combined neural network model achieved accuracy rates which were higher than that of the stand-alone neural network model.     

基于脑电多特征融合的癫痫发作预测方法

癫痫的发作会给患者的身体和精神造成极大的创伤,对癫痫发作的准确预测可以及时协助医生对患者采取治疗措施.为了准确预测癫痫发作,提出脑电特征和多通道脑电交互特征相融合的癫痫发作预测方法.首先,提出多尺度符号化排列传递熵对多通道脑电信号交互信息进行分析,生成同步矩阵,并通过显著性分析筛选与癫痫发作相关的重要脑电通道,减少不必要特征对分类的干扰;然后,对筛选通道后的脑电信号生成表征脑电信号特征的功率谱密度能量图(PSDED)和描述脑通道交互特征的同步矩阵图(SMD),将两个特征图融合,采用深度卷积神经网络(DCNN)对癫痫患者脑电信号进行分类识别,提高学习能力和泛化能力,分类准确率可达到96.825%;最后,在分类的基础上采用预测评价系统对癫痫发作预测性能进行评估,癫痫发作预测范围(SPH)为10 min和发作发生期(SOP)为10 min时,预测敏感性达到96.66%,误检率可达到0.03/h;当SPH为30min,SOP为10 min时,预测敏感性达到93.17%,误检率可达到0.05/h.与现有研究结果相比较,所提出方法具有较好的预测敏感度和较低的误检率.     

基于皮尔逊最优电极选择的ADHD患者脑电特征提取及分类研究

事件相关电位(ERP)可用于注意缺陷多动障碍儿童(ADHD)和正常儿童的脑电特征 提取与分类。首先,采用赌博任务范式,采集2 类儿童的脑电信号;其次,基于皮尔逊相关系 数算法选择最优电极,并预处理最优电极脑电信号;然后,提取预处理脑电信号的时域特征(均 值、方差、峰值)和频域特征(Theta 波段功率、Alpha 波段功率);最后,利用传统分类方法支持 向量机(SVM)、自适应增强(AdaBoost)、自举汇聚法(Bagging)、线性判别式分析(LDA)、反向传 播(BP)和组合分类器的分类方法(LDA-SVM,BP-SVM)完成对2 种脑电信号的分类。研究结果 表明,传统方法BP 分类器的分类准确率可达80.52%,组合分类器BP-SVM 的分类准确率可达 88.88%。组合分类方法能提高ADHD 儿童的分类准确率,为基于脑机接口技术的ADHD 神经 反馈康复治疗提供技术支持。     

Real-time removal of ocular artifacts from EEG based on independent component analysis and manifold learning

Frequent occurrence of ocular artifacts leads to serious problems in interpreting and analyzing the electroencephalogram (EEG). In the present paper, a novel and robust technique is proposed to eliminate ocular artifacts from EEG signals in real time. Independent Component Analysis (ICA) is used to decompose EEG signals. The features of topography and power spectral density of those components are extracted. Moreover, we introduce manifold learning algorithm, a recently popular dimensionality reduction technique, to reduce the dimensionality of initial features, and then those new features are fed to a classifier to identify ocular artifacts components. A k-nearest neighbor classifier is adopted to classify components because classification results show that manifold learning with the nearest neighbor algorithm works best. Finally, the artifact removal method proposed here is evaluated by the comparisons of EEG data before and after artifact removal. The results indicate that the method proposed could remove ocular artifacts effectively from EEG signals with little distortion of the underlying brain signals and be satisfied the real-time application.     

Least squares support vector machine employing model-based methods coefficients for analysis of EEG signals

The aim of the study is classification of the electroencephalogram (EEG) signals by combination of the model-based methods and the least squares support vector machines (LS-SVMs). The LS-SVMs were implemented for classification of two types of EEG signals (set A – EEG signals recorded from healthy volunteers with eyes open and set E – EEG signals recorded from epilepsy patients during epileptic seizures). In order to extract the features representing the EEG signals, the spectral analysis of the EEG signals was performed by using the three model-based methods (Burg autoregressive – AR, moving average – MA, least squares modified Yule–Walker autoregressive moving average – ARMA methods). The present research demonstrated that the Burg AR coefficients are the features which well represent the EEG signals and the LS-SVM trained on these features achieved high classification accuracies.     

基于能量熵的运动想象脑电信号分类

对脑电信号进行特征提取和分类是脑机接口研究的核心问题,利用不同运动想象脑电信号能量熵的变化,从能量熵中提取特征,利用自定义基于统计理论分类方法进行分类,结果均达到90%以上。     

基于能量特征的脑电信号特征提取与分类

为了快速、有效地提取脑电特征,提高分类正确率,采用带通滤波和小波包分析的方法提取Mu、Beta节律对应的脑电信号,在时域范围内,将信号幅度的平方作为能量特征值;在频域范围内,采用AR模型功率谱估计法所得的功率谱密度作为能量特征值.根据运动想象脑电信号特点,构造左右通道信号能量差值的符号特性作为分类判别依据,进行分类测试,方法简单.初步实验结果表明,所利用的两种方法的分类正确率达87.857%.     

Analysis of electrocardiographic changes in partial epileptic patients by combining eigenvector methods and support vector machines

Abstract: In the present study, the diagnostic accuracy of support vector machines (SVMs) on electrocardiogram (ECG) signals is evaluated. Two types of ECG beats (normal and partial epilepsy) were obtained from the Physiobank database. Decision making was performed in two stages: feature extraction by eigenvector methods and classification using the SVM trained on the extracted features. The present research demonstrates that the power levels of the power spectral densities obtained by eigenvector methods are features which represent the ECG signals well and SVMs trained on these features achieve high classification accuracies.     

脑电信号多维特征融合与分类算法研究

针对目前运动想象脑电信号特征提取单一,分类识别准确率低等现象,结合卷积神经网络分类器,提出了一种多维度特征加权融合的特征融合算法来提高运动想象脑电识别率。对预处理后的脑电信号进行小波包变换,提取其共空间特征、能量特征、边际谱熵特征以及非线性动力学特征,然后加权融合,使用卷积神经网络分类器分类。为验证算法的合理性,使用BCI-IV Dataset 2a数据集对提出的特征融合算法进行验证分析,结果表明,所提出的加权特征融合算法结合CNN分类器可以有效提高运动想象识别准确率。实验中,9位志愿者平均分类准确率达到75.88%,平均Kappa系数为0.70。     

基于图卷积网络的运动想象识别

运动想象识别将大脑的神经活动信号转为编码输出以实现意念控制,是脑机接口的一个重要研究方向.近年来深度学习算法的应用进一步提高了运动想象识别的准确率,但是当前基于深度学习的运动想象分析都将多路脑电信号作为二维矩阵信号,忽视了不同节点的空间关联信息.为了解决这个问题,将图卷积网络算法应用到运动想象分类中,通过多个节点脑电信...     

Complex networks approach for EEG signal sleep stages classification

Sleep stage scoring is a challenging task. Most of existing sleep stage classification approaches rely on analysing electroencephalography (EEG) signals in time or frequency domain. A novel technique for EEG sleep stages classification is proposed in this paper. The statistical features and the similarities of complex networks are used to classify single channel EEG signals into six sleep stages. Firstly, each EEG segment of 30 s is divided into 75 sub-segments, and then different statistical features are extracted from each sub-segment. In this paper, feature extraction is important to reduce dimensionality of EEG data and the processing time in classification stage. Secondly, each vector of the extracted features, which represents one EEG segment, is transferred into a complex network. Thirdly, the similarity properties of the complex networks are extracted and classified into one of the six sleep stages using a k-means classifier. For further investigation, in the statistical features extraction phase two statistical features sets are tested and ranked based on the performance of the complex networks. To investigate the classification ability of complex networks combined with k-means, the extracted statistical features were also forwarded to a k-means and a support vector machine (SVM) for comparison. We also compare the proposed method with other existing methods in the literature. The experimental results show that the proposed method attains better classification results and a reasonable execution time compared with the SVM, k-means and the other existing methods. The research results in this paper indicate that the proposed method can assist neurologists and sleep specialists in diagnosing and monitoring sleep disorders.     

Probabilistic neural networks combined with wavelet coefficients for analysis of electroencephalogram signals

Abstract: In this paper, the probabilistic neural network is presented for classification of electroencephalogram (EEG) signals. Decision making is performed in two stages: feature extraction by wavelet transform and classification using the classifiers trained on the extracted features. The purpose is to determine an optimum classification scheme for this problem and also to infer clues about the extracted features. The present research demonstrates that the wavelet coefficients obtained by the wavelet transform are features which represent the EEG signals well. The conclusions indicate that the probabilistic neural network trained on the wavelet coefficients achieves high classification accuracies (the total classification accuracy is 97.63%).     

Performance analysis of power and power variance for classification,detection and localization of epileptic multi-channel EEG

There are a large number of data sets of EEG signal for which, it is difficult to judge and monitor brain activity through observations. Epilepsy is a disorder in which a recurrent and sudden malfunction of the brain is characterized. It is proposed to classify, detect and localize Epileptic multi-channel EEG through various power and novel power variance features non-invasively. This work presents power spectral estimation (PSE) using time–frequency analysis of EEG signals in both parametric (FFT) and non-parametric methods (i.e. Welch, Burg, Covariance, MUSIC and Yule–Walker). To examine the robustness of power features for different methods, the analysis of p value is performed. The detection of epileptic seizure is classified using different kernels through SVM. It is observed from the PSE that the power features have higher values in epileptic subjects as compared to non-epileptic subjects. Amongst all the parametric and non-parametric methods, the MUSIC method gives the highest average power. Sensitivity, specificity, and classification accuracy are 100% for Welch, Burg, Covariance, and Yule–Walker methods while MUSIC and FFT methods deliver 98.73 and 99.52% respectively. The novelty is introduced through the quantification of power and power variance robust feature region/lobe-wise. This quantification is used for the localization of 25 epileptic subjects. Analysis of the parametric and non-parametric PSD methods for extraction of power and power variance features is not used by any study. These are effectively utilized for detection and localization of epilepsy non-invasively.

     

基于小波系数特征融合的小鼠癫痫脑电分类

采集癫痫小鼠模型在常态与致癫状态下的脑电信号以研究其癫痫脑电的自动分类。对经过噪声和伪迹消除预处理的脑电信号进行小波变换,获得不同频率子带的小波系数,对脑电信号及与癫痫特征波相关的小波系数提取相应的线性特征（标准差）和非线性特征（样本熵）;基于这些特征及其组合使用支持向量机分类器实现分类。实验发现基于小鼠脑电本身的标准差和样本熵的分类正确率分别为59.10%和58.00%;而融合各相关小波系数的标准差或样本熵,分类正确率分别达到86.60%和88.60%;融合全部相关小波系数的线性和非线性特征后分类正确率为99.80%。这些结果说明基于小波系数特征融合的分类算法性能有显著提升,能有效实现小鼠癫痫脑电的自动分类。     

抑郁症静息脑电的小波包节点功率谱熵分析

头皮脑电(EEG)信号反映了大脑皮层神经元细胞群自发性节律性的电生理活动,含有丰富的生理与病理信息,是临床脑神经与精神疾病诊断的重要依据.针对抑郁症的研究和诊断中缺少客观有效的量化参数和指标的状况,提出一种基于小波包分解节点重构信号的功率谱熵值(记为W值)的脑电信号分析方法,并利用此方法对静息态的脑电信号进行计算和分析.实验和分析结果表明:抑郁症患者脑电信号S32节点(频率24~32 Hz)的熵值(置信区间[0.0129,0.0176])在部分脑区显著大于正常健康人(置信区间[0.0246,0.0303]),显示抑郁症病人快波节律的能量分布存在弥散性,符合现在关于抑郁症患者自我调节能力减弱的发病机制.对结果进行了T检验统计分析,证明了这种辨别方法的准确性和可行性,将为抑郁症疾病检测诊断提供有效的量化物理指标.     

用于脑电数据增强和情绪识别的自注意力GAN

针对脑电信号（electroencephalogram,EEG）情绪识别中数据稀缺及由此导致的情感分类精度不高的问题,提出了一个引入自注意力机制的条件Wasserstein生成对抗网络（SA-cWGAN）,通过自注意力模块从训练数据学习长时上下文相关的全局特征,采用Wasserstein距离和梯度惩罚的Lipschitz约束对网络的损失函数进行优化,进而生成高质量的EEG数据对原有训练集进行增强。所提方法分别在DEAP和SEED数据集上进行了大量的二分类和三分类对比实验,生成了与EEG训练数据分布接近的微分熵（DE）和功率谱密度（PSD）特征,以此来增强EEG训练数据集,采用SVM分类器对增强后的EEG特征进行情绪分类。实验结果表明,在DEAP数据集上的唤醒度和效价维度下,增强后的DE、PSD特征较原有DE、PSD特征二分类准确率分别提高了16.63、17.55个百分点和6.48、8.34个百分点;在SEED数据集下,三分类准确率分别提高了4.64、5.18个百分点,证明所提方法生成的特征具有良好的鲁棒性,也表明通过对GAN网络引入自注意力机制生成的特征增强原有训练数据集能够有效提高E...     

Short-term emotion assessment in a recall paradigm

The work presented in this paper aims at assessing human emotions using peripheral as well as electroencephalographic (EEG) physiological signals on short-time periods. Three specific areas of the valence–arousal emotional space are defined, corresponding to negatively excited, positively excited, and calm-neutral states. An acquisition protocol based on the recall of past emotional life episodes has been designed to acquire data from both peripheral and EEG signals. Pattern classification is used to distinguish between the three areas of the valence–arousal space. The performance of several classifiers has been evaluated on 10 participants and different feature sets: peripheral features, EEG time–frequency features, EEG pairwise mutual information (MI) features. Comparison of results obtained using either peripheral or EEG signals confirms the interest of using EEGs to assess valence and arousal in emotion recall conditions. The obtained accuracy for the three emotional classes is 63% using EEG time–frequency features, which is better than the results obtained from previous studies using EEG and similar classes. Fusion of the different feature sets at the decision level using a summation rule also showed to improve accuracy to 70%. Furthermore, the rejection of non-confident samples finally led to a classification accuracy of 80% for the three classes.