基于空间频率与时间序列信息的多类运动想象脑电分类

结合共同空间模式（CSP）、离散小波变换（DWT）和长短期记忆网络（LSTM）方法,提出一种基于空间频率与时间序列信息的多类运动想象脑电特征提取方法。首先利用滑动矩形窗获得时间序列脑电信号,并采用DWT从每一段脑电信号提取运动想象脑电相关的子带小波系数,其次将小波系数通过一对多CSP进一步特征提取,得到的特征作为LSTM的输入,然后对LSTM的时间序列输出在时间步上进行平均,最后使用Softmax分类器进行分类。实验结果显示,新算法取得92.23%的准确率,相比CSP特征以及结合频率或时间序列信息的CSP特征有较大提升,表明空间、频率、时间序列信息的互补性和有效性。     

基于脑电信号的疼痛强度识别方法研究

目的:通过对疼痛患者的脑电信号进行特征提取和特征选择,实现对疼痛等级的量化评估。方法:对临床采集的脑电信号进行离散小波变换得到近似和细节系数,根据每层分解系数计算子带能量占比、系数统计特征、样本熵和锁相值,组成特征向量。利用随机森林进行特征选择和疼痛预测。结果:实现对疼痛等级的三分类,平均分类准确率为91.7%,其中无痛和重痛的分类准确率达100%。结论:本研究方法可以有效地对脑电信号进行特征提取和选择,以较高的准确率实现疼痛强度的识别,为临床疼痛的客观评估奠定基础。     

想象左右手运动的脑电特征提取及分类研究

针对想象运动的脑机接口(BCI)系统存在分类准确率低、抗干扰能力差等不足,提出一种将离散小波变换(DWT)和BP神经网络相结合的脑电识别方法(DWT-BP法).通过计算想象左、右手运动的C3、C4的平均功率,合理确定时间窗设置,对时间窗内的平均功率信号进行离散小波变换,并选取尺度6上的逼近系数A6的组合信号作为脑电信号特征,以BP神经网络为分类器实现对脑电观测数据的分析.实验结果表明,DWT-BP方法能够较准确地提取脑电信号的本质特征,具有较好的抗干扰能力和分类性能,以及识别运动想象脑电信号的有效性,同时为实现运动想象在线BCI系统打下基础.     

基于小波包分析的意识任务特征提取与分类

将基于小波包变换的多尺度分析方法应用于自发脑电 (EEG)的特征提取。在对 3种意识任务的脑电信号进行多级小波包分解的基础上 ,将不同尺度空间的能量信号作为特征值 ,组成不同意识任务的特征向量 ,并利用径向基函数神经网络进行分类测试。结果表明 ,小波包变换方法的分类正确率高于自回归模型方法。小波包分析方法可以作为不同意识任务脑电信号特征提取的一种新方法 ,具有较强的稳定性     

基于想象左右手运动脑电特征提取及其统计特性分析

目的 基于想象运动的脑-机接口(BCI)系统中,探讨如何获取思维脑电、提取特征并对其进行准确而有效地分类.方法 利用小波变换方法构建小波系数提取不同思维脑电特征,并从统计学的角度对这些特征进行分析.结果 在实际动作前0.5～1 s左右,想象左右手运动时C3和C4处各自具有明显不同的脑电特征,且这些特征存在显著性差别(P＜0.05)说明小波系数能很好的反映不同思维脑电的特征.结论 小波分析方法可以有效抑制或消除噪声和提取反映不同思维的脑电特征,通过对构建出的小波系数特征作统计学上的特性分析发现不同思维脑电具有显著性差别,该特征为后续分析中得以被准确地转换(识别分类)提供了更为可靠的保证.     

基于TQWT的癫痫脑电信号的识别

针对癫痫脑电(EEG)信号的识别问题,提出了一种基于可调品质因子小波变换(TQWT)的脑电特征提取方法。首先,利用TQWT将EEG信号进行分解,得到各个小波子波带;然后,根据癫痫异常波对应的频率范围,合理的选择小波子波带进行重构,提取有效值和峰峰值构成特征分量;最后,采用支持向量机进行分类。将所提出方法应用于癫痫脑电信号的识别中,以德国伯恩大学癫痫研究中心采集的典型脑电数据进行验证。实验分析结果表明,所提出的特征提取方法对正常和癫痫发作期EEG信号的分类准确率可达98%。     

基于典型相关分析和小波变换的眼龟伪迹去除

目的针对脑电信号中眼电伪迹去除尚存在的问题，提出一种基于典型相关分析与小波变换的（wavelet—enhanced canonical correlation analysis，wCCA）自动去除眼电伪迹的算法。方法首先，充分利用脑电信号和眼电伪迹的空间分布特征，将基于典型相关分析的盲源分离算法分别应用于左右脑区的混合信号中，从而保证典型相关分析分解得到的第一个典型相关变量（即左右脑区之间的最公共成分），就是眼电伪迹分量。然后为了恢复泄漏在该伪迹分量中的脑电成分，对伪迹分量进行小波阈值滤波，将高于某一阈值的小波系数置零，而保留低于阈值的系数。结果与其他三种基于盲源分离去除眼电伪迹的方法相比较，该方法在有效地自动去除眼电伪迹的同时，很好地保留了潜在的脑电信号，去除效果明显优于其他三种方法。结论由于该算法简单，处理速度较快，因此应用于实时的脑机接口系统中更具优越性，为后续脑电信号的特征提取和分类分析提供了良好的基础。     

基于小波包和共同空间模型的运动想象脑电信号特征提取算法

脑-机接口(brain-computer interface,BCI)为无法进行交流的人们提供了一种新的交流方式。传统的基于频率特征的脑电信号(electroencephalogram,EEG)特征提取方法只提取每个通道的能量特征,而忽略了不同通道之间的相关性信息。为了获得更好的特征提取结果,本研究采用了基于小波包和共同空间模型(common space pattern, CSP)的脑电信号特征提取方法。首先,在利用小波包对脑电信号分解前,对相关通道和频带进行辨别,提取运动想象脑电μ律和β节律,然后利用CSP算法进行空间滤波提取特征,选取相关节点计算小波包能量,最后通过支持向量机(support vector machine, SVM)将脑电信号分为左右手两种特征。为了验证本研究算法的可行性与有效性,在BCI竞赛数据集上进行了相应的实验,分类结果表明,所提出的特征提取算法能够有效提取运动想象特征,具有较高的分类精度。     

基于小波方差的ECoG分类EI北大核心CSCD

针对典型的基于皮层脑电图(ECoG)的脑-机接口(BCI)设计,被试任务为想象左手小指和想象舌头运动,提出了采用小波方差的特征提取方法。首先在小波变换的基础上,提出小波方差的计算方法及其意义,并以此作为特征,从64导联中获取特征较为明显的6个导联进行分析;然后对脑电(EEG)数据进行三层小波分解,根据ERD/ERS现象,提取包含Mu节律和Beta节律的小波系数方差作为特征,采用交叉验证的方法,利用classify进行线性分类。离线分析表明,对训练集和测试集的分类正确率达到90.24%和93.77%,小波方差作为BCI研究中特征提取的方法具有更加简单和有效的特性。     

基于CSSD和SVM的抑郁症脑电信号分类

从EEG脑电信号中提取与疾病相关的信息以实现对抑郁症的自动诊断。首先采用共空域子空间分解(CSSD)方法,对躁狂型抑郁症患者与健康人两组的16导联脑电信号进行特征提取,然后用支持向量机(SVM)分类器进行训练和分类测试。实验结果表明,相对于用小波变换提取的频率相关参数为分类特征的分类准确率为88%,采用CSSD方法提取特征参数进行分类可以取得更理想的效果为95%,后者的16导联脑电信号在空间模型上表现出较高的模式可分性。该研究成果对精神抑郁症的物理诊断和研究提供了新的视角。     

EEG-based motor imagery classification using enhanced active segment selection and adaptive classifier

In this study, an adaptive electroencephalogram (EEG) analysis system is proposed for a two-session, single-trial classification of motor imagery (MI) data. Applying event-related brain potential (ERP) data acquired from the sensorimotor cortices, the adaptive linear discriminant analysis (LDA) is used for classification of left- and right-hand MI data and for simultaneous and continuous update of its parameters. In addition to the original use of continuous wavelet transform (CWT) and Student's two-sample t-statistics, the 2D anisotropic Gaussian filter is proposed to further refine the selection of active segments. The multiresolution fractal features are then extracted from wavelet data by means of modified fractal dimension. The classification in session 2 is performed by adaptive LDA, which is trial-by-trial updated using the Kalman filter after the trial is classified. Compared with original active segment selection and non-adaptive LDA on six subjects from two data sets, the results indicate that the proposed method is helpful to realize adaptive BCI systems.     

Fuzzy support vector machine for classification of EEG signals using wavelet-based features

Translation of electroencephalographic (EEG) recordings into control signals for brain–computer interface (BCI) systems needs to be based on a robust classification of the various types of information. EEG-based BCI features are often noisy and likely to contain outliers. This contribution describes the application of a fuzzy support vector machine (FSVM) with a radial basis function kernel for classifying motor imagery tasks, while the statistical features over the set of the wavelet coefficients were extracted to characterize the time–frequency distribution of EEG signals. In the proposed FSVM classifier, a low fraction of support vectors was used as a criterion for choosing the kernel parameter and the trade-off parameter, together with the membership parameter based solely on training data. FSVM and support vector machine (SVM) classifiers outperformed the winner of the BCI Competition 2003 and other similar studies on the same Graz dataset, in terms of the competition criterion of the mutual information (MI), while the FSVM classifier yielded a better performance than the SVM approach. FSVM and SVM classifiers perform much better than the winner of the BCI Competition 2005 on the same Graz dataset for the subject O3 according to the competition criterion of the maximal MI steepness, while the FSVM classifier outperforms the SVM method. The proposed FSVM model has potential in reducing the effects of noise or outliers in the online classification of EEG signals in BCIs.     

Adaptive subject-based feature extraction in brain-computer interfaces using wavelet packet best basis decomposition

In this paper we discuss a subject-based feature extraction method using wavelet packet best basis decomposition (WPBBD) in brain-computer interfaces (BCIs). The idea is to employ the wavelet packet best basis algorithm to adapt to each subject separately. Firstly, original electroencephalogram (EEG) signals are decomposed to a given level by wavelet packet transform. Secondly, for each subject, the best basis algorithm is used to find the best-adapted basis for that particular subject. Finally, subband energies contained in the best basis are used as effective features. Adaptive and specific features of a subject are so obtained. Three different motor imagery tasks of six subjects are discriminated using the above features. Experiment results show that the subject-based adaptation method yields significantly higher classification performance than the non-subject-based adaptation and non-adaptive approaches.     

Epileptic seizure detection using DWT-based approximate entropy,Shannon entropy and support vector machine: a case study

In this work, we have used a time–frequency domain analysis method called discrete wavelet transform (DWT) technique. This method stand out compared to other proposed methods because of its algorithmic elegance and accuracy. A wavelet is a mathematical function based on time-frequency analysis in signal processing. It is useful particularly because it allows a weak signal to be recovered from a noisy signal without much distortion. A wavelet analysis works by analysing the image and converting it to mathematical function which is decoded by the receiver. Furthermore, we have used Shannon entropy and approximate entropy (ApEn) for extracting the complexities associated with electroencephalographic (EEG) signals. The ApEn is a suitable feature to characterise the EEGs because its value drops suddenly due to excessive synchronous discharge of neurons in the brain during epileptic activity in this study. EEG signals are decomposed into six EEG sub-bands namely D1–D5 and A5 using DWT technique. Non-linear features such as ApEn and Shannon entropy are calculated from these sub-bands and support vector machine classifiers are used for classification purpose. This scheme is tested using EEG data recorded from five healthy subjects and five epileptic patients during the inter-ictal and ictal periods. The data are acquired from University of Bonn, Germany. The proposed method is evaluated through 15 classification problems, and obtained high classification accuracy of 100% for two cases and it indicates the good classifying performance of the proposed method.     

Adaptive feature extraction for EEG signal classification

One challenge in the current research of brain–computer interfaces (BCIs) is how to classify time-varying electroencephalographic (EEG) signals as accurately as possible. In this paper, we address this problem from the aspect of updating feature extractors and propose an adaptive feature extractor, namely adaptive common spatial patterns (ACSP). Through the weighed update of signal covariances, the most discriminative features related to the current brain states are extracted by the method of multi-class common spatial patterns (CSP). Pseudo-online simulations of EEG signal classification with a support vector machine (SVM) classifier for multi-class mental imagery tasks show the effectiveness of the proposed adaptive feature extractor.     

基于形状纹理特征的食管癌和肝包虫病图像分类

目的:探讨用K最近邻(KNN)分类算法对食管癌X射线图像和肝包虫CT图像的Hu不变矩形状特征和小波变换纹理特征进行分类研究。方法:利用Hu不变矩算法和小波变换算法对食管癌X射线图像和肝包虫CT图像提取特征,用KNN分类器对特征值进行分类以验证所提取特征的分类能力。结果:对于食管癌X射线图像使用Hu不变矩算法提取形状特征具有较好的分类性能,对于肝包虫CT图像使用小波变换算法提取纹理特征具有较好的分类性能。结论:Hu不变矩形状特征结合KNN分类器的研究方法为新疆哈萨克族食管癌的分型提供一定的依据,小波变换纹理特征结合KNN分类器的研究方法为地方性肝包虫病的分型提供一定的依据,同时为计算机辅助诊断系统的研发奠定基础。     

基于迁移学习和空洞卷积的癫痫状态识别方法

癫痫患者脑电信号的自动检测和发作诊断对临床治疗癫痫具有重要意义。针对训练数据有限及训练与测试数据分布不一致的难点,采用领域间联合知识迁移学习方法,实现小训练数据量下的癫痫状态识别。首先对脑电信号进行4层小波包分解,提取小波包分解系数作为特征,通过边缘分布和联合分布迭代调整,完成源域和目标域特征之间的知识迁移,训练空洞卷积神经网络作为分类器,完成目标域癫痫状态识别。分别在波士顿儿童医院CHB-MIT脑电数据集(22 例被试,共计790 h)和波恩大学癫痫脑电数据集(5 组,每组100 个片段,每段23.6 s)上进行算法验证,实验结果表明,所提出的方法对复杂癫痫状态的平均识别准确度、敏感性、特异性在CHB-MIT数据集上达到96.8％、96.1％、96.4％;在波恩数据集上,平均识别准确率为96.9%,有效提高了癫痫状态识别综合性能,实现了癫痫发作稳定可靠检测。     

A statistical based feature extraction method for breast cancer diagnosis in digital mammogram using multiresolution representation

This paper presents a method for breast cancer diagnosis in digital mammogram images. Multi-resolution representations, wavelet or curvelet, are used to transform the mammogram images into a long vector of coefficients. A matrix is constructed by putting wavelet or curvelet coefficients of each image in row vector, where the number of rows is the number of images, and the number of columns is the number of coefficients. A feature extraction method is developed based on the statistical t-test method. The method is ranking the features (columns) according to its capability to differentiate the classes. Then, a dynamic threshold is applied to optimize the number of features, which can achieve the maximum classification accuracy rate. The method depends on extracting the features that can maximize the ability to discriminate between different classes. Thus, the dimensionality of data features is reduced and the classification accuracy rate is improved. Support vector machine (SVM) is used to classify between the normal and abnormal tissues and to distinguish between benign and malignant tumors. The proposed method is validated using 5-fold cross validation. The obtained classification accuracy rates demonstrate that the proposed method could contribute to the successful detection of breast cancer.     

利用整体经验模态分解和随机森林的脑电信号分类研究

癫痫脑电信号的自动监测与分类在临床医学上具有重要意义。针对脑电信号的非平稳特点,提出一种基于整体经验模态分解和随机森林相结合的脑电信号分类方法。选取波恩大学脑电信号数据集中癫痫发作间期和发作期的200个单通道信号,共819 400个数据作为样本。首先利用整体模态分解将癫痫脑电信号分解成多个固有模态函数,然后对各阶固有模态函数提取有效特征,最后分别用随机森林和最小二乘支持向量机对脑电信号的特征进行分类。将随机森林与最小二乘支持向量机分类正确识别率对比,结果表明,随机森林分类方法对发作期和发作间期的癫痫脑电信号的分类效果比较理想,识别精度为99.60％,高于最小二乘支持向量机的准确性。该方法的提出能有效提高临床癫痫脑电信号分析的效率。