基于时频混合域局部统计的帕金森病语音障碍分析方法研究

针对帕金森病语音检测问题,本文提出了一种基于时频混合域局部统计的帕金森病语音障碍分析方法。该方法首先将语音信号从时域转化为时频混合域,即进行时频化表示。在时频化表示方法中将语音信号进行分帧处理,再将每帧的语音进行傅里叶变换,通过计算得到能量谱,并将能量谱通过映射关系映射到图像空间进行可视化;其次统计信号每个能量数据在时间轴上和频率轴上的差分值,根据差分值计算该能量的梯度统计特征,用梯度统计特征来表示其不同时域与频域的能量值的突变情况;最后利用KNN分类器对提取的梯度统计特征进行分类。本文在不同的帕金森病语音数据集上进行实验,发现本文所提取的梯度统计特征在分类时有更强的聚类性。与基于传统特征与深度学习特征的分类结果相比,本文所提取的梯度统计特征在分类准确率、特异性和灵敏性上均优于前二者。实验证明了本文所提出的梯度统计特征在帕金森病语音分类诊断中的可行性。     

一种基于独立成份分析的fMRI时-空模型数据处理方法

独立成份分析(ICA)是信号处理领域中斯近发展起来的一种很有应用前景的方法，而脑功能磁共振(fMRI)信号的有效分离与识别是一个正在研究和试验之中的技术领域。因此，发展基于ICA的fMRI数据处理方法具有明显的理论价值和应用前景。本文首先介绍了ICA原理，分析了现行ICA—fMRI方法采用的信号与噪声的空域分布相互独立的信号模型所存在的明显不足，然后提出了微域中的信号与噪声的时域过程相互独立的fMRI信号模型，从而建立了一种新的fMRI数据处理方法：邻域独立成份相关法。合理的fMRI实验数据处理结果验证了新方法的合理性。     

独立分量分析的拟牛顿迭代算法及其在视觉诱发电位特征提取中的应用

少次叠加平均处理后的视觉诱发电位（VEP）中仍含有一定的背景噪声．对其进行进一步的提取与处理有重要的实用价值。独立分量分析（ICA）能够从混合信号中分离出最独立的成分，有效抑制噪声。本文尝试采用ICA的拟牛顿迭代算法进行VEP特征提取，介绍该方法的原理、实验和结果，并与采用牛顿迭代准则的快速独立孕量分析（Fast ICA）算法进行了比较。结果表明，基于拟牛顿法的ICA可以有效增强信号，从少次叠加平均的结果中提取出易于辨识的VEP的P300信号，具有较高的应用价值。     

基于声门闭合时刻估计的声调的提取

利用小波变换模极大值点同信号突变点之间的关系，以及声门闭合时刻(GCI)引起语音信号锐变的特点，通过检测两个相邻GCI的距离来估计语音的基音周期，由此求出基音频率对时间的变化曲线(即声调曲线)。通过仿真，该方法提取声调准确，而且抗噪音的能力比较强。     

基于独立分量分析的生理信号盲源分离

用于盲源分离的独立分量分析（ICA）和扩展ICA算法，基于极大似然估计，给出一个衡量输出分量统计独立的目标函数，最优化目标函数，得到一种用于独立分量分析的迭代算法。扩展ICA算法的优点在于迭代过程中不需要计算信号的高阶统计量，收敛速度快，同时适用于超高斯和亚高斯信号的分离。应用该算法实现了脑电、心电信号以及语音信号的分离，并给了实验结果。     

基于独立分量分析的生理信号盲源分离

用于盲源分离的独立分量分析 (ICA)和扩展ICA算法 ,基于极大似然估计 ,给出一个衡量输出分量统计独立的目标函数 ,最优化该目标函数 ,得到一种用于独立分量分析的迭代算法。扩展ICA算法的优点在于迭代过程中不需要计算信号的高阶统计量 ,收敛速度快 ,同时适用于超高斯和亚高斯信号的分离。应用该算法实现了脑电、心电信号以及语音信号的分离 ,并给出了实验结果     

基于声门闭合时刻估计的声调的提取

利用小波变换模极大值点同信号突变点之间的关系，以及声门闭合时刻(GCI)引起语音信号锐变的特点，通过检测两个相邻GCI的距离来估计语音的基音周期，由此求出基音频率对时间的变化曲线(即声调曲线)。通过仿真，该方法提取声调准确，而且抗噪音的能力比较强。     

基于语音转换技术的普通话电子喉语音增强方法研究

目的电子喉是喉切除患者使用最多的语音恢复工具,但是电子喉语音存在发声机械、音调单一、辐射噪声大等缺点,本文拟运用语音转换技术改善电子喉语音的发声效果,提高语音自然度和可懂度。方法选择200句分别以自然发声和电子喉发声的标准普通话日常用语作为训练语料,采用基于混合高斯模型(Gaussian mixed model,GMM)的语音转换方法对电子喉语音进行转换,转换参数为基频轨迹和声道谱参数(0~24阶梅尔倒谱系数),然后对转换后的语音质量进行主客观评价。结果转换语音的高频辐射噪声得到了有效抑制,基频变化出现。主观分析结果显示,转换语音的自然度和可接受度有所提高,但可懂度变化不大。结论使用语音转换技术可以降低电子喉语音的高频辐射噪声,改变声调和韵律信息,提高自然度和可接受度,对改善电子喉语音的听觉质量有较大帮助。     

独立分量分析在脑电信号处理中的应用及研究进展

独立分量分析(independent component analysis,ICA)方法是从一组观测信号中提取统计独立分量的方法.因为用这种方法分解出的各信号分量之间是相互独立的,而测得的脑电信号往往包含若干相对独立的成分,所以用它来分解脑电信号,所得的结果更具有生理意义,有利于去除干扰和伪差.本文简要地回顾了ICA的发展历史和主要算法,综述了它在脑电信号处理中的应用及研究进展,并指出了需要进一步研究解决的问题.     

基于带参考信号的ICA算法的脑电信号眨眼伪差的分离研究

独立分量分析（ICA）是一种从混合信号中提取统计独立的分量的一种方法.本研究提出了一种基于带参考信号的ICA算法的脑电信号眨眼伪差的分离方法,可以得到纯净的脑电信号.这个方法的主要思路是：先选取一导眨眼伪差比较明显的数据,从中获得眨眼伪差的参考信号,再用ICA方法把眨眼伪差第一个提取出来,最后得到消除伪差后的EEG信号.详细讨论了使用带参考信号的ICA算法消除眨眼伪差的方法与步骤,并给出了应用于真实信号的实验结果.     

Recovering dipole sources from scalp-recorded event-related-potentials using component analysis: principal component analysis and independent component analysis.

Principal component analysis (PCA) and independent component analysis (ICA) were examined in their ability to recover dipole sources from simulated data. Datasets of EEG segments were generated that contained cortical sources that were temporally overlapping or non-overlapping, and dipole sources with varying degree of spatial orthogonality. For temporal overlapping sources, both PCA and ICA resulted in components that required multiple-source equivalent current dipole models. The spatially overlapping sources affected the PCA method more than ICA, resulting in single PCA components in which all non-orthogonal sources were represented. For both PCA and ICA, dipole models with fixed-location dipoles successfully accounted for most of the variance in the component weights, even when the spatial or temporal overlap of the generating sources required multiple-dipole models.     

基于独立分量分析的脑电噪声消除

作为一种新的多元统计处理方法，独立分量分析(ICA)是解决盲源分离(BSS)问题的一个有效手段。在简要分析ICA理论及其算法的基础上，提出将其应用到脑电中的眼电伪迹的去除任务。实际采集的生理信号大多由相互独立的成分线性迭加而成，符合ICA要求源信号统计独立的基本假设。与传统方法相比，ICA这种空间滤波器不受信号频谱混迭的限制，消噪的同时能对有用信号的细节成分做到很好的保留，很大程度上弥补了时频域方法的不足。此外解混矩阵的逆可以用来反映独立源的空间分布模式，具有重要的生理意义。     

A hybrid method of application of independent component analysis to in vivo 1H MR spectra of childhood brain tumours

Independent component analysis (ICA) can automatically extract individual metabolite, macromolecular and lipid (MMLip) components from a series of in vivo MR spectra. The traditional feature extraction (FE)-based ICA approach is limited, in that a large sample size is required and a combination of metabolite and MMLip components can appear in the same independent component. The alternative ICA approach, based on blind source separation (BSS), is weak when dealing with overlapping peaks. Combining the advantages of both BSS and FE methods may lead to better results. Thus, we propose an ICA approach involving a hybrid of the BSS and FE techniques for the automated decomposition of a series of MR spectra. Experiments were performed on synthesised and patient in vivo childhood brain tumour MR spectra datasets. The hybrid ICA method showed an improvement in the decomposition ability compared with BSS-ICA or FE-ICA, with an increased correlation between the independent components and simulated metabolite and MMLip signals. Furthermore, we were able to automatically extract metabolites from the patient MR spectra dataset that were not in commonly used basis sets (e.g. guanidinoacetate).     

Extraction of gastric slow waves from electrogastrograms: Combining independent component analysis and adaptive signal enhancement

The electrogastrogram (EGG), a cutaneous measurement of gastric electrical activity, is a mixture of gastric slow waves and noise. To detect the propagation of gastric slow waves, it is desired to obtain gastric slow waves in each of multichannel EGGs. Recently, independent component analysis (ICA) has shown its efficiency in separating the gastric slow wave from noisy multichannel EGGs. However, this method is not able to recover gastric slow waves in each of the multichannel EGGs. In this paper, a twostage combined method was proposed for extracting gastric slow waves. First, ICA was performed to separate the gastric slow wave component from noisy multichannel EGGs. Second, adaptive signal enhancement with a reference input derived by the ICA in the first stage was employed to extract gastric slow waves in each channel. Quantitative analysis showed that, with the proposed method, the maximum root-mean-square error between the estimated time lag and its theoretical value in the simulations was only 0.65. The results from real EGG data demonstrated that the combined method was able to extract gastric slow waves from individual channels of EGGs which are important to identify the slow wave propagation. Therefore, the proposed method can be used to detect propagation of gastric slow waves from multichannel EGGs.     

Limitations and applications of ICA for surface electromyogram

Surface electromyogram (SEMG) has numerous applications, but the presence of artefacts and noise, especially at low level of muscle activity make the recordings unreliable. Spectral and temporal overlap can make the removal of artefacts and noise, or separation of relevant signals from other bioelectric signals extremely difficult. Individual muscles may be considered as independent at the local level and this makes an argument for separating the signals using independent component analysis (ICA). In the recent past, due to the easy availability of ICA tools, numbers of researchers have attempted to use ICA for this application. This paper reports research conducted to evaluate the use of ICA for the separation of muscle activity and removal of the artefacts from SEMG. It discusses some of the conditions that could affect the reliability of the separation and evaluates issues related to the properties of the signals and number of sources. The paper also identifies the lack of suitable measure of quality of separation for bioelectric signals and it recommends and tests a more robust measure of separation. The paper also reports tests using Zibulevsky's technique of temporal plotting to identify number of independent sources in SEMG recordings. The theoretical analysis and experimental results demonstrate that ICA is suitable for SEMG signals. The results identify the unsuitability of ICA when the number of sources is greater than the number of recording channels. The results also demonstrate the limitations of such applications due to the inability of the system to identify the correct order and magnitude of the signals. The paper determines the suitability of the use of error measure using simulated mixing matrix and the estimated unmixing matrix as a means identifying the quality of separation of the output. The work demonstrates that even at extremely low level of muscle contraction, and with filtering using wavelets and band pass filters, it is not possible to get the data sparse enough to identify number of independent sources using Zibulevs.ky's technique.     

How to stop or change a motor response: Laplacian and independent component analysis approach

Response inhibition is an essential control function necessary to adapt one's behavior. This key cognitive capacity is assumed to be dependent on the prefrontal cortex and basal ganglia. It is unresolved whether varying inhibitory demands engage different control mechanisms or whether a single motor inhibitory mechanism is involved in any situation. We addressed this question by comparing electrophysiological activity in conditions that require stopping a response to conditions that require switching to an alternate response. Analyses of electrophysiological data obtained from stop-signal tasks are complicated by overlapping stimulus-related activity that is distributed over frontal and parietal cortical recording sites. Here, we applied Laplacian transformation and independent component analysis (ICA) to overcome these difficulties. Participants were faster in switching compared to stopping a response, but we did not observe differences in neural activity between these conditions. Both stop- and change-trials Laplacian transformed ERPs revealed a comparable bilateral parieto-occipital negativity around 180 ms and a frontocentral negativity around 220 ms. ICA results suggested an inhibition-related frontocentral component which was characterized by a negativity around 200 ms with a likely source in anterior cingulate cortex. The data provide support for the importance of posterior mediofrontal areas in inhibitory response control and are consistent with a common neural pathway underlying stopping and changing of a motor response. The methodological approach proved useful to distinguish frontal and parietal sources despite similar timing and the ICA approach allowed assessment of single-trial data with respect to behavioral data.     

Organ-specific autoantibodies in HLA genotyped insulin-dependent diabetes mellitus families

The prevalence of cytoplasmic islet cell antibodies (ICA) and extrapancreatic antibodies (EPA), (stomach, adrenal and thyroid) was investigated in 132 juvenile onset diabetic patients, without personal or familial history of other autoimmune disease, and their 31 diabetic and 402 non-diabetic first degree relatives. The prevalence of ICA was 59% in index cases and 12% in the non-affected first degree relatives. The frequency of EPA was 23% and 16% respectively. There were no sex-related differences among the patients. However, among the non-affected relatives, an increased frequency of EPA was observed in females (23%) compared to males (8%) (P less than 10-4). There was a higher prevalence of ICA in healthy relatives bearing DR3 and/or DR4 antigen combinations compared to non-DR3 and non-DR4 individuals (14% versus 5%, P less than 0.05). Furthermore, ICA were more frequent in healthy siblings sharing two haplotypes compared with one or no haplotype (21% vs 10%, P less than 0.05). These results support the heterogeneity of the autoantibodies: ICA are related closely to diabetes, decline in frequency with the duration of the disease and show association with DR3 or DR4 and the number of HLA haplotypes shared with the proband; EPA are sex related, independent of the duration of diabetes, non-HLA linked, and clustered in families with parent-offspring overtransmission, reflecting an overlapping autoimmune background.     

A new approach to spike sorting for multi-neuronal activities recorded with a tetrode--how ICA can be practical

Multi-neuronal recording with a tetrode is a powerful technique to reveal neuronal interactions in local circuits. However, it is difficult to detect precise spike timings among closely neighboring neurons because the spike waveforms of individual neurons overlap on the electrode when more than two neurons fire simultaneously. In addition, the spike waveforms of single neurons, especially in the presence of complex spikes, are often non-stationary. These problems limit the ability of ordinary spike sorting to sort multi-neuronal activities recorded using tetrodes into their single-neuron components. Though sorting with independent component analysis (ICA) can solve these problems, it has one serious limitation that the number of separated neurons must be less than the number of electrodes. Using a combination of ICA and the efficiency of ordinary spike sorting technique (k-means clustering), we developed an automatic procedure to solve the spike-overlapping and the non-stationarity problems with no limitation on the number of separated neurons. The results for the procedure applied to real multi-neuronal data demonstrated that some outliers which may be assigned to distinct clusters if ordinary spike-sorting methods were used can be identified as overlapping spikes, and that there are functional connections between a putative pyramidal neuron and its putative dendrite. These findings suggest that the combination of ICA and k-means clustering can provide insights into the precise nature of functional circuits among neurons, i.e. cell assemblies.     

Electrophysiological analysis of the hippocampal-septal projections: I. Response and topographical characteristics

Summary The hippocampal input to the septal region was studied, in rats, by electrophysiological methods. The major observations were: 1. the hippocampal fibers are excitatory with respect to their target cells in the lateral septal nucleus (LSN); 2. the distribution of the subcallosal fornix is restricted to the dorsomedial regions of the LSN; 3. the fimbria contains fibers terminating throughout the dorsal aspect of the ipsilateral and contralateral septum; 4. there is an extensive overlapping in the distribution of the ipsi-and contra-lateral fimbria components with convergence upon single septal cells frequently seen; 5. the posterior part of hippocampal region CA1 contributes fibers to the fimbria as well as to the fornix.Abbreviations ACB Bed nucleus of the anterior commissure - AP Ref 0 The anterior-posterior and midline reference zero - CA Anterior commissure - CC Corpus callosum - CCA 3–4 Contralateral hippocampal fields CA-3–4 - Cd Pt Caudate putamen complex - CFim Contralateral fimbria - CMA Corticomedial maygdaloid division - GP Globus pallidus - Hipp Hippocampus - ICA1 (a) Anterior part of the ipsilateral hippocampal field CA1 - ICA1 (p) Posterior part of the ipsilateral hippocampal field CA1 - ICA3–4 Ipsilateral hippocampal field CA3–4 - IFim Ipsilateral fimbria - IFx Ipsilateral fornix - LSN Lateral septal nucleus - ME Microelectrode - MSN Medial septal nucleus - POA Pre-optic area This study was supported by U.S. Public Health Service Grants RR 5384 and NB 00405.     

Muscle Artifact Removal from Human Sleep EEG by Using Independent Component Analysis

Muscle artifacts are typically associated with sleep arousals and awakenings in normal and pathological sleep, contaminating EEG recordings and distorting quantitative EEG results. Most EEG correction techniques focus on ocular artifacts but little research has been done on removing muscle activity from sleep EEG recordings. The present study was aimed at assessing the performance of four independent component analysis (ICA) algorithms (AMUSE, SOBI, Infomax, and JADE) to separate myogenic activity from EEG during sleep, in order to determine the optimal method. AMUSE, Infomax, and SOBI performed significantly better than JADE at eliminating muscle artifacts over temporal regions, but AMUSE was independent of the signal-to-noise ratio over non-temporal regions and markedly faster than the remaining algorithms. AMUSE was further successful at separating muscle artifacts from spontaneous EEG arousals when applied on a real case during different sleep stages. The low computational cost of AMUSE, and its excellent performance with EEG arousals from different sleep stages supports this ICA algorithm as a valid choice to minimize the influence of muscle artifacts on human sleep EEG recordings.