A time-frequency approach for newborn seizure detection

Techniques previously designed for seizure detection in newborns using the electroencephalogram (EEG) have been relatively inefficient due to their assumption of local stationarity of the EEG. To overcome the problem raised by the nonstationarity of the EEG signal, current methods are extended to a time-frequency approach. This allows the analysis and characterization of the different newborn EEG patterns that are intended to be the first step toward an automatic time-frequency seizure detection and classification. An in-depth analysis of both the autocorrelation and spectrum seizure detection techniques identified the detection criteria that can be extended to the time-frequency domain. The selected method uses a high-resolution reduced interference time-frequency distribution referred to as the B-distribution (BD). Here, the authors present the various patterns of observed time-frequency seizure signals and relate them to current knowledge of seizures. In particular, initial results indicate that a quasilinear instantaneous frequency (IF) can be used as a critical feature of the EEG seizure characteristics     

An enhanced time-frequency-spatial approach for motor imagery classification.

Human motor imagery (MI) tasks evoke electroencephalogram (EEG) signal changes. The features of these changes appear as subject-specific temporal traces of EEG rhythmic components at specific channels located over the scalp. Accurate classification of MI tasks based upon EEG may lead to a noninvasive brain-computer interface (BCI) to decode and convey intention of human subjects. We have previously proposed two novel methods on time-frequency feature extraction, expression and classification for high-density EEG recordings (Wang and He 2004; Wang, Deng, and He, 2004). In the present study, we refined the above time-frequency-spatial approach and applied it to a one-dimensional "cursor control" BCI experiment with online feedback. Through offline analysis of the collected data, we evaluated the capability of the present refined method in comparison with the original time-frequency-spatial methods. The enhanced performance in terms of classification accuracy was found for the proposed approach, with a mean accuracy rate of 91.1% for two subjects studied.     

Preprocessing and time-frequency analysis of newborn EEG seizures

Neurological disease or dysfunction in newborn infants is often first manifested by seizures. Prolonged seizures can result in impaired neurodevelopment or even death. In adults, the clinical signs of seizures are well defined and easily recognized. In newborns, however, the clinical signs are subtle and may be absent or easily missed without constant close observation. This article describes the use of adaptive signal processing techniques for removing artifacts from newborn electroencephalogram (EEG) signals. Three adaptive algorithms have been designed in the context of EEG signals. This preprocessing is necessary before attempting a fine time-frequency analysis of EEG rhythmical activities, such as electrical seizures, corrupted by high amplitude signals. After an overview of newborn EEG signals, the authors describe the data acquisition set-up. They then introduce the basic physiological concepts related to normal and abnormal newborn EEGs and discuss the three adaptive algorithms for artifact removal. They also present time-frequency representations (TFRs) of seizure signals and discuss the estimation and modeling of the instantaneous frequency related to the main ridge of the TFR     

基于图像增强处理的CDVS匹配算法

紧凑型视觉描述子（CDVS）的目标是针对移动端的图像检索以及匹配应用提供一套标准化的比特流语法。CDVS标准算法对于光照条件良好的图像具有很好的匹配效果,但是对于昏暗条件下拍摄的图像,匹配的准确度表现不足。因此提出一种基于直方图均衡化的CDVS匹配算法,对昏暗条件下的图像通过直方图均衡算法进行质量提升,增加关键点匹配数,然后再利用CDVS标准算法对图像进行匹配。另一方面,提出一种基于同态滤波的CDVS匹配算法,对昏暗图像进行频域变换,突出高频信号,抑制低频信号,增加图像对比度。实验对比了昏暗条件下图像匹配结果与处理之后图像匹配结果,验证了本算法的有效性。     

基于 RSVP 的面向不良信息检测人机协作系统研究

针对复杂的环境背景下不良信息的快速准确检测问题,提出了基于快速序列视觉呈现( rapid serial visual presentation, RSVP)的面向不良信息检测人机协作系统。 首先利用快速佩戴便携式采集系统采集了 12 名受试者的脑电数据;然后采用 Mallat 算法提取较低维度的时频特征,使用人工神经网络(ANN)和支持向量机(SVM)两种模型分类对比;最后在训练集中引入 不同次数的叠加平均数据以改善模型的分类性能。 实验结果表明,在含有 3 个目标的 60 张图像中平均正确输出至少 2 张目 标,AUC 值达到了 0. 9。 该系统在小批量数据集、环境变化复杂的不良图像信息检测中有着良好的性能,相较于人工检测提高 了效率。     

HHT方法在转子振动故障诊断中的应用

传统的振动故障诊断主要是基于频谱分析的方法，而Hilbert-Huang变换得出的时频图是分析故障信号奇异性的有效工具，也是非平稳信号比较有力的分析工具。Hilbert- Huang变换方法以经验模态分解为基础，使信号变换后得到的瞬时频率具有物理意义。该文通过几种时频分析方法如魏格纳-维尔(Wigner-Ville)分布、小波变换等比较，发现Hilbert-Huang变换的时频分析更能够清楚给出时频分布情况，准确反映系统的固有特性。对转子实验台几种典型振动故障信号进行了分析研究，结果表明，利用Hilbert-Huang变换的时频分析不仅能直观检测信号中的微弱奇异成分，而且可以有效地进行故障诊断，实现早期故障预报。该方法为旋转机械状态检测和故障诊断提供了新的手段。     

Nonstationary Motor Fault Detection Using Recent Quadratic Time–Frequency Representations

As the use of electric motors increases in the aerospace and transportation industries where operating conditions continuously change with time, fault detection in electric motors has been gaining importance. Motor diagnostics in a nonstationary environment is difficult and often needs sophisticated signal processing techniques. In recent times, a plethora of new time-frequency distributions has appeared, which are inherently suited to the analysis of nonstationary signals while offering superior frequency resolution characteristics. The Zhao-Atlas-Marks distribution is one such distribution. This paper proposes the use of these new time-frequency distributions to enhance nonstationary fault diagnostics in electric motors. One common myth has been that the quadratic time-frequency distributions are not suitable for commercial implementation. This paper also addresses this issue in detail. Optimal discrete-time implementations of some of these quadratic time-frequency distributions are explained. These time-frequency representations have been implemented on a digital signal processing platform to demonstrate that the proposed methods can be implemented commercially.     

Wavelet transform-based signal waveform discrimination for inspection of rotating machinery

This paper presents a new discrimination procedure of signal waveforms based on wavelet theory for the inspection of rotating machinery. The wavelet transform decomposes signals into time-frequency space, rather than mere frequency space, limited by the uncertainty principle. This decomposition permits time-frequency analyses and provides a more flexible means of signal processing than before. To examine a rotary compressor pump, particular waves in the rotational load torque signals that correlate with failure modes are discriminated from one another and evaluated. To extract the focal waves, the signal is decomposed with wavelets and then only particular waves, such as impulses, are reconstructed from a selected set of wavelet coefficients. This is called time-frequency space filtering. The wavelet local modulus maxima are used to pen a time-frequency window through which only the focal waves can pass with high fidelity. The maxima have information of the inflection points of the wave at each resolution that represent its waveform. The experimental results show the effectiveness of the procedure.     

一种基于重排二次型时频分布的电能质量检测新方法

电能质量扰动信号是一种典型的非平稳信号,采用二次型时频分布能够获得其时间频率联合特性。提出一种基于重排二次型时频分布的电能质量检测新方法,首先采用瞬时无功功率理论和广义形态滤波器将电能质量信号的基波成分和扰动成分分离,再利用重排二次型时频分布对扰动分量进行分析,从而获得时频聚集型更好的扰动分量的时频联合分布,直观地表达出扰动信号的时频特性。仿真算例验证了此方法对各种常见电能质量扰动和交叉电能质量扰动的检测和特征提取是有效的。     

Unconstrained monitoring of body motion during walking

Discusses using the matching pursuit algorithm to characterize time-frequency patterns of body motion in poststroke hemiplegic patients. We have been working on the quantification of body motions in healthy young and elderly subjects, patients with Parkinson's disease (PD), and poststroke hemiplegic (PSH) patients using an accelerometry technique and advanced signal processing methods. In this article, we use the matching pursuit (MP) algorithm to characterize the time-frequency patterns of the acceleration signal recorded from both healthy subjects and poststroke hemilpegic patients. The MP algorithm was chosen since it provides better time and frequency resolutions than other time-frequency analysis methods and is an algorithm that decomposes any signal into several already-known time-frequency patterns, which are called atoms. It also provides detailed information about each time-frequency pattern including its energy, time and frequency localization, and phase and scale (time duration), which can be used for the comparison and the statistical analysis.     

Extracting evoked potentials with the singularity detectiontechnique

For a better and faster method of extracting EPs, we study the difference between the EP signal singularities and the EEG noise singularities. The ensemble-averaging operation is based on the fact that the EEG can be looked upon as white noise. The singularity detection (SD) technique that we discuss can adequately remove white noise from the signal. We found that there was a very large difference between the EP signal singularities and the EEG noise singularities. The local maxima of the wavelet-transform modulus provide enough information to analyze these singularities. We can extract the EP signal components from the EEG noise by selecting the wavelet-transform modulus maxima that correspond to the EP signal singularities. After removing the modulus maxima of the EEG noise fluctuations, we are able to reconstruct a denoised EP signal     

小波－奇异值分解在异步电机转子故障特征提取中的应用

针对电流信号中异步电机的转子故障特征分量经常被电源频率分量淹没而无法准确检测的缺点,提出了一种基于小波-奇异值分解的转子故障特征提取方法。通过连续小波变换将电流信号中的各特征频率分量转换到时频分布空间中,对该时频空间进行奇异值分解将各特征频率分量分解到不同的正交特征子空间中,对特征子空间的选择重构可以有效地滤除电源频率分量而提取出转子故障特征分量。模拟数据和实际故障信号的应用表明,该方法提供了一种可实际应用的异步电机转子故障诊断方法。     

HHT在复合材料损伤声发射信号处理中的应用

为了研究Hilbert-huang在复合材料损伤声发射信号处理领域的可行性和有效性,介绍了小波变换和Hilbert-huang变换的基本思想和特点,然后对碳/环氧复合材料进行了拉伸损伤实验,提取其声发射信号,并分别利用HHT和小波变换进行了处理.通过对比发现,HHT在处理复合材料声发射信号方面具有更强的自适应性和更高的...     

Artifact cleaning of motor imagery EEG by statistical features extraction using wavelet families

Electroencephalogram (EEG) and its sub-bands represent electrical pattern of human brain. EEG signal contains transient components, spikes, and different types of artifacts due to eye blinking, movement of the person, anxiety, and so forth, during EEG capture. Wavelet transforms are powerful mathematical tool for sampling approximation to get clean EEG. It also helps in filtering, sampling, interpolation, noise reduction, signal approximation and signal enhancement, and feature extraction. In this paper, we have analyzed artifact cleaning via PSD graphs and statistical features extracted from motor imagery EEG-like standard deviation variance. For this, we considered 19 channels EEG signal and applied orthogonal Daubechies wavelet, bi-orthogonal rbio wavelet and Coifman wavelets to check the better performance of different wavelets. Coifman wavelet uses both scaling function and vanishing moments for sampling approximation and hence give smooth sampling compared to rbio and Daubechies wavelet transforms. Coif is a compactly supported wavelet system which also helps in smooth sampling approximations than other wavelets in the state of arts. The detailed coefficients and approximate coefficients can be further used for extracting features from EEG and classification purposes. Artifacts cleaning is thus observed better in coif wavelet analysis compared to other wavelets from the power distributions as power spectral density (PSD) graphs, standard deviation and variance obtained. Matlab R2013b is used for filtering and sampling EEG. Python 2.7 is used for statistical features extraction.     

Time--frequency analysis of biosignals

The wavelet transform has a powerful time-frequency analysis and signal-coding tool suitable for use in the manipulation of complex nonstationary signals. This article provides an overview of the emerging role of wavelet-transform analysis in biomedical signal processing and analysis. It also provides a brief overview of the theory of the transform in its two distinct and very different forms: continuous and discrete. In conclusion, it has been shown that the wavelet transform is a flexible time-frequency decomposition tool that can form the basis of useful signal analysis, and coding schemes. It is envisaged that the future will see further application of the wavelet transform to biomedical signal analysis, as the emerging technologies based on them are honed for practical purposes.     

Coupled Oscillatros: Complex But Not Complicated Cardiovascular and Brain Interactions

In coupled nonlinear oscillators approach, the framework that has been used for the studies of cardiovascular and brain oscillations. As background, it describes the human CVS and present results of time-frequency analysis using wavelet transforms of several noninvasive measurements of cardiovascular signals. Studies of neuronal oscillations have been undertaken since the first human electroencephalographic (EEG) recording, and the recent resurgence of interest in neuronal oscillations. It is concluded that interactions occur between the oscillatory processes, both within and between the cardiovascular and the neuronal systems. The strengths and directions of these interactions may be used, in principle, for characterization of the state of the organism as demonstrated here for the case of deep anesthesia.     

Digital processing of EEG signals

The digitization of electroencephalogram (EEG) signal data is the essential first step in using computers to analyse and manipulate EEG data. EEG signals are inherently complicated due to their nonGaussian, nonstationary, and often nonlinear nature as shown by most of the articles of this special issue. On top of that, the small amplitude of these signals reinforce their sensitivity to various artifacts and noise sources. The aim of this special issue is to shed light onto the recent digital techniques for processing EEG signals ranging from storage and artifact removal to event detection/classification and prediction issues     

Performance Evaluation of a Precorrelation Interference Detection Algorithm for the GNSS Based on Nonparametrical Spectral Estimation

This paper deals with a signal processing precorrelation method to detect the presence of interference at a global navigation satellite system (GNSS) receiver site. In particular, a nonparametric spectral estimation approach based on the Welch windowed periodogram will be considered here. The performance of the proposed detector, in terms of detection probability, for a given false-alarm probability value, will be derived by means of an analytical approach and resorting to computer simulations. A performance comparison with previously proposed precorrelation methods will also be presented in order to highlight the better behavior of the proposed approach that makes it suitable for safety of life (SoL) applications.