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     

基于新模态-小波包分解的超宽带雷达生命体征信号去噪算法

超宽带雷达具有高分辨率,穿透能力强,低功耗等优势,工作时人体无需接触任何电极或传感器,可以穿透衣服、废墟等非金属介质在较远的距离内检测人体生命体征信息,在非接触式生命体征检测方面具有很重要的应用价值。由于人类心跳信号容易被呼吸谐波和其他噪声干扰,为了准确提取人体生命体征信号,提出一种基于改进的自适应噪声集合经验模态分解（ICEEMDAN）与小波包分解（WPD）结合的生命体征信号去噪方法。先通过超宽带雷达测量待测者的生命体征,获取人体所在空间位置提取出体表微动信号,对体表振动信号进行补偿与欠采样处理;利用ICEEMDAN-WPD的阈值去噪方法对微动信号进行模态分解,选取合适的模态分量去噪并进行重构,获取人体心跳微动信号的时频特征。实验结果表明,该算法相较于传统的去噪算法将相关系数提高到了0.940 5,信噪比提高到了9.093 8 dB,保留更多的生命体征信息的同时拥有更高的信噪比,可有效应用于生命体征检测领域。     

用于多分量信号分析的自适应广义S变换算法研究

时频分析是处理非平稳信号强有力的工具,S变换作为传统的时频分析方法之一,其窗函数的尺度可以随频率改变。但是,其时频窗函数尺度变化是固定的,无法适用不同信号的局部特性,导致能量聚集性较差。本文提出了一种自适应的广义S变换算法,设计了由4个调节参数控制的广义高斯窗函数,采用浓度测量自适应优化调节参数,以寻求最佳的时频表征效果。并针对时频分析结果,采用瞬时频率重组和分量重构方法,得到各个分量的瞬时频率,同时进行平滑处理,最终实现多分量信号的参数估计。仿真实验说明,本文提出的自适应广义S变换算法,结合瞬时频率重组和分量重构信号方法,极大地提升了多分量信号的时频分辨率和信号分离的准确性。     

Signal processing in prolonged EEG recordings during intensive care

Methods for analyzing and displaying EEG signals are discussed. The increasing availability and affordability of powerful computer equipment makes possible the use of ever more sophisticated signal processing techniques, which extract relevant (but not readily discernible) information from long-term EEG recordings and can easily identify important features in the EEG. Whether these techniques are actually taken up in clinical practice is heavily dependent on how well they match clinical requirements. This article concentrates on requirements set in the context of long-term recordings in the ICU that demand the ability to process short-term discrete events as well as long-term trend information. A huge range of potentially useful signal processing techniques exists. This article illustrates the value of some of these techniques for ICU signals using the EEG recordings collected during the IMPROVE project     

基于离散小波变换的卷积自编码运动想象脑电信号的分类

左右手运动想象脑电信号(MI-EEG)分类准确率低,制约了相关脑-机接口技术的发展。实验采集了16名健康受试者的运动想象脑电信号,提出了一种基于离散小波变换(DWT)和卷积自编码(CAE)的运动想象脑电信号分类算法。利用离散小波变换将EEG转换成时频矩阵,输入到卷积自编码网络中进行脑电信号的特征分类。该算法在实验数据集和公开数据集上测试都得到了较好的分类结果,静息-想象左手、静息-想象右手、想象左手-想象右手3组EEG在实验数据集上分类准确率分别为97.36%、97.27%、86.82%,在公开数据集上分类准确率分别为99.30%、98.23%、92.67%。离散小波变换结合卷积自编码网络模型在左右手运动想象脑电信号分类应用中比其他深度学习方法(CNN、LSTM、STFT-CNN)性能更优。     

Time-frequency analysis of electrophysiology signals in epilepsy

Many of the most powerful and effective algorithms in signal processing start with the assumption of stationarity. In addition, the deterministic portion of the signal is often assumed to be composed of complex exponentials that are the solutions to linear time-invariant (LTI) differential equations. These assumptions are often valid enough to yield good results when the signals and systems involved result from engineering design which often assures compliance with these conceptualizations. Signals of biological origin often do not comply with these assumptions, however, resulting in disappointment when conventional techniques are used. Newly emerging techniques of time-frequency (t-f) analysis can provide new insights into the nature of biological signals. This article describes some results using reduced interference distributions (RIDs) in the analysis of biosignals recorded in human epilepsy. It is shown that RID analysis of these signals results in insights and research hypotheses which would be difficult or impossible to obtain using conventional techniques. This is not a general t-f review article, and it is beyond the scope of this article (and space limits) to discuss the many new t-f tools that are now appearing in the literature. This article demonstrates one application of RID analysis     

基于STFT的高压电气设备局放信号时频分析

为了分析高压电气设备的局部放电信号,介绍了基于短时傅里叶变换(STFT)的时频分析方法及其重要性,并使用2种基于Matlab的方法仿真分析了局放信号。结果显示,基于trfstft方法的STFT时频分析可很好的给出局放信号的时频变化规律及其强度并从噪声中分离出局放信号。这提供了一种对含噪的模拟局放信号去噪处理的方法,可帮助有针对性地减少其危害,并进行目标识别和故障诊断。     

基于ADALINE神经网络的电能质量监测终端设计

谐波频谱检测是电能质量监测仪器的核心功能,其检测频谱是进行各种电能质量特征值运算的前提。传统以STFT为检测算法的电能质量终端由于时间窗固定,不具有暂态情况下的谐波分析能力。而诸如小波变换、S变换等算法则由于运算量巨大不利于谐波实时在线监测。针对这种情况,设计了一种以DSP为处理器,基于自适应线性神经网络(ADALINE)的电能质量监测终端。详细介绍信号接口电路、调理电路、PLL倍频电路、AD转换电路的硬件设计,给出了自适应线性神经算法推导和DSP数据处理框图。实验表明,所构建系统在运算量不大的情况下增强了暂态谐波测量能力,同时利用ADALINE误差信号可对电压暂降进行精确时间点定位。     

Adaptive control of non‐linear systems with unknown virtual control coefficients

This paper deals with state feedback adaptive control of parametric‐strict‐feedback (triangular) non‐linear systems with unknown virtual control coefficients. A priori knowledge of the signs of the virtual coefficients is not required, and control signals and adaptive laws are smooth. Asymptotic tracking of smooth reference signals is achieved while all the variables remain bounded. The proposed algorithms make use of backstepping and tuning functions, and enlarge the class of non‐linear systems with unknown parameters for which asymptotic output tracking can be achieved. Copyright © 2000 John Wiley & Sons, Ltd.     

Time-Frequency Excision for GNSS Applications

Interference detection and mitigation in global navigation satellite systems (GNSSs) are important issues for both military and civilian applications. In this paper, a novel time-frequency algorithm for GNSS application is proposed. The use of infinite impulse-response notch filters for the interference excision is introduced and analytical formulas for the detection of the disturbing signals are derived. The proposed method is tested by simulations and compared with time-frequency excision algorithms reported in literature, proving its effectiveness for interference removal.     

Proposed network structures and combined adaptive algorithms for electrocardiogram signal denoising

An electrocardiogram (ECG) signal is a record of the electrical activities of heart muscle and is used clinically to diagnose heart diseases. An ECG signal should be presented as clear as possible to support accurate decisions made by doctors. This article proposes different combinations of combined adaptive algorithms to derive different noise-cancelling structures to remove (denoise) different kinds of noise from ECG signals. The algorithms are applied to the following types of noise: power line interference, baseline wander, electrode motion artifact, and muscle artifacts. Moreover, the results of the suggested models and algorithms are compared with those of conventional denoising tools such as the discrete wavelet transform, an adaptive filter, and a multilayer neural network (NN) to ensure the superiority of the proposed combined structures and algorithms. Furthermore, the hybrid concept is based on dual, triple, and quadruple combinations of well-known algorithms that derive adaptive filters, such as the least mean squares, normalized least mean squares and recursive least squares algorithms. The combinations are formulated based on partial update, variable step-size (VSS), and second iterative VSS algorithms, which are considered in different combinations. In addition, biased NN and unbiased linear neural network (ULNN) structures are considered. The performance of the different structures and related algorithms are evaluated by measuring the post-signal-to-noise ratio, mean square error, and percentage root mean square difference.     

A time-series prediction approach for feature extraction in a brain-computer interface.

This paper presents a feature extraction procedure (FEP) for a brain-computer interface (BCI) application where features are extracted from the electroencephalogram (EEG) recorded from subjects performing right and left motor imagery. Two neural networks (NNs) are trained to perform one-step-ahead predictions for the EEG time-series data, where one NN is trained on right motor imagery and the other on left motor imagery. Features are derived from the power (mean squared) of the prediction error or the power of the predicted signals. All features are calculated from a window through which all predicted signals pass. Separability of features is achieved due to the morphological differences of the EEG signals and each NNs specialization to the type of data on which it is trained. Linear discriminant analysis (LDA) is used for classification. This FEP is tested on three subjects off-line and classification accuracy (CA) rates range between 88% and 98%. The approach compares favorably to a well-known adaptive autoregressive (AAR) FEP and also a linear AAR model based prediction approach.     

采用Hilbert-Huang变换和支持向量机的过电压分类识别方法(英文)

This paper proposes an effective method for over-voltage classification based on the Hilbert-Huang transform(HHT) method.Hilbert-Huang transform method is composed of empirical mode decomposition(EMD) and Hilbert transform.Nine kinds of common power system over-voltages are calculated and analyzed by HHT.Based on the instantaneous amplitude spectrum,Hilbert marginal spectrum and Hilbert time-frequency spectrum,three kinds of over-voltage characteristic quantities are obtained.A hierarchical classification system is built based on HHT and support vector machine(SVM).This classification system is tested by 106 field over-voltage signals,and the average classification rate is 94.3%.This research shows that HHT is an effective time-frequency analysis algorithms in the application of over-voltage classification and identification.     

基于AOK TFR理论的电力系统暂态信号分析新方法

提出一种基于自适应最优核时频分布(AOKTFR)理论的电力系统暂态信号特征分析新方法。该方法首先对信号求取模糊函数,并在模糊域采用自适应最优高斯核函数抑制交叉项,然后求模糊函数的二维傅里叶变换,从而得到仅有自项分量的时频分布,并在此基础上实现时频平面脊信息的提取。分别以绕线式异步电动机转子绕组单相开焊故障和变压器内部、外部故障的区分为例进行仿真。仿真结果表明:该方法适用于多分量、时变的非平稳信号的分析,具有很好的时频聚集性、强自适应性和抗噪声性能,且不受交叉项的影响;脊信息的提取进一步提高了时频分辨率。     

A Prior Neurophysiologic Knowledge Free Tensor-Based Scheme for Single Trial EEG Classification

Single trial electroencephalogram (EEG) classification is essential in developing brain-computer interfaces (BCIs). However, popular classification algorithms, e.g., common spatial patterns (CSP), usually highly depend on the prior neurophysiologic knowledge for noise removing, although this knowledge is not always known in practical applications. In this paper, a novel tensor-based scheme is proposed for single trial EEG classification, which performs well without the prior neurophysiologic knowledge. In this scheme, EEG signals are represented in the spatial-spectral-temporal domain by the wavelet transform, the multilinear discriminative subspace is reserved by the general tensor discriminant analysis (GTDA), redundant indiscriminative patterns are removed by Fisher score, and the classification is conducted by the support vector machine (SVM). Applications to three datasets confirm the effectiveness and the robustness of the proposed tensor scheme in analyzing EEG signals, especially in the case of lacking prior neurophysiologic knowledge.     

EEG changes accompanying learned regulation of 12-Hz EEG activity

We analyzed 15 sessions of 64-channel electroencephalographic (EEG) data recorded from a highly trained subject during sessions in which he attempted to regulate power at 12 Hz over his left- and right-central scalp to control the altitude of a cursor moving toward target boxes placed at the top-, middle-, or bottom-right of a computer screen. We used infomax independent component analysis (ICA) to decompose 64-channel EEG data from trials in which the subject successfully up- or down-regulated the measured EEG signals. Applying time-frequency analysis to the time courses of activity of several of the resulting 64 independent EEG components revealed that successful regulation of the measured activity was accompanied by extensive, asymmetrical changes in power and coherence, at both nearby and distant frequencies, in several parts of cortex. A more complete understanding of these phenomena could help to explain the nature and locus of learned regulation of EEG rhythms and might also suggest ways to further optimize the performance of brain-computer interfaces.     

Evaluating time-varying heart-rate variability power spectraldensity

A multiple weighted-least-square (WLS) identification process is presented for recognizing changes in ICU patient status. An adaptive scheme for the WLS is proposed in which the forgetting factor is automatically driven by the signal characteristics. Generally, adaptive algorithms are more complex and time-consuming than standard WLS, but they show a high tracking performance combined with the benefit of parameter smoothing. Nevertheless, the use of parameter-explicit filtering significantly reduces the computation time. This is a relevant advantage for real-time implementation. This adaptive approach also provides additional information to identify the signal variation speed, which can be used to localize transient phenomena. This article presents the algorithm performance in individuating and tracking the modifications of the cardiac autonomic control. To make data interpretation easier, the time-frequency distributions obtained are displayed as spectrograms. In addition, the signal speed variation is used to draw the attention of the physician to transient episodes     

A non-linear spatial hearing model based on bases pursuit algorithm

In the research on spatial hearing and realization of virtual auditory space, it is important to effectively model the head-related transfer functions (HRTFs) or head-related impulse responses (HRIRs). In our study, we managed to carry out adaptive non-linear approximation in the field of wavelet transformation. The results show that the HRIRs’ adaptive non-linear approximation model is a more effective data reduction model, is faster, and is 5 dB on average better than the traditional principal component analysis (PCA) (Karhunen-Loève transform) model based on relative mean square error (MSE) criterion. Furthermore, we also discussed the best bases’ choice for the time-frequency representation of HRIRs, and the results show that local cosine bases are more propitious to HRIRs’ adaptive approximation than wavelet and wavelet packet base. However, the improved effect of local cosine bases is not distinct. Here, for the sake of modeling the HRIRs more truthfully, we consider choosing optimal time-frequency atoms from redundant dictionary to decompose this kind of signals of HRIRs and achieve better results than all the previous models.