基于心音传感阵列ICA 信号处理的冠心病诊断的研究

通过研究冠脉血流动力学和心脏心音产生的机理，首次提出了将独立分量分析（ICA）方法应用于心音信号处理并达到自动检测冠心病的目的。在本系统中，信号采集系统采用了高灵敏度传感器列阵对正常人及冠心病患者胸部的多个部位进行检测。经预处理后的信号最后通过计算机进行数据采集。应用独立分量分析的方法将心脏舒张期的心音信号进行分离，并将各心音分量的统计特征参数作为输入参量输入到径向其函数网络（RBF网络）进行训练和识别。实验结果说明，独立分量分析结合人工神经网络的心音信号的分析方法是一种较为有效的诊断冠状动脉疾病的无创伤方法。     

基于双门限的第一、第二心音自动识别方法

为提高心音检测算法对异常心音的识别率,提出一种基于双门限的第一心音(S1)和第二心音(S2)自动识别方法,通过海明窗进行滤波预处理,采用改进型希尔伯特-黄变换提取心音包络,利用双门限法对心音进行分段,使用临床知识对S1和S2进行自动识别。实验结果表明,该方法能够准确识别正常心音和异常心音中的S1和S2。     

基于Hilbert-Huang Transform的心音信号谱分析

心音信号是一种典型的非平稳信号,传统信号处理方法的应用受到很大限制.针对此本文提出了基于Hilbert-Huang Transform(HHT) 的心音信号的分析方法,对冠心病患者的心音信号进行了分析.通过把心音信号分解为内蕴模式函数,利用Hilbert变换建立了心音信号的时间-频率-能量三维Hilbert谱分布以及边界谱分布;Hilbert谱及其边界谱在时域以及频域以较高的分辨率表征了心音信号的时频变化特性,揭示了冠心病患者心音信号的病理特征;为冠心病的早期无损诊断奠定了坚实基础,临床实践中有较大的指导价值.     

基于Hilbert2Huang Transform 的心音信号谱分析

心音信号是一种典型的非平稳信号,传统信号处理方法的应用受到很大限制。针对此本文提出了基于 Hilbert - Huang Transform(HHT) 的心音信号的分析方法,对冠心病患者的心音信号进行了分析。通过把心音信号分 解为内蕴模式函数,利用Hilbert 变换建立了心音信号的时间- 频率- 能量三维Hilbert 谱分布以及边界谱分布; Hilbert 谱及其边界谱在时域以及频域以较高的分辨率表征了心音信号的时频变化特性,揭示了冠心病患者心音信 号的病理特征;为冠心病的早期无损诊断奠定了坚实基础,临床实践中有较大的指导价值。     

基于ARM9和CPLD的四导心音采集显示系统的设计

介绍一个基于ARM920T内核的S3C2440A微处理器和CPLD芯片EPM570T144C5N实现的四导心音采集显示系统。系统的模拟电路部分对心音信号进行放大去噪;数字部分通过CPLD控制高速A/D转换并通过模拟SPI接口将数据传给ARM9。选用Linux2.6.30作为操作系统,设计了ARM9的SPI驱动和基于Qt4.7以及Qwt的应用程序。实际运行情况表明,本系统运行流畅,实现了实时采集心音信号、显示波形、储存心音数据。     

基于HHT和PPA的心音分段算法研究

心音是人体的一种重要的生理信号,它含有大量关于心脏病理状况的相关信息,反映了心脏及心血管结构及生理和病理信息。针对能否有效地提取第一心音(S1)、第二心音(S2),从而判断心脏是否病变,并且作为后续研究的基础,提出基于HHT和PPA的心音分段算法,包括首先利用希尔伯特-黄变换(HHT)进行心音包络的提取,然后利用中值滤波对包络进行平滑处理,最后通过峰逐层算法(PPA)来消除多余的低幅度峰值。通过对40例心音进行分段处理,可以对其中的39例进行正确分段。结果证明这种方法可以有效地提取心音信号的S1、S2,为后期的识别研究奠定了良好的基础。     

基于小波分析和神经网络的心音信号研究

针对传统的冠心病诊断方法具有不准性或有创性问题,积极广泛开展冠心病无损检测的研究,提高诊断准确性,为大众提供方便可行的检测手段是十分必要的。在分析冠状动脉堵塞与心音信号关系的基础上,研究心音信号的预处理,对心音信号进行去噪和定位分段;利用ARMA模型及功率谱估计对心音信号进行分析研究,提取冠心病病理特征;通过神经网络对心音信号进行分类,实现冠心病的智能无损诊断。实验结果表明,采用上述方法进行冠心病无损诊断准确率达到85.1%,为临床上的冠心病的无损诊断提供了应用基础。     

西门子传动产品线在2-3年内将作重大调整

在第八届国际现代工厂/过程自动化技术与装备展览会(2004FA/PA)期间，西门子公司召开专题产品推广会，并在会上透露，在未来的2-3年内西门子公司的传动产品线将作重大调整，现有的所有变频传动产品(如MM3系列、MM4系列、611D等)都将被淘汰，取而代之的是SINAMICS品牌，该品牌又分为G系列(G110、G130、G150)和S系列(S120、S150)，     

基于DSP的数字心音信号检测系统设计

分析了心音信号的产生机理、信号成分及心音的临床诊断价值。根据人体心音信号噪声强、信号弱、随机性强、容易受到外界干扰等特点,设计了基于DSP的心音信号数字检测系统,该系统由心音传感器、放大电路、滤波电路、A/D转换和DSP等部分组成;使用该系统先后在多家医院进行了临床心音信号采集,300多例心音样本采集实验表明,本系统可实现对微弱心音数据的实时采集、放大与有效滤波,采集系统可以满足对心音信号的检测要求。     

S3正式迈入PCI Express时代GammaChrome S18抢先测试

在携DeltaChrame S4/S8系列重返独立型图形芯片市场后，S3日前又推出了首款PCI Express(以下简称PCI-E）界面的图形核心——GrmmaChrome C18(以下简称S18)，它是S3正式进入PCI-E市场的里程碑。S18定位于中端主流市场，S3宣称，     

Feature determination for heart sounds based on divergence analysis

Heart auscultation (the interpretation of heart sounds by a physician) is a fundamental component of cardiac diagnosis. It is, however, a difficult skill to acquire. In decision making, it is important to analyze heart sounds by an algorithm to give support to medical doctors. In this study, two feature extraction methods are comparatively examined to represent different heart sound (HS) categories. First, a rectangular window is formed so that one period of HS is contained in this window. Then, the windowed time samples are normalized. Discrete wavelet transform is applied to this windowed one period of HS. Based on the wavelet detail coefficients at several bands, the time locations of S1–S2 sounds are determined by an adaptive peak detector. In the first feature extraction method, sub-bands belonging to the detail coefficients are partitioned into ten segments. Powers of the detail coefficients in each segment are computed. In the second feature extraction method, the power of the signal in a window which consists of 64 samples is computed without filtering the HSs. In the study, performances of these two feature extraction methods are comparatively examined by the divergence analysis. The analysis quantitatively measures the distribution of vectors in the feature space.     

A new neural network structure for detection of coronary heart disease

We investigate the application of neural networks for the detection of Coronary Heart Disease (CHD). We have used a Neural Network (NN) on data from a self- applied questionnaire to implement a decision system designed to seek out high risk individuals in a large population. A Multi- Layered Perceptron (MLP) was trained with risk factors to distinguish CHD. We also describe a modification to the architecture of the neural network in which an extra layer of neurons is added at the input. We present possible interpretations of the weights of these neurons, and show how they can be used as a selection criteria for which questions to use as inputs. The technique is compared against other statistical methods. We go on to demonstrate the system's capability for detecting both the symptomatic and asymptomatic patient.     

ANFIS based decision support system for prenatal detection of Truncus Arteriosus congenital heart defect

Congenital Heart Disease or Defect (CHD) is one of the most crucial causes of neonatal mortality. According to the consensus reported by Cardiological society of India, CHD is responsible for around 10% of infant mortality in India. Clinical investigation of CHD is normally performed with ultrasound (US) imaging modality. It captures biological internal structures with improper boundary due to inherent speckle noise. The fetal heart particularly has thin wall chambers and hence this fact protrudes to be a main motivation to contrive a new Computer Aided Diagnostic Support System (CADSS) to diagnose prenatal CHD from US images. This proposed CADSS is the first framework implemented to diagnose the prenatal Truncus Arteriosus congenital heart defect (TACHD) from 2D US images. The system starts with pre-processing the clinical data-set utilizing Probabilistic Patch Based Maximum Likelihood Estimation (PPBMLE). Then the anatomical structures are highlighted from the pre-processed information, utilizing the Fuzzy Connectedness based image segmentation process. Then 32 diagnostic features are extracted by utilizing seven different feature extraction models. Amongst, a subset of potential features are selected by applying Fisher Discriminant Ratio (FDR) analysis. Finally, Adaptive Neuro Fuzzy Inference System (ANFIS) is built with the selected feature subset as classifier, to perceive and show clinical results of prenatal TACHD. The performance analysis of various classifiers is evaluated by using 10-fold cross validation process for the image data-set. Comparative results prove that the proposed classifier has the potential to produce the higher classification accuracy than existing classifiers.     

卷积神经网络应用于先心病心音信号分类研究

心脏听诊是先心病初诊和筛查的主要手段。传统心音分类算法普适性差，过程复杂，不利于将来实时化决策。采用1 800个心音信号对几种时间序列分类的主流深度学习网络进行训练，结果显示循环神经网络易出现过拟合；长短时记忆网络分类损失值0.257，准确率0.872；卷积神经网络损失值0.25，准确率0.896。实验表明卷积神经网络相比较其他两种网络具备更大的潜力。基于卷积神经网络的先心病分类算法，因训练样本量大，使网络普适性得到了保证。与其他分类器相比，CNN的另一个优势是其可自动提取特征。该研究有望用于机器辅助听诊。     

Online discovery of Heart Rate Variability patterns in mobile healthcare services

Recent years, advances in day-to-day wearable sensors have led to the development of low powered physiological sensor platforms, which can be integrated in body area networks, a new enabling technology for real-time health monitoring. The bottleneck in health state awareness is the algorithm that has to interpret the sensor data. Nowadays Coronary Heart Disease (CHD) is still the leading cause of death. Many classification techniques such as decision tree and neural networks proposed for an early detection of individual at risk for CHD are not able to continuously detect heart state based on sensor data stream. In this study, we propose an online three-layer neural network to recognize Heart Rate Variability (HRV) patterns related to CHD risk in consideration of daily activities. ECG sensor data is preprocessed using Poincaré plot encoding. Incremental learning is utilized to train the network with new data without forgetting the previously learned patterns. The algorithm is named Poincaré-based HRV patterns discovering Incremental Artificial neural Network (PHIAN). When a sample is presented, the nodes in the hidden layer of PHIAN compete for determining the node with the highest similarity to the input. Error variables associated with the neuron units are used as criteria for new node insertion in hopes of allowing the network to learn new patterns and reducing classification error. However, the node insertion has to be stopped in the overlapping decision areas. We suppose that the overlaps between classes have lower probability than the centric part of the classes. Therefore, after a period of learning we remove the nodes with no neighbor. Plus, the error probability density is taken into account instead of input probability density. Finally, the predictive capability of PHIAN is compared with three previous classification models, namely Self-Organizing Map (SOM), Growing Neural Gas (GNG), and Multilayer Perceptron (MLP) in terms of classification error and network structure. The results show that PHIAN outperforms the existing techniques. Our proposed model can be efficiently applied to early detection of abnormal conditions and prevent the abnormal becoming serious.     

基于主动学习不平衡多分类AdaBoost算法的心脏病分类

针对不平衡分类中小类样本识别率低问题,提出一种基于主动学习不平衡多分类AdaBoost改进算法。首先,利用主动学习方法通过多次迭代抽样,选取少量的、对分类器最有价值的样本作为训练集;然后,基于不确定性动态间隔的样本选择策略,降低训练集的不平衡性;最后,利用代价敏感方法对多分类AdaBoost算法进行改进,对不同的类别给予不同的错分代价,调整样本权重更新速度,强迫弱分类器"关注"小类样本。在临床经胸超声心动图（TTE）测量数据集上的实验分析表明：与多分类支持向量机（SVM）相比,心脏病总体识别率提升了5.9%,G-mean指标提升了18.2%,瓣膜病（VHD）识别率提升了0.8%,感染性心内膜炎（IE）（小类）识别率提升了12.7%,冠心病（CAD）（小类）识别率提升了79.73%;与SMOTE-Boost相比,总体识别率提升了6.11%,G-mean指标提升了0.64%,VHD识别率提升了11.07%,先心病（CHD）识别率提升了3.69%。在TTE数据集和4个UCI数据集上的实验结果表明,该算法在不平稳多分类时能有效提高小类样本识别率,并且保证其他类别识别率不会大幅度降低,综合提升分类器性能。     

An economic analysis of an automated system for interpreting electrocardiograms

For the past year routine electrocardiograms (ECGs) from the Heart Station at U.S. Naval Hospital, San Diego, California have been automatically interpreted using the Smith/Mayo computer program with an IBM 360/65 at the Naval Electronics Laboratory Center, San Diego, California. The ECGs are batch processed. In the near future, an on-line, automated ECG interpretation system will be centralized in the Heart Station and have the capability to process ECGs for all U.S. Navy medical facilities in Southern California. This paper discusses the economic implications of such a system and compares its costs with those of a manual system, the existing batch mode system and commercial ECG service bureaus for similar ECG workload volumes. Finally, an evaluation of the accuracy of the Smith/Mayo program's interpretation of 4,628 normal and abnormal ECGs is presented.     

Identification of sleep stages from heart rate variability using a soft-decision wavelet-based technique

This work is concerned with a new technique to find identification factors for the different sleep stages based on a soft-decision wavelet-based estimation of power-spectral density (PSD) contained in the main frequency bands of Heart Rate Variability (HRV).A wavelet-based PSD distribution of HRV in different sleep stages is implemented on an epoch basis. Four sleep stages (S1–S4), “REM sleep” (with “rapid eye movements”), and wakefulness are considered in this work. The data used, including electro-cardiograms and sleep stage monitoring hypnograms, are provided by the sleep laboratory of the department of Psychiatry and Psychotherapy of Christian-Albrechts University Kiel, Germany. The data, taken from 12 healthy people and containing enough epochs of the above 5 different sleep stages plus the wake state, is divided into almost equal sets for training and test.The results show that the PSD of the very-low-frequency (VLF) band and the low-frequency (LF) band are reduced as sleep stages vary from the wake state to REM sleep and further to light sleep (S1–S2) and deep sleep (S3–S4). The variation of the PSD in the high-frequency (HF) band is almost the opposite. The ratio of the VLF/HF PSD is found to be a good identification factor between the different sleep stages, showing better results than other, commonly used factors such as the LF/HF and VLF/LF PSD ratios.     

Economic load dispatch problems with valve-point loading using natural updated harmony search

In this paper, the update process of harmony search (HS) algorithm is modified to improve its concept of diversity. The update process in HS is based on a greedy mechanism in which the new harmony solution, created in each generation, replaces the worst individual in the population, if better. This greedy process could be improved with other updates mechanisms in order to control the diversity perfectly. Three versions of HS have been proposed: (1) Natural Proportional HS ; (2) Natural Tournament HS; (3) Natural Rank HS. These three HS versions employed the natural selection principle of the “survival of the fittest”. Instead of replacing the worst individual in population, any individual can be replaced based on certain criteria. Four versions of economic loading dispatch (ELD) problems with valve point have been used to measure the effect of the newly proposed HS versions. The results show that the new HS versions are very promising for ELD domain. This claim is proved based on the comparative evaluation process where the new HS versions are able to excel the state-of-the-art methods in almost ELD problems used.