基于二维云模型的心电信号ST段分析方法研究

发布时间：2018-03-14 05:34

本文选题：心电信号　切入点：ST段　出处：《郑州大学》2012年硕士论文　论文类型：学位论文

【摘要】：急性心肌梗死是危害中老年人身体健康的常见疾病,在其发病初期主要是通过心电图情况来进行检测,目前针对该病症的有效诊断及防治已成为医学界所面临的一项新的课题。以往常规的心电信号采集系统中均采用12导联进行采集,但多数患者在疾病突发时往往不在医院,会对诊断及救护带来不便,因此,本研究在开发一种便携式单导联心电监护手机的背景下,进行了单导联心电信号检测技术算法研究。该技术可以让患者随时随地的对心电图进行监测和诊断,成为了心电监护设备的一种发展方向。急性心肌梗死早期主要体现为心电图的ST段变化,因此心电信号ST段的正确识别对于急性心肌梗死的诊断具有重要的意义。ST段代表心室除极完成后复极过程的电位变化,极易受到心肌缺血、噪声等外界干扰,此外ST段形态在同一个体的不同导联情况下也存在差异,因此ST段的计算机自动识别成熟度远低于QRS波识别技术。正确识别ST段起始点、终止点、ST段形态特征及其电平测量较困难,且目前尚无统一测量标准。针对于ST段形态特征和检测难点,本文提出了一种基于二维云模型理论的心电信号ST检测方法。具体研究内容如下： (1)根据ST段所受的噪声特点,本文使用了零相位数字滤波方法,设计出一个9阶Chebyshev带通数字滤波器来对心电信号进行滤波降噪,通过滤波前后的信号频谱分析,验证了该滤波器在保持心电信号形态不失真(主要是ST段形态)的条件下,能够有效消除低频基线漂移和工频干扰。 (2)在心电信号特征参数的提取方面,对于心电信号QRS波,本文采用差分阈值法进行检测,并综合运用时间移动窗口、自适应等技术,提高了检测精度,克服了小波变换计算量大、神经网络模板训练时间长等缺点,比较适合于便携式单导联心电监护手机的开发；同时采用局域变换算法对ST段起始点、终止点进行提取,仿真实验结果显示算法具有较高的精度。 (3)针对心电数据模糊性和随机性较大的特点,本文提出了一种基于云模型的心电信号ST段的检测方法,能够通过待测信号对判别规则云隶属度大小的判断,来进行心电信号ST段形态判定。首先采用云模型对ST段内的大量采样点特征(数据点的电位值、一阶导数和二阶导数)所出现的频率进行聚类分析,获得具有自身特性的ST段特征综合云,进而利用云模型来描述几种ST段的不同特征。之后通过云变换以生成ST段判别规则云,将待检测的ST段数据特征作为输入,通过其对判别规则云隶属度的判断,来进行ST段形态的判别。本研究利用欧盟CSE心电数据库平台,采用Matlab对算法进行测试,结果验证了该算法有效可行,利用云模型所得到的ST段判别结果符合医学诊断逻辑思维。
[Abstract]:Acute myocardial infarction (AMI) is a common disease that endangers the health of middle-aged and old people. It is mainly detected by electrocardiogram in the early stage of the disease. At present, the effective diagnosis and prevention of the disease has become a new subject in the medical field. In the past, 12 leads were used in the conventional ECG acquisition system, but most of the patients were not in the hospital when the disease broke out. This study is based on the development of a portable, single-lead ECG monitoring cell phone. The algorithm of single lead ECG signal detection is studied, which enables patients to monitor and diagnose ECG at any time and anywhere, which has become a developing direction of ECG monitoring equipment. In the early stage of acute myocardial infarction (AMI), the changes of St segment are mainly reflected in the changes of St segment of electrocardiogram. Therefore, the correct recognition of St segment of ECG signal is of great significance for the diagnosis of acute myocardial infarction. St segment represents the potential changes in the repolarization process after ventricular depolarization is completed. It is easy to be interfered by external disturbance such as myocardial ischemia, noise and so on. In addition, the shape of St segment is different in different leads of the same body. Therefore, the maturity of automatic recognition of St segment is much lower than that of QRS wave recognition technology. It is difficult to measure the shape and level of St segment at the termination point, and there is no uniform measurement standard at present. In this paper, a novel ECG St detection method based on two-dimensional cloud model theory is proposed. 1) according to the characteristics of the noise in St segment, a 9-order Chebyshev band-pass digital filter is designed to filter and reduce the noise of ECG signal by using the zero-phase digital filter. The spectrum of the signal before and after filtering is analyzed. It is verified that the filter can effectively eliminate the low frequency baseline drift and power frequency interference under the condition that the ECG signal shape is not distorted (mainly St segment shape). In the aspect of extracting characteristic parameters of ECG signal, this paper uses differential threshold method to detect QRS wave of ECG signal, and synthetically uses time moving window and adaptive technology to improve detection accuracy. It overcomes the disadvantages of large computation of wavelet transform and long training time of neural network template, so it is more suitable for the development of portable single-lead ECG monitoring mobile phone. At the same time, local transform algorithm is used to extract St segment starting point and termination point. The simulation results show that the algorithm has high accuracy. 3) in view of the fuzzy and randomness of ECG data, this paper presents a method of ECG St segment detection based on cloud model, which can judge the membership degree of discriminant rule cloud by the signal to be tested. Firstly, the cloud model is used to cluster the frequency of a large number of sample points (potential value, first derivative and second derivative) in St segment. The St segment feature synthesis cloud with its own characteristics is obtained, and then the different features of several St segments are described by using cloud model, and then the St segment discriminant rule cloud is generated by cloud transformation, and the St segment data feature to be detected is used as input. By judging the membership degree of the discriminant rule cloud, the St segment shape is judged. In this study, the algorithm is tested by using the CSE ECG database platform of EU and Matlab. The results show that the algorithm is effective and feasible. The result of St segment discrimination obtained by cloud model accords with the logical thinking of medical diagnosis.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH776;R318.0

【参考文献】