基于脉搏波传导时间变异性的冠心病识别方法研究

发布时间：2018-10-18 18:16

【摘要】：冠心病以其发病率高、治愈率低等特点成为威胁人类身体健康的最大隐患之一,如果不能有效地预防治疗,冠心病将成为人类未来发展面临的严峻问题。随着智能医疗仪器的快速发展,给冠心病的诊治带来了便利,但是冠心病识别还存在准确率低、检测耗时较长且费用昂贵等缺陷。因此,研究一种实时性好、准确率高的冠心病识别方法显得尤为重要。心电信号和脉搏信号蕴含丰富的人体生理系统生理和病理信息,可以作为冠心病预防和识别的指标。结合心电脉搏信号检测出脉搏波传导时间变异性信号,脉搏波传导时间变异性信号能够反映冠脉病变的严重程度,以及自主神经系统的调节机制。实时分析脉搏波传导时间变异性得到相关信息,对冠心病的实时监护及预警具有重要意义。从冠心病的病理机理和临床诊断出发,综述了冠心病临床诊断方法和冠心病识别方法的国内外研究现状。在综述方法的基础上,利用冠心病发病过程中冠脉病变严重程度和自主神经调控原理,提出采用脉搏波传导时间变异性来实现冠心病的实时、准确的识别。解决了现有脉搏波传导时间变异性信号分析方法实时性和准确性顾此失彼的问题,以及现有只通过自主神经系统调控原理识别冠心病包涵信息单一的问题。本文主要工作如下:1)研究脉搏波传导时间变异性信号的提取方法,通过数据特征和现有方法的分析对比,确定同步心电信号R波峰值至脉搏信号主波峰值间的时间间隔序列,即为脉搏波传导时间变异性。针对心电脉搏信号在采集过程中引入的各种噪声和干扰,采用实时性较强的整系数滤波器进行滤波。2)针对目前脉搏波传导时间变异性分析方法的主观性强、实时性差等问题,结合脉搏波传导时间变异性的特点,在时域分析和非线性分析的基础上,采用滑窗迭代的思想对其改进。得到实时时域特征和实时非线性特征,通过实验分析得到改进后的特征具有较好的实时性,并且大部分特征具有较好的准确性。同时分析了脉搏波传导时间变异性的频谱信息,较心率变异性频谱能量分布更明显。3)根据脉搏波传导时间变异性信号的特点确定了各识别算法的模型参数,并通过识别准确率和算法运行时间进一步说明参数的重要性。采用t检验和主成分分析进行特征选择,有效保留原始特征信息的同时消减了数据维数,从而降低了识别算法的复杂度。通过实验对比分析,提出了一种兼顾准确性和实时性的冠心病识别方法。
[Abstract]:Coronary heart disease (CHD) has become one of the biggest hidden dangers to human health because of its high incidence and low cure rate. If it cannot be effectively prevented and treated, coronary heart disease will become a severe problem in the future development of human beings. With the rapid development of intelligent medical instruments, it brings convenience to the diagnosis and treatment of coronary heart disease, but the recognition of coronary heart disease still has the defects of low accuracy, long time consuming and expensive detection. Therefore, it is very important to study a recognition method of coronary heart disease with good real-time and high accuracy. ECG and pulse signals contain abundant physiological and pathological information of human physiological system and can be used as indicators of prevention and recognition of coronary heart disease. Pulse wave time variability signal can be detected by ECG pulse signal. Pulse wave time variability signal can reflect the severity of coronary artery disease and the regulation mechanism of autonomic nervous system. Real-time analysis of pulse wave time variability is of great significance to real-time monitoring and early warning of coronary heart disease. Based on the pathological mechanism and clinical diagnosis of coronary heart disease (CHD), the research status of clinical diagnosis and recognition of CHD at home and abroad is reviewed. On the basis of the review methods, using the severity of coronary artery disease and the regulation principle of autonomic nerve during the course of coronary heart disease, the pulse wave conduction time variability is proposed to realize the real-time and accurate recognition of coronary heart disease. The problem of real-time and accuracy of pulse wave time variability signal analysis method is solved, and the problem of single information recognition of coronary heart disease by the principle of autonomic nervous system regulation is also solved. The main work of this paper is as follows: 1) the extraction method of pulse wave conduction time variability signal is studied. Through the analysis and comparison of data characteristics and existing methods, the time series between the peak R wave of synchronous ECG signal and the peak value of main wave of pulse signal is determined. That is, pulse wave conduction time variability. Aiming at all kinds of noise and interference introduced in ECG pulse signal acquisition process, the filter with high real time integral coefficient is used to filter it. 2) aiming at the problems of strong subjectivity and poor real-time performance of current pulse wave conduction time variability analysis method, etc. According to the characteristics of pulse wave conduction time variability, based on the time domain analysis and nonlinear analysis, the sliding window iteration is used to improve it. The real-time domain feature and the real time nonlinear feature are obtained. The improved feature has good real-time performance and most of the features have good accuracy. At the same time, the spectrum information of pulse wave conduction time variability is analyzed, which is more obvious than heart rate variability spectrum energy distribution. 3) according to the characteristics of pulse wave conduction time variability signal, the model parameters of each identification algorithm are determined. The importance of the parameters is further explained by the recognition accuracy and the running time of the algorithm. T test and principal component analysis (PCA) are used for feature selection, which can effectively preserve the original feature information and reduce the dimension of the data, thus reducing the complexity of the recognition algorithm. Based on the comparative analysis of experiments, a recognition method of coronary heart disease (CHD) with both accuracy and real time is proposed.
【学位授予单位】：兰州理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R541.4;TN911.7

【参考文献】