【作者】 朱振宇；

【导师】 周洪利；

【作者基本信息】 北京交通大学 ， 计算机科学与技术， 2010， 硕士

【摘要】 近年来,心血管疾病已经成为威胁人类健康的主要疾病之一。ECG信号能直观反映心律波动状况,成为分析研究的热门领域。本课题是心电信号数据提取算法的设计与实现,与硬件检测相对应,这是实现心电数据检测与提取的软件手段。心电图中的QRs波群包含了心脏状态的重要信息,是进行其它心电参数检测(如P波、T波)的前提,因此QRS波群的准确识别与检测对于心脏疾病的预防和诊断具有重要的意义。本文对心电数据提取算法的研究是基于一款微型心电仪的应用需求,紧紧围绕心电仪的两种数据处理模式,即单机检测模式与联机数据处理模式展开算法的设计与实现。针对两种模式不同的应用环境与各异的运算存储能力,分别设计并实现了差分阈值法与小波变换法对心电信号中QRS波群的检测与定位。在差分阈值法的设计中,依据实时处理数据的要求,设计以短时程为单位的数据块处理模式,实现以时程不断更新提取数据的模式。在消除R波误判方面,实现了R波半波宽法的检测,提高了R波的识别准确度。算法在仿真环境中通过了运行检测。在小波变换法的设计中,运用小波多尺度分析的特征,使用二次微分Marr小波变换对心电数据按照Mallat算法进行R波的检测与提取。运用综合检测方法进行复检,如补偿法,从而在特定尺度上识别QRS波群的同时有效抑制高频噪声,如工频干扰,从而提高了R波检测的正确率。对于P波,T波的检测提出了思路与方法。本文使用国际上广泛采用的MIT-BIH心电数据作为算法处理的数据与验证算法准确度的标准。通过算法运行处理并将提取的数据结果与MIT-BIH提供的标准信息做比较,对算法的准确度做出客观科学的评价。测试结果表明,本文设计的两种算法准确率较高,能满足实际应用的要求。更多还原

【Abstract】 In recent years, as one of the major diseases, cardiovascular disease has become a threat to human health. ECG analysis is a hot research area because it can reflect the state of the heart fluctuations intuitively. This thesis is about the study and implementation of the algorithm in detecting characteristic points of ECG signals. Opposite to the hardware detection, it is the method of software in ECG data detection. QRS waves contain important information about the heart condition and it is the prerequisite of detecting the other ECG data such as the P wave and T wave, so accurately identifying the QRS waves make the important significance for heart disease prevention and diagnosis.The algorithms are based on the application of a Micro ECG Device for the two data processing patterns, one is single instrument detection and the other is detection with PC connection. The two different environments have different computing and storage capacity, so the differential threshold method and wavelet transform method are adapted to the detection and location of QRS waves in two patterns.In the implementation of the differential threshold method, based on real-time data processing, designed short-range time as a unit of data block mode, to achieve continuously updated the characteristic points. Using the half width of R wave detection method to eliminate the mistake of the R wave detection, the method can improve the R wave detection accuracy.In the implementation of the wavelet transform method, using the character in multi-scale analysis of wavelet, designed the Mallat algorithm based on dyadic Marr wavelet transform to detect the QRS waves. Using an integrative method to identify the result, it can detect the QRS waves in the specific scale with effective inhibition of high-frequency noise, such as power-line interference, the method can improve the R wave detection accuracy. The P wave and T wave detection ideas and methods are proposed.This thesis introduced the widely used MIT-BIH ECG data as the object of processing algorithms and standard to measure the accuracy of the algorithm. Making objective scientific evaluation with the algorithm by checking result compared with the standard annotation of MIT-BIH. Test result shows that the two designed algorithms can meet the practical applications with high accuracy.更多还原