正交小波变换的快速算法在心电QRS波检测中的应用  

Detection of ECG QRS Wave Using Fast Arithmetic of Orthogonal Wavelet Transforms

作　　者：林宛华[1] 汪家旺[2] 杨涛 陈真诚[1] 

机构地区：[1]中南大学信息物理工程学院生物医学工程研究所,湖南长沙410083 [2]南京医科大学第一附属医院,江苏南京210032 [3]南京普澳医疗设备有限公司,江苏南京210032

出　　处：《中国医学物理学杂志》2008年第6期903-906,910,共5页Chinese Journal of Medical Physics

基　　金：南京市政府科技计划项目(200701032)

摘　　要：目的:研究基于小波变换的心电QRS波检测的准确率、抗干扰性和实时性,论证其在实际工程应用中的可行性。方法:作者在比较了不同小波基的检测准确率之后,采用一种基于三次B样条小波变换的心电QRS波检测算法,利用离散正交二进小波的快速算法-Mallat算法进行分解滤波,再利用小波变换与信号奇异点的关系,在23尺度下识别R波峰值,在21尺度上检测QRS波的起点和终点,QRS波的起点和终点对应于小波变换的一对符号相反的模极大值,R波的峰点对应于介于这对模极大值之间的小波变换过零点,并用美国MIT/BIH心电标准数据库分析该算法的准确率、抗干扰性和实时性。结果:该方法具有比较理想的检测准确率,在99%以上;对肌电、工频、基漂等常见的心电信号干扰有较好的容限度,即使心电序列伴有严重的基漂和高频、工频、肌电等干扰,也不影响QRS波的检测;此外,三次B样条小波基的滤波器个数少,提高了运算速度,采样11.4s的数据进行分析,耗时为0.2s^0.3s,实时效果较明显。结论:可以满足实际工程应用的需要。Objective： Research the veracity,anti-jamming and real time of ECG QRS detection based on wavelet transform, demonstrate the feasibility of application to real project. Method： After comparing the different veracity of wavelet detective, we adopt a QRS wave detection arithmetic based on 3-order B-Spline wavelet transforms. In the processing, we use the fast arithmetic of discrete binary wavelet transforms-Mallat arithmetic. After detecting the peak of R wave in scale 2^3, using the relationship between wavelet transforms and signal singularity point, we detect the begin and the end of QRS wave in scale 2^1. The begin and the end of QRS wave correspond to a pair of modulus maxima with opposite sign, the R wave correspond to the zero-crossing of wavelet transform between this pair of modulus maxima. In the end, we analyse the veracity ,the ability of anti-jamming and the rate of this arithmetic through American MIT/BIH ECG database. Result： This method has a relatively high detective veracity, upwards of 99%. This method has a high tolerance to familiar ECG signal disturbance like electromyography,industrial frequency,base-line excursion. It would not influence the detection of QRS even the ECG sequence with serious base-line excursion,high frequency,industrial frequency and electromyography disturbanc. Besides, 3-order B-Spline wavelet filter has a small quantity of coefficient, that improve the calculate speed. It takes 0.2 s-0.3 s to analyse 11.4 seconds long sampling data, and it＇s real time effect is quite evidence. Conclusion： The characteristic of this arithmetic can satisfy the application in real project.

关 键 词：小波变换 QRS波检测 抗干扰 

分 类 号：R318.04[医药卫生—生物医学工程]