摘要
QRS形态分析是心电信号自动分析中的关键步骤之一。现有QRS形态分析算法大多是基于时频变换分析或基于基元线段句法识别的。其中,基于时频变换分析的方法难以处理复杂波形和异常波形;而基于基元线段句法识别的算法则易丢失波形细节。针对这一问题,本文提出了一种基于QRS波群关键点和有限自动机的QRS形态分析算法。首先,采用二次多项式曲线对输入的心电信号进行分段最小二乘法逼近。然后,通过分析各段曲线的单调性、陡峭程度以及曲线方向等特征,实现QRS波群中峰点、谷点、边界点等关键点的检测,同时提取各关键点的幅度和时间信息,并判断出关键点的类型。最后,构造了一个有限自动机,以带属性的QRS波群关键点序列作为输入,实现了QRS形态的识别。经MIT-BIH心律失常数据库验证,本文算法可处理含有任意多个子波的QRS波群,正确识别出其各种复杂形态,准确分析出如切迹、顿挫等形态细节。
Analysis of QRS morphology is one of crucial steps in automatic electrocardiogram (ECG) analysis. Most existing methods for QRS morphological analysis are based on time-frequency transform or syntactic recognition with the primitives of line segments. The algorithms based on time-frequency transform can not correctly deal with complex QRS morphology especially of abnormal shapes;while syntactic algorithms based on line segment primitives are used to ignore details. This paper introduces an algorithm based on a finite automaton with the key points of QRS complex as inputs for QRS morphological analysis. At first, the input ECG signal is piecewise fitted with quadratic polynomial curves according to least square error. Then, in each segment, the features of monotonicity, steep degree, and curve direction are analyzed to detect key points such as peaks, valleys, and boundaries in QRS complex. The amplitude and time of each detected key point are extracted, and all the points are classified into several types. Finally, a finite automaton is constructed to analyze the QRS morphology with the inputs of attributed QRS key points. The proposed method has been tested with the MIT-BIH arrhythmia database. Experiment results show that it can deal with the QRS complex with arbitrary number of waves, describe various complex shapes correctly,and accurately detect morphological details such as notches and slurs.
出处
《信号处理》
CSCD
北大核心
2009年第11期1680-1685,共6页
Journal of Signal Processing
关键词
心电分析
QRS波群
关键点提取
有限自动机
形态分析
ECG analysis
QRS Complex
Key Point Extraction
Finite Automaton
Morphological Analysis