A correlation analysis-based detection and delineation of ECG characteristic events using template waveforms extracted by ensemble averaging of clustered heart cycles
A correlation analysis-based detection and delineation of ECG characteristic events using template waveforms extracted by ensemble averaging of clustered heart cycles
复制标题
DOI：
10.1016/j.compbiomed.2013.10.024
复制
发表时间：
2014-01-01
影响因子：
7.7
通讯作者：
Naseri, H.
中科院分区：
文献类型：
--
作者：
Homaeinezhad, M. R.;ErfanianMoshiri-Nejad, M.;Naseri, H.
本研究的目的是介绍一种简单，标准和安全的程序来检测和描绘P波和T波的心电图（ECG）信号在真实的条件。所提出的方法包括四个主要步骤：（1）一个安全的QRS波检测和描绘算法，（2）一个模式识别算法，旨在区分不同的心电图集群之间发生的连续R波，（3）提取模板的主导事件的每个集群波形和（4）应用相关分析，以自动描绘的P-和T-波在噪声条件下。所提出的P和T检测描绘算法的性能特性进行评估与各种ECG信号的质量从最好的情况下改变到最坏的情况下，基于随机游走噪声理论。此外，该方法被应用到MIT-BIH心律失常和QT数据库的一些部分的性能特征进行比较与一些P和T检测划定algorithm.The进行的评估表明，在低质量值约0.6的信号，所提出的方法检测P和T事件的灵敏度Se=85%和P+= 89%，分别。此外，在相同质量下，与这些ECG事件相关联的平均描绘误差分别为45和63 ms。稳定的描绘误差、高检测准确度和高抗噪能力是所提出方法开发过程中考虑的最重要方面。(C)2013爱思唯尔有限公司保留所有权利。
The goal of this study is to introduce a simple, standard and safe procedure to detect and to delineate P and T waves of the electrocardiogram (ECG) signal in real conditions. The proposed method consists of four major steps: (1) a secure QRS detection and delineation algorithm, (2) a pattern recognition algorithm designed for distinguishing various ECG clusters which take place between consecutive R-waves, (3) extracting template of the dominant events of each cluster waveform and (4) application of the correlation analysis in order to delineate automatically the P- and T-waves in noisy conditions. The performance characteristics of the proposed P and T detection-delineation algorithm are evaluated versus various ECG signals whose qualities are altered from the best to the worst cases based on the random-walk noise theory. Also, the method is applied to the MIT-BIH Arrhythmia and the QT databases for comparing some parts of its performance characteristics with a number of P and T detection-delineation algorithms.The conducted evaluations indicate that in a signal with low quality value of about 0.6, the proposed method detects the P and T events with sensitivity Se=85% and positive predictive value of P+=89%, respectively. In addition, at the same quality, the average delineation errors associated with those ECG events are 45 and 63 ms, respectively. Stable delineation error, high detection accuracy and high noise tolerance were the most important aspects considered during development of the proposed method. (C) 2013 Elsevier Ltd. All rights reserved.