心电信号多形态T波检测方法研究

Study on Detection Method of T Waves with Different Morphologies in Electrocardiogram

目的研究一种T波检测新方法,以提高检测形态异常T波的准确率。方法基于小波变换和支持矢量机技术,设计了一种准确性高且适用于检测心电信号多形态T波的新方法,并从QT数据库下载了50组真实心电信号中5000个形态多样的T波数据验证新方法的性能。结果将实验结果与库中专家标注的信息对比,在误差容忍度20 ms内,检测准确率达到96.38%,敏感度达到98.04%,阳性预测率达到100%。与现有其他类似方法比较,新方法的平均检测精确度和敏感度分别高出6.73%和4.48%。结论本文方法对于各种形态的T波均具有高的检测准确率,且具有良好的抗噪性和可扩展性。

Objective To propose a novel T wave detection method for increasing the detection accuracy of T waves with abnormal morphologies. Methods Based on wavelet transform and support vector machine technolo- gy, a new method for detecting T waves with different morphologies in electrocardiogram （ECG） was pro- posed. The 5000 beat data from 50 sets of real ECG signals with various T wave morphologies downloaded from the QT database were applied to validate the proposed method. Results Compared with the results annotated by the cardiologists on QT database, the accuracy, sensitivity and positive predictive accuracy were 96.38%, 98.04% and 100% respectively with the error tolerance being of 20 ms. The average detection accuracy and sensitivity were 6.73% and 4.48% higher respectively than those of other related methods. Conclusion The novel method is of high accuracy to detect T waves with various morphologies and has good scalability and noise immunity.