Abstract: Electrocardiogram（ECG） is one of the basic detection programs in the diagnosis of myocardial ischemia, myocardial infarction and other cardiovascular diseases, in which the ST-T segment reflects the process of ventricular repolarization. Firstly, the wavelet transform-based Multi-Resolution Analysis（MRA） is applied to correct the baseline drift and filtering of the original ECG signals. Then, by combining the convexity operator and dynamic search window, the location of the ST-T segment is obtained. Through all the data in the European ST-T database, the average recognition rate of the algorithm is more than 93%. For all regular ECG signals, the recognition rate is above 90%. This algorithm avoids the location difficulty for T wave peak. It can be effectively applied to the problem of location of R wave inversion, T wave inversion and T wave rose changes. Furthermore, since the computation complexity of this algorithm is lower, it can be well embedded into practical engineering applications.

Key words: ST-T segment, Multi-Resolution Analysis（MRA）, convexity operator, dynamic search windows

摘要： 心电图是心肌缺血、心肌梗塞等心血管疾病的首要检查项目之一，其ST-T段反映了心室复极的过程。利用小波变换的多分辨率分析对原始心电信号进行基线漂移校正和滤波，再结合凸点算子和动态搜索窗方法，对ST-T段定位。通过European ST-T数据库中的所有数据验证，该算法的平均识别率在93%以上。对于所有较为规律的心电信号，其识别率均高于90%。该方法避免了T波峰值定位难这一问题，有效地解决R波倒置、T波倒置和T波高尖等的定位。该算法计算的复杂度较低，能很好地嵌入到实际工程应用中。

关键词: ST-T段, 多分辨率分析, 凸点算子, 动态搜索窗