基于相关分析方法的非稳态心电T波交替检测研究

目的：研究T波段幅度、形态逐拍交替变化的心电变异现象检测。方法：本研究首先使用Mexican．hat小波检测R峰并对心电信号进行预处理；在提取T波矩阵方面为减少心拍间内差，采用点乘最大法，最大程度地对齐T波；最后基于时域相关分析方法检测T波交替幅度、交替心拍，追踪非稳态心电信号中短暂的交替数据段。结果：利用相关分析法对样本数据所测交替幅度与谱分析法相比更加显著，并且可以检测出谱分析方法无法检测的交替心拍。结论：时域相关分析方法能够更精确地追踪T波交替随时间变化的现象，但其对输入数据要求较高，因此在检测中可以先通过谱分析方法检测为阳性TWA的基础上，再对心电信号进行相关分析，从而确定非稳态交替时间段和更加准确的交替幅度。

Research on Detecting the Nonstationary T-wave Alternans（TWA） Based on Correlation Method

Objective： To research the technology of detecting T-wave altemans （TWA） which is a variation in the amplitude or morphology of the T wave of the beat-to-beat ECG. Methods： The study firstly uses Mexican-hat wavelet transform to detect R-peak and then preprocesses the ECG signal; Dot product alignment is adopted to minimize inter-beat variance which should result in aligning the T wave to an extreme on the resolution of the T wave matrix; Finally, the study detects the alternans amplitude and heart beats to track the transient TWA episodes of the nonstationary ECG signal based on the time-domain correlation method（CM.）. Results： In comparison with the accepted spectral method（SM）, CM is able to detect the more obvious altemans amplitude and identify the alternans heart beats which cannot be detected by SM. Conclutions： The correlation method tracks the time-varying TWA phenomenon more accurately, however, it has the stringent data requirements. Therefore we process the signal by the specatral technique firstly, then, if the significant TWA is detected, we adopt the correlation method to confirm transient TWA episodes and the more accurate amplitude.