中文摘要：

心电的T波交替现象与室性心律失常有关，已成为预测心脏猝死的一个无创的临床指标。当前T波交替分析常用谱分析方法，为弥补其不具备时间分辨率的不足，本研究提出一种鲁棒的基于时频分析的微伏级T波交替检测算法：通过心电信号的短时傅里叶变换，提取时域T波序列，计算其在选定时频区域的能量谱；然后运用Wilcoxon秩和检验统计分析方法，检测微伏级T波交替现象。经仿真实验、欧洲sT—T数据及临床检测实验，本算法对T波交替检测的平均灵敏性达90．4％，正确预测率达92．0％，在30dB及以上信噪比情况下，实现了100％的正确检测。仿真实验还表明本算法也支持短时心电数据的TWA准确检测。

英文摘要：

T-wave ahernans （TWA） is associated with the ventricular arrhythmia, and it has been regarded as an independent and noninvasive indicator of sudden cardiac death （SCD）. The spectral method is commonly used to detect TWA but does not provide temporal resolution. In this paper a novel algorithm based on the time- frequency analysis was developed to detect microvoh TWA with both temporal and frequency resolution. After computing the short-time Fourier transformatio （STFT） of the electrocardiogram （EGG） segment, the T wave sequences were extracted and the total energy of each T wave contained within the selected time-frequency region was calculated, and then the Wilcoxon rank-sum test was applied to the energy sequences of T waves to detect TWA. Simulated ECGs containing artificial TWA were used to test the algorithm and subsequently the ECGs from European ST-T database and healthy objects were tested by the algorithm, which achieved a sensitivity of 90.4% and a accuracy rate of 92.0%. And if the SNR was not less than 30dB, the accuracy rate of detection by the algorithm achieved 100%. The simulated experiment indicates that TWA of the small samples of ECG also can be detected by the algorithm accurately.