Jones D L, Touvannas J S, Lander P, Albert D E
Coordinated Science Laboratory, University of Illinois, Urbana 61801.
J Electrocardiol. 1992;25 Suppl:188-94. doi: 10.1016/0022-0736(92)90099-l.
Frequency-domain techniques have been extensively investigated for the analysis of high-resolution electrocardiograms (ECGs), although the merit of frequency-domain analysis is still subject to controversy. Time-frequency analysis methods, which estimate the frequency content of a signal as a function of time, potentially provide even more information for improved ECG analysis. Some researchers report impressive results in predicting the outcome of electrophysiologic studies using the short-time Fourier transform (spectrogram). Other time-frequency representations, such as the Wigner distribution, short-time spectral estimators, and the wavelet transform, have also been investigated. The authors present a unified overview of time-frequency representations, showing that only four classes characterize most time-frequency representations. The authors describe the advantages and drawbacks of the various approaches and speculate on their promise for ECG analysis. Very preliminary experiments in applying some of these techniques to the prediction of the outcome of electrophysiologic studies have suggested some possible new research directions.
频域技术已被广泛用于分析高分辨率心电图（ECG），尽管频域分析的优点仍存在争议。时频分析方法可将信号的频率成分作为时间的函数进行估计，有可能为改进心电图分析提供更多信息。一些研究人员报告称，使用短时傅里叶变换（频谱图）预测电生理研究结果取得了令人瞩目的成果。其他时频表示方法，如维格纳分布、短时谱估计器和小波变换，也已得到研究。作者对时频表示方法进行了统一概述，表明大多数时频表示方法仅分为四类。作者描述了各种方法的优缺点，并推测了它们在心电图分析中的前景。将其中一些技术应用于电生理研究结果预测的非常初步的实验提出了一些可能的新研究方向。
