Guo Z, Durand L G, Allard L, Cloutier G, Lee H C, Langlois Y E
Biomedical Engineering Laboratory, Clinical Research Institute of Montreal, Quebec, Canada.
Med Biol Eng Comput. 1993 May;31(3):237-41. doi: 10.1007/BF02458042.
The normality (Gaussian property) and stationarity of the cardiac Doppler blood-flow signal were evaluated on short-time segments distributed over the cardiac cycle. The basic approaches used to perform statistical tests on the nonstationary and quasiperiodic cardiac Doppler signal are presented. The results obtained from the data of ten patients having a normal aortic valve and ten patients having a stenotic valve indicate that a complex Gaussian random process is an acceptable approximation for the clinical cardiac Doppler signal. For segments of 10 ms or less, 82 per cent of them were accepted to be stationary with a significance level of 0.05, whereas for durations greater than 40 ms, the percentage of stationary segments was less than 75 per cent. It was concluded that the 10 ms window generally used in practice is a good choice for Doppler spectrogram estimation, but a shorter time interval would be preferable.
在心动周期分布的短时间段上评估心脏多普勒血流信号的正态性（高斯特性）和平稳性。介绍了对非平稳和准周期心脏多普勒信号进行统计测试所使用的基本方法。从十名主动脉瓣正常患者和十名瓣膜狭窄患者的数据中获得的结果表明，复高斯随机过程是临床心脏多普勒信号的可接受近似。对于10毫秒或更短的时间段，其中82%在显著性水平为0.05时被接受为平稳的，而对于持续时间大于40毫秒的情况，平稳段的百分比小于75%。得出的结论是，实际中通常使用的10毫秒窗口是多普勒频谱图估计的良好选择，但更短的时间间隔会更可取。
