摘要
本文介绍了两种对超声多普勒血流声谱国进行波形自动分类决策的方法。方法一，与统计得到的标准波形组对比，按适当判推进行自动分类；方法二，送入经训练的人工神经网络进行自动分类。待分类波形由计算多普勒信号的最大频率提取声谱包络并经分段、平均和归一化等处理得到。文中给出了计算机模拟的结果。
Two methods for automatic classification and decision of Doppler waveforms for flow sonography analysis system are described. The envelope of the flow sonogram is obtained by estimating the maximum frequency of Doppler signal, from which we can draw out the curve to be classified' C1assification is carried out by comparing with the standard waveform group in the first method and by the use of the error back-propagation type neural network in the second method. One can then make some diagnosis decision.Result of a computer simulation is given.
出处
《应用声学》
CSCD
1996年第3期7-11,共5页
Journal of Applied Acoustics
关键词
超声多普勒
声谱包络
分类决策
血流声谱
Doppler ultrasound, Envelope of flow sonogram, Classification and decision,Neural network
