音频多普勒信号的多特征分类决策  

MULTI-CHARACTERISTICS CLASSIFICATION OF AUDIO DOPPLER SIGNALS

作　　者：陈斯中[1] 汪源源[1] 王威琪[1] 

机构地区：[1]复旦大学电子工程系,上海200433

出　　处：《中国生物医学工程学报》2002年第4期308-314,共7页Chinese Journal of Biomedical Engineering

基　　金：教育部优秀年轻教师基金资助项目 ( 9710 99);上海市青年科技启明星计划资助项目 ( 97QD14 0 35 )

摘　　要：本文将音频多普勒信号的几种特征提取方法结合起来 ,用多元分析的方法对血流状况进行多特征的分类决策 ,所用的特征提取方法包括 :传统的声谱参数法 ,音频信号的零极点模型法 ,分形特征分析法及Teager能量法。将这种多特征分类决策法用于 71例胎儿脐血流音频多普勒信号的分析 ,结果表明 :该方法对胎儿宫内异常等疾病具有敏感性 。Several characteristics extraction methods of audio Doppler signals were combined to judge the normal or abnormal blood flow states with a multianalysis method. The characteristics extraction methods included the conventional method of spectrogram parameters, the zero-pole model method of audio Doppler signals, the fractal characteristics analysis method and the Teager energy method. The analysis results of 71 cases of audio Doppler signals from the umbilical blood flow with this multianalysis method showed that the method was sensitive to the finding of fetal intrauterine diseases and it might provide a satisfactory diagnostic means for the clinical medicine.

关 键 词：音频多普勒信号 多元分析 声谱 零极点 分形 Teager算法 

分 类 号：R445.1[医药卫生—影像医学与核医学] R318.04[医药卫生—诊断学]