检索历史 清除

摘要采用动态三维心音频谱分析技术对16例确诊为二尖瓣狭窄病人的心音进行了分析。将心音分成四段，即第一心音段、收缩期段、第二心音段和舒张期段。用AR参数模型估计四段的功率谱，计算包括时间域、频率域和强度域三方面的16个启发性特征参数。结果表明，二尖瓣狭窄病人与正常人上述参数有明显的差别（除3个参数外，其余参数均 P<0.05）。用Pocket算法学习训练的以这16个特征参数为模式向量的线性分类器可以用来作正常和异常的二分类识别。并且，舒张期杂音响度分级与舒张期相对性能量参数［lg（rd1）］呈正相关。

abstractsWe studied heart sounds of 16 cases of rheumatic mitral valve stenosis by means of dynamic 3-D power spectrum analysis. The PCG signal in a cardiac cycle could be divided into four segments that were the first heart sound period, systolic period, the second heart sound period and diastolic period. Then applied AR modeling method, frequency domain features could be derived from power spectrum of the four segments. The result suggested that there was signficent difference of 16 heuristic heart sounds characteristics between mitral stenostic patients and normal persons ( P<0.05 eccept 3 features). The linear classifying which was trained by Pocket algorithm with these 16 dimensional feature vector could be used classifying normal and abnormal. Moreover, there was positive correlation between the diastolic murmur degree and the logarithms of instantaneous sound intesity of diastolic period, ie lg (rdl) .