摘要
通过研究冠脉血流动力学和心脏心音产生的机理 ,首次提出了将独立分量分析 (ICA)方法应用于心音信号处理并达到自动检测冠心病的目的。在本系统中 ,信号采集系统采用了高灵敏度传感器列阵对正常人及冠心病患者胸部的多个部位进行检测 ,经预处理后的信号最后通过计算机进行数据采集。应用独立分量分析的方法将心脏舒张期的心音信号进行分离 ,并将各心音分量的统计特征参数作为输入参量输入到径向基函数网络 (RBF网络 )进行训练和识别。实验结果说明 。
This paper studies the mechanism of coronary artery blood dynamics and heart sounds and we develop a novel method of applying independent component analysis (ICA) to heart sounds attained by sensor array. This system can diagnose coronary artery diseases (CAD) automatically. In this system, heart sounds of healthy people and CAD patients were detected by highly sensitive sensor array placed on body chest and the signals were acquired by computer. After using ICA to separate various components of sounds from diastolic heart sounds, all the characteristic parameters of each component of sounds compose a 20 parameters' input vector for radial basis function networks (RBF) to learn and identify. The result shows that the application of ICA and neural networks to heart sounds is an efficient noninvasive method to diagnose CAD.
出处
《传感技术学报》
CAS
CSCD
2003年第1期16-20,共5页
Chinese Journal of Sensors and Actuators