基于图像识别句法分析的心电图计算机自动分析  

Electrocardiogram Computer Automatic Analysis Based on Pattern Recognition Syntax Analysis

作　　者：陈长亮[1] 杨伟杰[1] 陈振生[1] 

机构地区：[1]山东大学控制科学与工程学院,济南250100

出　　处：《中国医疗器械信息》2006年第3期17-18,22,共3页China Medical Device Information

摘　　要：心电图的计算机分析一直热门话题,但临床上心电图的计算机分析应用并不广泛,其原因是心电自动分析的精度较低。随着DSP的广泛应用和计算机数字图像处理技术的发展,我们具备了采用更高精度算法的软硬件条件。心电图作为波形图像处理时,包含了比单纯电压信号更多的信息,对波形图进行分析和识别可以对心电图形变异的捕捉更加敏感和精确。作者通过采用标准正交Hermite核函数来扩展生成基元,并将基元的参数作为或然神经网络的输入,神经网络的输出作为判别结果,取得了令人欣喜的效果。Electrocardiogram（ECG） Computer Analysis has been a hot research topic for a long time. but because of the deficiency of its accuracy, it cannot he used instead of manpowr at all. Today as DSP is applied to many areas, and DiGital image processin is developed very well, which give us the conditions for adopting more precise methods and algorithms. When we take ECG as waveform pattern, much more information can be detected than the pule voltage signal series. By the image pattern recognition methods, this paper describes how to analyze the EUG in the more effective and sensible way.

关 键 词：模式识别 基元 聚类分析 或然神经网络 

分 类 号：R540.1[医药卫生—心血管疾病]