B型超声心动图中心脏组织及结构的识别  

RECOGNITION OF CARDIAC TISSUE AND STRUCTURE FROM ECHOCARDIOGRAMS

作　　者：程义民[1] 王以孝[1] 沈勇[1] 张冬青[1] 

机构地区：[1]中国科学技术大学电子科学与技术系,计算机视觉实验室合肥230026

出　　处：《中国生物医学工程学报》2001年第6期502-509,共8页Chinese Journal of Biomedical Engineering

摘　　要：本文描述了一种从B型超声心动图识别心脏组织及结构的方法。该方法融合了空间特征信息及纹理特征信息 ,并用经神经网络进行识别进而给出心脏组织及结构的类别标号图象。在空间特征提取中 ,采用了基于模型的模糊分割方法 ,该方法可将各类组织交界附近隶属度带到识别过程中并融合纹理特征 ,因而有较好的识别率。该方法已在一台PentiumIIPC机上进行了模拟 ,并获得了较好的实验结果。We presented a new method for recognizing of cardiac tissue and structure from echocardiograms. In the method, using a Neural Network, the space feature and texture feature of the scene were fused, and the type-labeled image of cardiac tissue and structure could be obtained. In the process of extracting space features, a fuzzy segmentation based on model was used to improve the classification of tissue pixels nearby the boundary of diffirent tissue. Furthermore, the recognition result was refined with the texture feature. The method was simulated on a Pentium II PC and some good experiment results were obtained.

关 键 词：超声心动图 模糊分割 空间特征 纹理特征 神经网络 

分 类 号：R318.04[医药卫生—生物医学工程] R540.45[医药卫生—基础医学]