基于PCA和改进的BP网络的肺结节识别研究  

Identification research of pulmonary nodules based on PCA and the improved BP neural network

作　　者：胡向丽[1] 范立南[1] 孙申申[1] 谷文娟[1] 

机构地区：[1]沈阳大学信息工程学院,沈阳110044

出　　处：《仪器仪表用户》2011年第6期3-7,共5页Instrumentation

基　　金：辽宁省自然科学基金(20102154);辽宁省教育厅科研项目计划(L2010376)

摘　　要：提出了一种基于PCA和改进的BP神经网络的CT影像肺结节检测方法。将已经获取的肺部影像中肺结节和血管的特征数据经PCA处理后,应用改进的BP神经网络进行分类,经过竞争选择,获得分类结果。实验证明该识别方法实现简单、识别率高、训练速度与识别速度较快,训练好的网络有很好的分类能力。Putting forward a detection method of pulmonary nodules which is applied to CT image based on PCA and the improved BP neural network. Characteristics of nodules and vascular which are processed by PCA have been obtained, classified by applying the improved BP neural network, choiced by competition, recognized by results. Experiment results show that implementation of the recognition model is simple, recognition rate is high, speed of training and recognition is fast, and the trained network has very good classification ability.

关 键 词：主成分分析(PCA) BP神经网络 肺结节 血管 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]