一种基于胸部CT图像感兴趣区域提取的检索技术  

A Method of Chest CT Image Retrieval Based on ROI Extraction

作　　者：陈琴[1] 邰晓英[1] 

机构地区：[1]宁波大学信息科学与工程学院,浙江宁波315211

出　　处：《宁波大学学报（理工版）》2009年第3期354-358,共5页Journal of Ningbo University：Natural Science and Engineering Edition

基　　金：国家自然科学基金(60472099)

摘　　要：探索高效的医学图像检索技术,可以方便医务人员快速高效地查找相似的图像.针对胸部CT图像,提出一种采用改进的区域生长法提取感兴趣区域,在此基础上再提取纹理特征作为图像的特征向量,并以特征向量作为图像检索的依据来进行检索.实验结果表明:在有效提取出病灶区域之余,针对该区域提取图像特征起到了良好的检索效果.将此方法应用于医学图像检索系统可有助于提高医生的诊断效率.Improvement of the retrieval efficiencies for medical images can benefit medical staff members in fast search of similar images and CT chest images. In this paper, an improved regional growing algorithm is proposed to retrieve the region of interest. Based on the regions detected, the texture features are further extracted as the characteristic vectors of the image, with which as the tracking evidence to retrieve the image from the image set for classification purposes. The test results show that, other than efficiently extracting the diagnosed regions, the proposed method can also improve the image retrieval efficiency in terms of detecting the image characteristics. The proposed algorithm can be applied in real-time medical image retrieval system, and uplift the diagnose efficiency for medical staff members.

关 键 词：医学图像检索 感兴趣区域提取 纹理特征 胸部CT图像 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]