基于多阈值单水平集方法的医学图像分割  

Medical image segmentation based on multi-threshold single level set

作　　者：张娜[1] 张建勋[1] 王慧娟[2] 史瑞芝[1] 

机构地区：[1]南开大学机器人与信息自动化研究所,天津300071 [2]北华航天工业学院计算机科学与工程系,廊坊065000

出　　处：《高技术通讯》2009年第11期1164-1169,共6页Chinese High Technology Letters

基　　金：863计划(2007AA04Z235)资助项目

摘　　要：针对Chan-Vese的无边界主动轮廓模型(CV模型)只能区分前景与背景的缺点,提出了一种基于多阈值单水平集的医学图像分割方法,并将此方法应用于微创手术的预处理中。由于医学图像结构复杂,具有器官轮廓多连接等特点,因此使用常规的水平集方法进行分割往往不能取得理想的效果,而该方法采用修改目标泛函的方式引入多类分割,具有多区域分割的特点,只需经过一次单水平集的迭代循环,即可将图像根据灰度不同划分为多个区域,具有精确、快速等优点。对不同的合成图像和医学图像的实验结果表明,该方法实现了快速精确的多区域分割,能很好地提取到医学图像中的骨骼轮廓,分割效果达到了预期水平。This paper proposes a multi-threshold single level set model for medical image segmentation based on the Chan-Vese model. It can be applied to the pre-processing in minimal invasive surgery. The structures of medical images are always complicated, and the contours of organs are always muhi-conneeted. The general level set algorithms for medical image segmentation usually can not attain good results. Chan-Vese model can only distinguish two types of intensities. By this new model, after a circle of calculating the single level set, images are segmented to multi-regions exactly and efficiently according to the intensities of them. The results of applying this model to different synthetic ar/d real medical images approved that this model can partition images to multi regions, and detect bone contours very efficiently.

关 键 词：医学图像分割 单水平集 Chan-Vese(CV)模型 多区域分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP751.1[自动化与计算机技术—计算机科学与技术]