基于图割与快速水平集的腹部CT图像分割

杨昌俊; 杨新

基于图割与快速水平集的腹部CT图像分割

杨昌俊^1,2,
杨新^1,2, ,

1. 上海交通大学自动化系, 上海 200240;
2. 上海交通大学系统控制与信息处理教育部重点实验室, 上海 200240

基金项目:

上海市基础研究重点项目(09jc1410502)

国家重点基础研究发展计划(973计划)(2010CB732506)

详细信息

作者简介:
杨昌俊(1987-),男,上海交通大学模式识别与智能系统专业硕士研究生,主要研究方向为医学图像分析和三维可视化,Tel:13917565651,E-mail:changjunyang@gmail.com;杨新(1950-),男,上海交通大学自动化系博士生导师,主要研究方向为医学图像分析、三维可视化、图像处理中的偏微分方程,E-mail:yangxin@sjtu.edu.cn

通讯作者:
杨新(1950-),男,上海交通大学自动化系博士生导师,主要研究方向为医学图像分析、三维可视化、图像处理中的偏微分方程,E-mail:yangxin@sjtu.edu.cn

中图分类号: R814
计量
- 文章访问数: 915
- HTML全文浏览量: 7
- PDF下载量: 7
出版历程
- 收稿日期: 2011-05-26
- 网络出版日期: 2022-12-14

Abdominal CT Image Segmentation Based on Graph Cuts and Fast Level Set

YANG Chang-jun^1,2,
YANG Xin^1,2, ,

1. Department of Automation, Shanghai Jiao Tong University, Shanghai 200240, China;
2. Key Laboratory of System Control and Information Processing, Ministry of Education of China, Shanghai Jiao Tong University, Shanghai 200240, China

摘要

摘要: 针对腹部CT图像中组织分割的问题,提出了一种基于图割与改进的快速水平集的交互式分割方法。首先对人工给定的一个待分割目标的初始轮廓作膨胀运算,将所得内部边界所有像素点作为源点、外部边界像素点作为汇点构造图,并通过图割方法对CT图像进行初步分割,然后以膨胀所得内部边界作为初始轮廓,通过基于区域竞争主动轮廓模型(RCAC)的快速水平集算法对初步分割后的图像进行精确分割,克服了传统水平集方法运算速度较慢及易产生边界泄漏的问题。在此基础上将二维分割推广到三维分割,最终完成腹部不同器官的完整三维分割。进行三维重建,使医生能直观地观测到各器官的三维相对位置和拓扑结构。实验结果表明,该方法交互简单、鲁棒性强、准确性高,能有效辅助医生实现诊断和手术规划。
- 腹部CT图像分割 /
- 图割 /
- 水平集方法
Abstract: An interactive segmentation method based on graph cuts and improved fast level set is proposed to segment abdominal CT image.In our approach,an initial contour is sketched and dilated,a graph for graph cuts method is constructed while inner contour vertices based on morphologic dilation are identified as source and outer contour vertices as sink,the Pre-CT image segmentation is achieved by graph cuts method roughly,and then with the initial inner contour of dilation,the fast level set algorithm based on the Region Competition Based Active Contour(RCAC)model is applied to re-segment the result image by graph cuts.Thus,the low speed and leakage error in traditional level set methods are avoided.Furthermore,we extend this segmentation method to three dimensions,and several 3-D abdominal organs are segmented.Doctors could straightforwardly visualize the organs'relationship and structure topologically after 3-D reconstruction.Experimental results show that the proposed method is with interactive simply,robustness and high accuracy,and can support doctors for diagnosis and surgical planning effectively.
- abdominal CT image segmentation /
- graph cuts /
- level set method