中国组织工程研究

中国组织工程研究 ›› 2011, Vol. 15 ›› Issue (30): 5607-5610.doi: 10.3969/j.issn.1673-8225.2011.30.024

• 数字化骨科 digital orthopedics • 上一篇    下一篇

基于区域生长超声图像分割血管边界反映血管病变

税  雪，刘  奇   

  1. 四川大学医学信息工程系，四川省成都市  610065
  • 收稿日期:2011-04-11 修回日期:2011-06-18 出版日期:2011-07-23 发布日期:2011-07-23
  • 通讯作者: 刘奇，博士，副教授，硕士生导师，四川大学电气信息学院医学信息工程系，四川省成都市610065 liuqii@scu.edu.cn
  • 作者简介:税雪★，女，1986年生，四川省泸州市人，汉族，四川大学在读硕士，主要从事医学图像与信号处理研究。 Shuixue111@126.com
  • 基金资助:

    国家自然科学基金生命科学部（30700781），课题名称：麻醉期应激反应对在体血管压力-容积曲线的影响。

Ultrasound image segmentation based on region-growing algorithm for reflection of vascular lesions

Shui Xue, Liu Qi   

  1. Department of Medical Information Engineering, Sichuan University, Chengdu  610065, Sichuan Province, China
  • Received:2011-04-11 Revised:2011-06-18 Online:2011-07-23 Published:2011-07-23
  • Contact: Liu Qi, Doctor, Associate professor, Master’s supervisor, Department of Medical Information Engineering, Sichuan University, Chengdu 610065, Sichuan Province, China liuqii@scu.edu.cn
  • About author:Shui Xue★, Studying for master’s degree, Department of Medical Information Engineering, Sichuan University, Chengdu 610065, Sichuan Province, China Shuixue111@126.com
  • Supported by:

    the National Natural Science Foundation of China, No. 30700781*

PDF

510

被引次数

7

摘要：

背景：通过分析超声血管图像能反映血管的病变情况。
目的：采用区域生长理论对超声图像进行图像分割，分析边界点的相对位移。
方法：先对视频图像分帧，将动态图像转换为静态图像，采用Gabor滤波、自适应直方图量化去除超声图像噪声，然后运用区域生长法对图像做分割，接着通过开闭运算、sobel算子检测图像边界，最后提取出两条血管边界。
结果与结论：通过Gabor滤波、区域生长法等手段，得到了比较好的分割结果。区域生长法在处理速度上满足了实时性要求，具有一定的通用性。并且通过分析边界点的相对位移曲线，一定程度上反映血管的病变。

关键词: 血管内膜超声图像, Gabor滤波, 区域生长, 血管形态学, 静态图像

Abstract:

BACKGROUND: It can reflect pathological changes of the blood vessel through the analysis of vascular images in ultrasound.
OBJECTIVE: Based on region-growing algorithm to segment ultrasound images and analyze the boundary point relative displacement
METHODS: Firstly, we decompressed video images into frames, transformed dynamic images into static, and then applied Gabor filtering and adaptive histogram quantization to reduce speckle noise in ultrasound images and segmented the images through region-growing algorithm. Combined with open and close operation and sobel edge detection, we have finally extracted two boundaries of the blood.
RESULTS AND CONCLUSION: The method we proposed obtains the quite good segmentation result, and the method of region-growing meets the requirement of real-time in terms of processing speed. Furthermore, the boundary point relative displacement curve can reflect pathological changes of the blood vessel to some extent, which lays a solid foundation for researches on morphology of the blood vessel and haemodynamics and has important meanings for pathophysiology. 　

中图分类号: