English

应用科学学报 >

2012 , Vol. 30 >Issue 1: 89 - 95

DOI: https://doi.org/10.3969/j.issn.0255-8297.2012.01.014

信号与信息处理

离散轮廓点集法提取超声图像前列腺边缘

展开
  • 1. 复旦大学电子工程系，上海200433
    2. 新华医院超声诊断科，上海200092
通信作者：汪源源，教授，博导，研究方向：医学信号处理和医学超声工程，E-mail: yywang@fudan.edu.cn

收稿日期: 2011-08-02

  修回日期: 2011-09-13

  网络出版日期: 2012-01-30

基金资助

国家自然科学基金(No.10974035)；上海市优秀学科带头人计划基金(No.10XD1400600)资助

收起

Prostate Ultrasound Image Boundary Extraction Using Discrete Contour Point Set

Expand
  • 1. Department of Electronic Engineering, Fudan University, Shanghai 200433, China
    2. Department of Ultrasound Diagnoses, Xinhua Hospital, Shanghai 200092, China

Received date: 2011-08-02

  Revised date: 2011-09-13

  Online published: 2012-01-30

Fold

摘要

 提出一种基于离散轮廓点集的超声图像前列腺边缘提取方法，将经直肠超声图像中的腺体准确分割出来. 该方法根据前列腺的轮廓特征对候选轮廓点进行多级筛选，再除去由于各种干扰产生的非边缘点，得到位于实际边缘的离散轮廓点集，由此确定粗糙的腺体边缘. 随后利用一种快速水平集的曲线演化方法对初始边缘进行小
范围调整，使其更接近实际的腺体轮廓. 对临床应用中的经直肠超声图像进行边缘提取实验，结果表明该方法能克服腺体外部和内部区域灰度不均一以及边缘模糊等问题，提取的前列腺边缘位置准确，轮廓完整.

关键词： 超声图像; 前列腺; 边缘提取; 离散轮廓点集; 曲线演化

本文引用格式

原宗良1， 汪源源1， 余锦华1， 陈亚清2 . 离散轮廓点集法提取超声图像前列腺边缘[J]. 应用科学学报, 2012 , 30(1) : 89 -95 . DOI: 10.3969/j.issn.0255-8297.2012.01.014

Abstract

 A boundary extraction method based on the discrete contour point set (DCPS) is proposed for transrectal ultrasound (TRUS) prostate images. Several features of the prostate boundary are used to search candidate contour points, and a point set is picked up from them as the actual boundary by eliminating nonboundary
candidate points. A coarse prostate boundary is defined by the DCPS and adjusted in a small scale to get a closer prostate boundary using a curve evolution method with a fast level set. Results of experiments for a series of TRUS images show that the proposed method can effectively overcome problems of inhomogeneous intensity distributions inside and outside the prostate, and blur or missing edges, to extract an accurate and fully-formed boundary of the prostate.
Keywords:

Key words： ultrasound images; prostate; boundary extraction; discrete contour point set; curve evolution

Options
文章导航

/

〈 〉