C-V模型在乳腺X影像分割中的研究和应用

Mammogram Segmentation Based on the Chan-Vese Model

本文以数字乳房X片图像为对象,研究了医学图像分割的理论和算法。首先,文章对水平集方法以及主动轮廓模型的基本理论及其在图像分割领域的应用做了简单的概述。然后重点研究了C-V多相水平集方法,该方法具有理想的区域划分方案,可以分割分段常值和分段光滑图像,可以自然地避免多个水平集函数的重叠和"真空"问题。文章还指出初始位置对曲线演化的速度的影响,并且针对该问题提出了利用阈值分割技术对水平集函数进行初始化,通过优化水平集函数的初始位置来加快C-V模型的演化速度。基于C-V的图像分割算法作为医学诊断的辅助手段有极其重要的意义。

The medical image segmentation theory and algorithms based on the digital mammogram was studied.Firstly,the level set method,and active contour model in the basic theory and application of image segmentation field was simply overuied.Then C-V multiphase level set method was studied emphatically.This method has ideal region partition scheme which could segment constant and smooth images.It was also pointed out that the initial position on the influence of the curve evolution speed and proposed a threshold segmentation technology to initialize the level set function.The C-V model based segmentation algorithm has very important significance as a medical diagnosis auxiliary method.