计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (7): 14-16.

• 博士论坛 • 上一篇    下一篇

基于模糊马尔可夫场的脑部MR图像分割算法

李彬 陈武凡 颜刚   

  1. 南方医科大学生物医学工程学院
  • 收稿日期:2006-09-06 修回日期:1900-01-01 出版日期:2007-03-01 发布日期:2007-03-01
  • 通讯作者: 李彬

A Novel Algorithm for Segmentation of Brain MR Images Using Fuzzy Markov Random Field Model

Bin Li WuFan Chen Gang Yan   

  • Received:2006-09-06 Revised:1900-01-01 Online:2007-03-01 Published:2007-03-01
  • Contact: Bin Li

摘要: 磁共振(MR)图像由于部分容积效应使其表现出一定的模糊性。由于传统的马尔可夫随机场模型仅能处理确定问题,因此在运用该模型分割MR图像时效果不佳。本文在传统马尔可夫场模型的基础上,建立了模糊马尔可夫场模型。通过对模型的分析得出图像像素对不同类的隶属度计算公式,提出了一种高效、无监督的图像分割算法,从而实现了对脑部MR图像的精确分割。通过对模拟脑部MR图像和临床脑部MR图像分割实验,表明本文提出的新算法比传统的基于马尔可夫场的图像分割算法和模糊C-均值等图像分割算法有更精确的图像分割能力。

关键词: 图像分割, 马尔可夫场, 模糊, 磁共振图像

Abstract: The magnetic resonance (MR) images behavior fuzziness duo to the artifact of partial volume effects. The conventional Markov random field (MRF) model can only deal with the deterministic problem, so the segmentation results are often undesired using conventional MRF model. In this paper, a fuzzy MRF model is developed based on the conventional MRF model. By analyzing the fuzzy MRF model, the formula of determining the membership values for each voxel to indicate the partial volume degree is derived. We also proposed an efficient and unsupervised algorithm to realize the accurate segmentation for MR brain images. The simulated brain images and real clinical images are selected to test the proposed algorithm. The experimental results show that the proposed algorithm can segment the brain images more accurately than the conventional model-based algorithms and the fuzzy C-mean do as well.

Key words: image segmentation, Markov random field, fuzzy, magnetic resonance images