摘要
本文介绍一种结合贝叶斯分类的水平集方法用于人体MR图像分割。本方法首先通过贝叶斯分类模型计算出水平集曲线位于边界上的概率;其次,将与此概率相关联的区域决策影响因子添加在水平集函数方程中;最终,实现利用图像的区域信息提高水平集曲线识别边界能力的目的。分割实验结果表明,该方法较好地克服了传统水平集方法在MR图像分割中存在的边界泄露问题。与其他MR图像分割方法比较,本方法具有更优的分割结果。
In this paper a level set segmentation algorithm based on Bayesian classification for human Magnetic Resonance Images (MRI)was proposed. Firstly the probability that level set curve lies on boundary was calculated by using Bayesian classification model Secondly the term of region strategy which is correlated with the forenamed probability was appended to the level set function Finally by using the images regional information we improve the level set method on identifying the boundary. Results of the segmentation tests prove that the proposed segmenting method preferably solve the leaking problem in MRI segmentation for traditional level set method and obtain a better result than other MRI segmentation methods.
出处
《中国医学物理学杂志》
CSCD
2005年第4期568-571,共4页
Chinese Journal of Medical Physics
基金
国家重点基础研究发展计划(973计划)资助项目(No:2003CB716100)
广东省自然科学基金资助项目(No:32891)
关键词
MRI
图像分割
水平集
贝叶斯分类
窄带法
MRI image segmentation level set method bayesian classification narrow band
