计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (4): 183-185.DOI: 10.3778/j.issn.1002-8331.2009.04.052

• 图形、图像、模式识别 • 上一篇    下一篇

基于旋转不变测度的扩散张量图像配准

李 文,江贵平   

  1. 南方医科大学 生物医学工程学院,广州 510515
  • 收稿日期:2008-07-24 修回日期:2008-10-17 出版日期:2009-02-01 发布日期:2009-02-01
  • 通讯作者: 李 文

Registration of diffusion tensor images based on rotation invariants

LI Wen,JIANG Gui-ping   

  1. Department of Biomedical Engineering,Southern Medical University,Guangzhou 510515,China
  • Received:2008-07-24 Revised:2008-10-17 Online:2009-02-01 Published:2009-02-01
  • Contact: LI Wen

摘要: 扩散张量成像(Diffusion Tensor Magnetic Resonance Imaging,DT-MRI)是一种新兴的磁共振成像技术,作为非侵入性的分析大脑内部结构的重要工具,DTI在神经外科学等临床领域的研究中发挥着重要的作用,与此同时不同个体以及同一个体不同时间采集的DTI数据的配准需求也与日俱增,为满足实际需求提出了一种基于旋转不变测度的扩散张量图像的配准方法,首先对扩散张量图像进行空间变换,使其空间位置一一对应,考虑张量数据的方向性特点,再对每个张量进行重定向,从而保证了配准后的图像其扩散张量方向与周围组织的解剖结构一致。实验结果表明,该方法计算复杂度低,是一种有效的张量图像配准方法。

关键词: 扩散张量, 配准, 旋转不变测度, 互信息, 重定向

Abstract: As a newly developed non-invasive technique of brain structure analysis,Diffusion Tensor Magnetic Resonance Imaging(DT-MRI) has been playing a more and more significant role in neurological and neurosurgical applications.Accordingly,the requirement for registration of DTI is increasing.In this paper a new method of registration of DT images based on rotation invariants is presented.First,rotation invariants images calculated from source and target DT data are registered using multi-channel mutual information.Second,every tensor is reoriented to assure that tensor orientations of images after registration preserve its physical meaning,that is,diffusion must be oriented in the same direction as fibers.

Key words: diffusion tensor, registration, rotation invariants, mutual information, reorientation