Qin Binjie, Zhuang Tiange
Deptartment of Biomedical Engineering, Shanghai Jiaotong University, Shanghai 200030.
Sheng Wu Yi Xue Gong Cheng Xue Za Zhi. 2003 Jun;20(2):233-6.
An effective MR and PET image registration method designed for clinical use is given in this paper. A new correlation ratio similarity measure for voxel intensity based medical image registration was adopted. Detailed discussion on methods of rigid transformation, resampling, multi-resolution medical image representation and optimization for accelerated multi-resolution registration scheme is also given in this paper. At last, experimental results based on clinical MR and PET medical images show that correlation ratio based multi-resolution registration method works well and is faster than most existing direct voxel intensity based registration methods under clinical scenario.
本文给出了一种专为临床应用设计的有效磁共振成像（MR）与正电子发射断层显像（PET）图像配准方法。采用了一种基于体素强度的医学图像配准新相关比相似性度量。本文还详细讨论了刚体变换、重采样、多分辨率医学图像表示以及加速多分辨率配准方案优化的方法。最后，基于临床MR和PET医学图像的实验结果表明，基于相关比的多分辨率配准方法效果良好，且在临床场景下比大多数现有的基于直接体素强度的配准方法更快。
