计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (19): 167-170.

• 图形、图像、模式识别 • 上一篇    下一篇

牙周膜胶原纤维组织切片的显微图像配准

杨  玲1,2,孙翠刚2,王中科3,饶妮妮1   

  1. 1.电子科技大学 生命科学与技术学院,成都 610054
    2.成都信息工程学院 电子工程学院,成都 610225
    3.成都信息工程学院 网络工程学院,成都 610225
  • 出版日期:2012-07-01 发布日期:2012-06-27

Micrographs image registration of collagen fibers slices in periodontal ligament

YANG Ling1,2, SUN Cuigang2, WANG Zhongke3, RAO Nini1   

  1. 1.School of Life Science and Technology, University of Electronic Science and Technology of China, Chengdu 610054, China
    2.College of Electronic Engineering, Chengdu University of Information Technology, Chengdu 610225, China
    3.College of Networks Engineering, Chengdu University of Information Technology, Chengdu 610225, China
  • Online:2012-07-01 Published:2012-06-27

摘要: 针对牙周膜胶原纤维组织切片在制作过程中对力度和方向非常敏感,极易造成组织变形和空间位移的问题,采用了一种有效的图像配准方法,以便后期对胶原纤维组织空间结构的重建。在Masson染色制成的兔牙周膜石蜡切片的基础上,利用高倍光学显微镜获得了牙周膜的序列切片图像;再采用基于正规步长梯度下降的二维刚性配准和基于对数域对称Demons微分同胚非刚性配准相结合的方法,对牙周膜序列切片图像进行级联配准。实验结果表明,相对于有限元非刚性配准而言,对数域对称Demons微分同胚非刚性配准方法具有较好的优势,其配准每张图片的平均时间为有限元法的3.1%,而配准均方误差平均为有限元法的89%。

关键词: 显微图像, 胶原纤维, 图像配准

Abstract: For it is very sensitive to the strength and direction in the production process of paraffin serial sections, it easily results in periodontal ligament tissue deformation and spatial displacement. This paper proposes an efficient image registration method to help reconstruct its spatial structure. On the basis of the Masson staining paraffin sections of rabbit Perio Dontal Ligament(PDL), the serial section images of PDL are obtained in the high optical microscope. And a new registration method that is combined rigid registration in 2D based on regular step gradient descent optimizer and non-rigid registration based on the symmetric log-domain diffeomorphic Demons algorithm is introduced and used for image cascade registration of slice images. The results show that the symmetric log-domain diffeomorphic Demons algorithm is better than Finite Element Method(FEM) registering each slice with 3.1% average processing time of FEM, and 89% average MSE of FEM.

Key words: micrographs image, collagen fiber, image registration