Loading...

Non-Rigid Medical Image Registration Based on Information Theory

Khorsandi, Rahman | 2011

2062 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 41457 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Fatemizadeh, Emadoddin
  7. Abstract:
  8. The registration of images is a fundamental task in numerous applications in medical image processing. The importance of medical image registraiton due to the imaging systems development in last decades is obvious to every one. Especially the wide employment and different capabilities of these systems has caused more attention to this field of image processing. Th e application of medical image registraiton is extended from clinical diagnosis and treatment evaluation, to image guided surgery. The dimension of images as well as modalities of imaging and imaging subjectshas made a wide variety of problems in this branch of image processing. Registration, briefly speaking, is a geometrical transformation that aligns same structures in two images spatially. If the images tobe registered are from rigid tissues or there is not significant deformation in two images or the structures in two images are one by one, this transform is a rigid transform. However, most of the time, the transform is non-rigid. Two new approach of image registration based on theories of mutual information and image’s features like landmark has been proposed in this thesis. t takes advantage of the information contributed by the overall intensity values of images as well as the extracted landmarks. The experimental results show that the proposed methods is more accurate than only Mutual Information (MI) optimization method and time of process is reduced significantly. Also, user can identify the region of interest for registration which makes more applicable and useful for practical uses
  9. Keywords:
  10. B-Spline Function ; Information Theory ; Mutual Information ; Image Registration ; Nonrigid Transformation

 Digital Object List

  • محتواي پايان نامه
  •   view

 Bookmark

...see more