Loading...

A Novel Structural Based Similarity Measure for MRI and Ultrasound Registration

Moaven, Aria | 2018

630 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 51189 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Fatemizadeh, Emadodin
  7. Abstract:
  8. One of the most important issues in medical image processing is the registration of images with various imaging modalities, because in this case, one can take advantage of these imaging modalities and sometimes fuse and use the useful information of each one in the form of a single image.As it was said, MRI and ultrasound images each have their own disadvantages and advantages, and by considering these two modalities, they have tried to integrate the good features of these two. As we know, one of the destructive cases in the MRI image is the inhomogeneity of the image, a inhomogeneity due to the fact that the main magnetic field is not constant and makes the parts of the image brighter or darker than normal situation. In this study, new similarity measure have been introduced to adapt MRI and ultrasound images, as well as MRI T1 and MRI T2 images, which could allow for an appropriate convergence in terms of inhomogeneity in MRI images. These new similarity measure, using the general framework introduced in the LC2 similarity measure, and using image structure extraction techniques such as SVD Face, FA-SVD Face, LBP, RTV, and LTV, can be more reliable than previous research results. For image inhomogeneity, which sometimes cause error in LC2 similarity measure, the results of this research can lead to a correct results. The results obtained for the proposed methods, based on the similarity measure used in this study for comparing different methods, have improved the compatibility of inhomogeneity MRI images and made-up ultrasound images up to 33% compared to the original LC2 method, and for The matching of images with MRI T1 and MRI T2 inhomogeneity, which the original LC2 method was unable to register these images, the proposed methods in this thesis can successfully complete the registration of two images
  9. Keywords:
  10. Similarity Measure ; Image Registration ; Medical Images ; Medical Ultrasound Images ; Linear Combination Linear Correlation ; Magnetic Resonance Imagin (MRI)Inhomogeneity ; Image Structure Extraction Methods

 Digital Object List

 Bookmark

No TOC