Medical image registration methods have been evolving dramatically over the last two decades from being perceived as a rather minor precursor to some medical imaging applications to becoming a major tool itself in medical image analysis. Image registration is a crucial step in many medical image analysis algorithms such as segmentation and labeling, as well as being an important step in population studies involving shape, volume, and functional changes in both health and disease.
This dissertation makes three major contributions: First, a new approach to volumetric intersubject deformable image registration method is proposed for magnetic resonance (MR) images of the human brain. The method is a significant extension of the highly successful method HAMMER by Shen and Davatzikos. The method introduces new image features in order to better identify anatomical correspondences between subjects. A novel approach to generate a dense displacement field based upon the weighted diffusion of the automatically derived feature correspondences is introduced. An extensive validation of the new algorithm was performed on T1 weighted MR images of the human brain. The results were compared with results generated by HAMMER and are shown to yield significant improvements in terms of anatomical alignment, particularly of the brain cortex, as well as in reduced computation time. In the second contribution, the registration algorithm has been adapted to register computed tomography (CT) images of the human pelvis. Furthermore, an approach is presented to register a statistical atlas comprising a point distribution model based on a tetrahedral mesh to a subject's CT scan. This new approach comprises a further augmentation of the core method to maintain the topology of the atlas mesh after deformation as well as incorporating statistical shape information from an atlas to make the registration more robust. Results on CT images of the human pelvis demonstrate the benefits of incorporating prior shape information from the atlas into a registration framework. The third and final contribution of this dissertation presents an approach to preserve the topology of multiple segmented structures as well as their connectivity relationships to each other during registration by incorporating digital homeomorphism into the registration framework.
|Advisor:||Prince, Jerry L.|
|School:||The Johns Hopkins University|
|School Location:||United States -- Maryland|
|Source:||DAI-B 69/04, Dissertation Abstracts International|
|Subjects:||Biomedical engineering, Electrical engineering|
|Keywords:||Brain, Deformable image registration, Image registration, Pelvises|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be