Date on Master's Thesis/Doctoral Dissertation


Document Type

Doctoral Dissertation

Degree Name

Ph. D.


Electrical and Computer Engineering

Committee Chair

Farag, Aly A.

Committee Co-Chair (if applicable)

Graham, James

Committee Member

Inanc, Tamer

Committee Member

Hockenbury, R. Todd

Committee Member

Ng, Chin


Diagnostic imaging; Computer vision in medicine


This dissertation looks at two different categories of the registration approaches: Shape registration, and Image registration. It also considers the applications of these approaches into the medical imaging field. Shape registration is an important problem in computer vision, computer graphics and medical imaging. It has been handled in different manners in many applications like shapebased segmentation, shape recognition, and tracking. Image registration is the process of overlaying two or more images of the same scene taken at different times, from different viewpoints, and/or by different sensors. Many image processing applications like remote sensing, fusion of medical images, and computer-aided surgery need image registration. This study deals with two different applications in the field of medical image analysis. The first one is related to shape-based segmentation of the human vertebral bodies (VBs). The vertebra consists of the VB, spinous, and other anatomical regions. Spinous pedicles, and ribs should not be included in the bone mineral density (BMD) measurements. The VB segmentation is not an easy task since the ribs have similar gray level information. This dissertation investigates two different segmentation approaches. Both of them are obeying the variational shape-based segmentation frameworks. The first approach deals with two dimensional (2D) case. This segmentation approach starts with obtaining the initial segmentation using the intensity/spatial interaction models. Then, shape model is registered to the image domain. Finally, the optimal segmentation is obtained using the optimization of an energy functional which integrating the shape model with the intensity information. The second one is a 3D simultaneous segmentation and registration approach. The information of the intensity is handled by embedding a Willmore flow into the level set segmentation framework. Then the shape variations are estimated using a new distance probabilistic model. The experimental results show that the segmentation accuracy of the framework are much higher than other alternatives. Applications on BMD measurements of vertebral body are given to illustrate the accuracy of the proposed segmentation approach. The second application is related to the field of computer-aided surgery, specifically on ankle fusion surgery. The long-term goal of this work is to apply this technique to ankle fusion surgery to determine the proper size and orientation of the screws that are used for fusing the bones together. In addition, we try to localize the best bone region to fix these screws. To achieve these goals, the 2D-3D registration is introduced. The role of 2D-3D registration is to enhance the quality of the surgical procedure in terms of time and accuracy, and would greatly reduce the need for repeated surgeries; thus, saving the patients time, expense, and trauma.