Date of Award
Master of Engineering Science
Electrical and Computer Engineering
Dr. Roy Eagleson
Image registration is the process of spatially aligning two or more images of a scene into a common coordinate system. Research in image registration has yielded a number of rigid and non-rigid image registration methods capable of registering images of a scene between modalities. In addition, techniques of information visualization have been applied to medical image registration research to produce an atlas based image registration method. This method is capable of registration medical images of a same modality between subjects for comparative studies.
This thesis aims to extend research in image registration by adding to it the visual encoding of time. The visual encoding of time furthers image registration research by enabling the simultaneous analysis of the spatial and temporal relationships that exist between images. The benefit ofregistering images with respect to both space and time is shown through the development of a software application capable of presenting a time space narrative of x-ray images representing a patient’s medical history. This time-space narrative is assembled by performing rigid atlas based image registration on a set of x-ray images and by visually encoding their timestamps to form of an interactive timeline. The atlas based image registration method was selected to ensure that images can be registered to a common coordinate system in cases where images do not overlap. Rigid image registration was assumed to be sufficient to provide the desired visual result.
Subsequent to its implementation, an analysis of the measured uncertainty of the image registration method was performed. The error in manual point pair correspondence selection was measured at more than +/- 1.08 pixels under ideal conditions and a method to calculate the unique standard error of each image registration was presented.
Filion, Jean-Michel, "DESIGN OF A USER INTERFACE FOR THE ANALYSIS OF MULTI-MODAL IMAGE REGISTRATION" (2010). Digitized Theses. 3741.