Return to search

Experimental identification of X-ray CT system characteristics for an improved understanding of image processing

The first goal of this study was to develop experimental tools to estimate the blur and noise characteristics of X-ray CT systems. We evaluated the noise power spectrum (NPS) using the averaged periodogram technique. As predicted by theory, varying the image reconstruction filter produced a shift in the frequency content while slice thickness only affected the amplitude of the noise. / In order to quantify the blur, we developed and evaluated two non-linear parametric models of the point spread function (PSF): the Gaussian and damped cosine models. We fitted these models to images of spatially distributed point sources and thus quantified the pattern of shift-variance. We found that the system produced a rotating blur and observed a loss of PSF radial symmetry as the input point source moved away from the center of the field of view. We validated the use of point sources by comparing non-parametric PSF estimates obtained with this input to those found using a correlation-based technique (Wiener-Hopf equation). We gained insight into the design of the input signal, which consisted of pseudo-randomly located holes, through an exhaustive simulation. / The second goal was to investigate how this information could be used to process CT images. We formulated and evaluated a coordinate transformation for shift-invariant restoration of CT images. We developed a simple evaluation procedure which proved beneficial in delimiting the usefulness and detecting limitations of the method. We also formulated a number of recommendations regarding the use of the threshold and Laplacian of a Gaussian segmentation operators taking the shape of the PSF and of the NPS into account.

Identiferoai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:QMM.39338
Date January 1992
CreatorsDoré, Sylvie
PublisherMcGill University
Source SetsLibrary and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada
LanguageEnglish
Detected LanguageEnglish
TypeElectronic Thesis or Dissertation
Formatapplication/pdf
CoverageDoctor of Philosophy (Department of Mechanical Engineering.)
RightsAll items in eScholarship@McGill are protected by copyright with all rights reserved unless otherwise indicated.
Relationalephsysno: 001287955, proquestno: NN74838, Theses scanned by UMI/ProQuest.

Page generated in 0.0018 seconds