Use this URL to cite or link to this record in EThOS:
Title: Non-uniform resolution and partial volume recovery in tomographic image reconstruction methods
Author: Ahmad, M.
Awarding Body: University College London (University of London)
Current Institution: University College London (University of London)
Date of Award: 2014
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Acquired data in tomographic imaging systems are subject to physical or detector based image degrading effects. These effects need to be considered and modeled in order to optimize resolution recovery. However, accurate modeling of the physics of data and acquisition processes still lead to an ill-posed reconstruction problem, because real data is incomplete and noisy. Real images are always a compromise between resolution and noise; therefore, noise processes also need to be fully considered for optimum bias variance trade off. Image degrading effects and noise are generally modeled in the reconstruction methods, while, statistical iterative methods can better model these effects, with noise processes, as compared to the analytical methods. Regularization is used to condition the problem and explicit regularization methods are considered better to model various noise processes with an extended control over the reconstructed image quality. Emission physics through object distribution properties are modeled in form of a prior function. Smoothing and edge-preserving priors have been investigated in detail and it has been shown that smoothing priors over-smooth images in high count areas and result in spatially non-uniform and nonlinear resolution response. Uniform resolution response is desirable for image comparison and other image processing tasks, such as segmentation and registration. This work proposes methods, based on MRPs in MAP estimators, to obtain images with almost uniform and linear resolution characteristics, using nonlinearity of MRPs as a correction tool. Results indicate that MRPs perform better in terms of response linearity, spatial uniformity and parameter sensitivity, as compared to QPs and TV priors. Hybrid priors, comprised of MRPs and QPs, have been developed and analyzed for their activity recovery performance in two popular PVC methods and for an analysis of list-mode data reconstruction methods showing that MPRs perform better than QPs in different situations.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available