IISc Logo    Title

etd AT Indian Institute of Science >
Division of Electrical Sciences >
Electrical Engineering (ee) >

Please use this identifier to cite or link to this item: http://hdl.handle.net/2005/766

Title: Truncated Data Problems In Helical Cone-Beam Tomography
Authors: Anoop, K P
Advisors: Rajgopal, K
Keywords: Tomography
Medical Tomography
Medical Imaging
Cone-Beam Tomography
Axial Truncation
Lateral Truncation
Helical Scan Computed Tomography
Computed Tomography - Truncation
Non-stationary Linear Model
Lewitt’s Simple Completion Method
Feldkamp Method (FDK)
Submitted Date: Jun-2008
Series/Report no.: G22617
Abstract: This report delves into two of the major truncated data problems in helical cone-beam tomography: Axial truncation and Lateral truncation. The problem of axial truncation, also classically known as the Long Object problem, was a major challenge in the development of helical scan tomography. Generalization of the Feldkamp method (FDK) for circular scan to the helical scan trajectory was known to give reasonable solutions to the problem. The FDK methods are approximate in nature and hence provide only approximate solution to the long object problem. Recently, many methods which provide exact solution to this problem have been developed the major breakthrough being the Katsevich’s algorithm which is exact, efficient and also requires lesser detector area compared to Feldkamp methods. The first part of the report deals with the implementation strategies for methods capable of handling axial truncation. Here, we specifically look at the exact and efficient Katsevich’s solution to long object problem and the class of approximate solutions provided by the generalized FDK formulae. The later half of the report looks at the lateral truncation problem and suggests new methods to handle such truncation in helical scan CT. Simulation results for reconstruction with laterally truncated projection data, assuming it to be complete, gives severe artifacts which even penetrates into the field of view (FOV). A row-by-row data completion approach using Linear Prediction is introduced for helical scan truncated data. An extension/improvement of this technique known as Windowed Linear Prediction approach is introduced. Efficacy of both these techniques are shown using simulation with standard phantoms. Various image quality measures for the resulting reconstructed images are used to evaluate the performance of the proposed methods against an already existing technique. Motivated by a study of the autocorrelation and partial autocorrelation functions of the projection data the use of a non-stationary linear model, the ARIMA model, is proposed for data completion. The new model is first validated in the 2D truncated data situation. Also a method of incorporating the parallel beam data consistency condition into this new method is considered. Performance evaluation of the new method with consistency condition shows that it can outperform the existing techniques. Simulation experiments show the efficacy of the ARIMA model for data completion in 2D as well as 3D truncated data scenario. The model is shown to work well for the laterally truncated helical cone-beam case.
URI: http://hdl.handle.net/2005/766
Appears in Collections:Electrical Engineering (ee)

Files in This Item:

File Description SizeFormat
G22617.pdf3.78 MBAdobe PDFView/Open

Items in etd@IISc are protected by copyright, with all rights reserved, unless otherwise indicated.


etd@IISc is a joint service of NCSI & IISc Library ||
|| Powered by DSpace || Compliant to OAI-PMH V 2.0 and ETD-MS V 1.01