UBC Theses and Dissertations
Fourier regularization applied to spect reconstruction Togane, Dylan Dalmar
Fourier Regularized Emission Computed Tomography (FRECT), a reconstruction method for Emission Computed Tomography (ECT), is presented and compared with other reconstruction methods. The compromise which characterizes ECT reconstruction is between image resolution and stability with respect to noise. FRECT reconstruction consists of minimizing an objective function that has a "goodness of fit" term which fits the data to a specified resolution and a "Fourier regularization" term which stabilizes the reconstruction process. Goodness of fit terms that can be coupled with the Fourier regularizer include the Least Squares term (FRLS) and the Maximum Likelihood term (FRML). Measures are taken to ensure that the fit and regularization terms act on separate areas of the Fourier domain. In this work FRECT is applied to Single Photon Emission Computed Tomography (SPECT) data. Analytically simulated data, Monte-Carlo simulations, phantom experiments and clinically acquired SPECT data are reconstructed. The compromise between resolution and stability is measured for Filtered Back-Projection (FBP), Ordered Subset Expectation Maximization (OSEM), and two variations of FRECT. The sensitivity with respect to experimental noise in the data is found empirically for all reconstruction methods, and analytically for those methods that allow such analysis.
Item Citations and Data