• Monte Carlo Simulation
• Parallel Computing
• Parallelhole collimator
• Septal penetration
• SPECT

Abstract

Single Photon Emission Computed Tomography (SPECT) could be considered as a milestone in terms of biomedical imaging technique, which visualizes Functional processes in-vivo, based on the emission of gamma rays produced within the body. The most distinctive feature of SPECT from other imaging modalities is that it is based on the tracer principle, discovered by George Charles de Hevesy in the first decade of the twentieth century.

As known by everyone, the metabolism of an organism is composed of atoms within a molecule which can be replaced by one of its radioactive isotopes. By using this principle, we are able to follow and detect pathways of the photons which are emitted from the radioactive element inside the metabolism.

SPECT produces images by using a gamma camera which consists of two major functional components, the collimator and the radiation detector. The collimator is a thick sheet of a heavy metal like lead, tungsten of gold with densely packed small holes and is put just in front of the photon detector. The radiation detector converts the gamma rays into scintillation light photons.

In conventional SPECT, scanners utilize a parallel hole collimator. Defining a small solid angle, each collimator hole is located somewhere along this line and the photons might reach the detector by passing through these holes. Subsequently, we can create projection images of the radioisotope distribution. The quantity of photons which come to the radiation detector through the collimator holes specifies the image quality regarding signal to noise ratio.

One of the crucial parts of all SPECT scanners is the collimator design. The main part of this dissertation is to investigate performance characteristics of YAP-(S)PETII scanner collimator and to obtain collimator characteristics curves for optimization purposes.

Before starting the collimator performance investigation of YAP-(S)PETII scanner, we first performed simulation of it in SPECT mode with point source Tc-99m to measure collimator and system efficiency by using GATE–the Geant4 Application for Emission Tomography. GATE is an advanced, flexible, precise, opensource Monte Carlo toolkit developed by the international OpenGATE collaboration and dedicated to the numerical simulations in medical imaging. We obtained the results of collimator and system efficiency in terms of collimator length, radius and septa by using GATE_v4. Then, we compared our results with analytical formulation of efficiency and resolution. For those simulation experiments, we found that the difference between the simulated and the analytical results with regard to approximated geometrical collimator efficiency formulation of H. Anger, is within 20%. Then, we wrote a new ASCII sorter algorithm, which reads ASCII output of GATE_v4 and then creates a sinogram and reconstructs it to see the final simulation results.
At the beginning, we used the analytical reconstruction method, filtered back projection (FBP), but this method produces severely blurred images. To solve this problem and increase our image quality, we tried different mathematical filters, like ramp, sheep-logan, low-pass cosine filters. After all of those studies mentioned above, we learned that GATE_v4 is not practical to measure collimator efficiency and resolution. On the other hand, the results of GATE_v4 did not show directly septal penetrated photon ratio. Under the light of these findings, we decided to develop a new user-friendly ray tracing program for optimization of low energy general purpose (LEGP) parallelhole collimators.

In addition, we tried to evaluate the image quality and quantify the impact of high-energy contamination for I-123 isotope imaging. Due to its promising chemical characteristics, Iodine-123 is increasingly used in SPECT studies. 159 keV photons are used for imaging, however, high-energy photons result in an error in the projection data primarily by penetration of the collimator and scattering inside the crystal with energy close to the photons used for imaging. One of the way to minimize this effect is using a double energy window (DEW) method, because, it decreases noise in main (sensitive) energy window. By using this method, we tried to determine the difference between simulated and experimental projection results and scattered photon ratio (Sk) value of YAP-(S)PETII scanner for I-123 measurements.

The main drawback of GATE simulations is that they are CPU-intensive. In this dissertation to handle this problem, we did the feasibility study of the Fully Monte Carlo based implementation of the system matrix derivation of YAP-(S)PETII scanner by using XtreemOS platform. To manage lifecycle of the simulation on the top XtreemOS, we developed a set of scripts. The main purpose of our study is to integrate a distributed platform like XtreemOS to reduce the overall simulation completion time and increase the feasibility of SPECT simulations in a research environment and establish an accurate and fast method for deriving the system matrix of the YAP-(S)PETII scanner by using Monte Carlo simulation approach.

We developed also the ML-EM Algorithm to the reconstruct our GATE simulation results and to derive the system matrix directly from GATE output. In addition to the accuracy consideration, we intend to develop a flexible matrix derivation method and GATE output reconstruction tool.