Statistically Modified Subset Order Subset Expectation Maximization (SMS-OSEM) SPECT Image Reconstruction
| dc.contributor.advisor | Dawit, Assefa (PhD) | |
| dc.contributor.author | Emebet, Hailu | |
| dc.date.accessioned | 2021-10-21T08:41:14Z | |
| dc.date.accessioned | 2023-11-04T15:22:11Z | |
| dc.date.available | 2021-10-21T08:41:14Z | |
| dc.date.available | 2023-11-04T15:22:11Z | |
| dc.date.issued | 2021-12 | |
| dc.description.abstract | Single photon emission computed tomography (SPECT) imaging is widely implemented in nuclear medicine as its clinical role in the diagnosis and management of several diseases is, many times, very helpful. The goal of reconstruction in tomographic images is as much as possible to recreate the exact image of an object being scanned. Nevertheless, the quality of the reconstructed image depends on the performance of the reconstruction algorithm. One of the most commonly used SPECT reconstruction algorithms in clinical practice is ordered subset expectation maximization (OSEM). It uses a subset of projections to shorten the reconstruction time to reach maximum reconstruction accuracy. However, only few studies are made on the use of statistical measurements to ordered subsets for better reconstruction performance. Hence, the aim of this thesis work is to develop a scheme for use in better SPECT image reconstruction in such a way that there will be improvement on the quality of the reconstructed image as well as reconstruction time. First the SPECT imaging system is simulated and projections of different phantoms (Shepp-Logan, Jaszczak, and Thorax) are calculated. Then the SPECT imaging system matrix is computed using main geometrical parameters of the SPECT instrument. Following this, the projections are grouped into subsets depending on the phantom used. The main contribution of the thesis work is that the subsets are ordered in decreasing order using statistical measurements such as variance, standard deviation and entropy. This measurement allows to find more information about the image being reconstructed. Consequently, a better and faster SPECT image reconstruction method known as Statistically Modified Subset OSEM (SMS-OSEM) is developed. The performance of the SMS-OSEM scheme was compared against the traditional OSEM by varying the number of iterations, the number of subsets and noise levels. The overall performance of the proposed algorithm was checked with different number of iterations (1, 10, 30, 50 and 100) using three different types of phantoms. Based on the phantoms tested using SMS-OSEM algorithm, it was observed that OSEM with variance based subset ordering was able to increase accuracy of the traditional OSEM reconstruction by 15.02% when the number of iterations is low between 1 and 20. For higher number of iterations, the accuracy was increased up to 67.88% depending of the phantom type. In addition, the reconstruction time was reduced by 12.52% for lower number of iterations and by 33.03% for higher number of iterations. The degree of tolerance of the SMS-OSEM method towards noise was tested by adding different amount of Gaussian noise and the algorithm offered better performance than the traditional OSEM scheme. | en_US |
| dc.identifier.uri | http://etd.aau.edu.et/handle/123456789/28270 | |
| dc.language.iso | en_US | en_US |
| dc.publisher | Addis Ababa University | en_US |
| dc.subject | Single-photon emission computed tomography | en_US |
| dc.subject | SPECT | en_US |
| dc.subject | Tomography | en_US |
| dc.subject | OSEM | en_US |
| dc.subject | Statistically modified subset | en_US |
| dc.subject | SMS-OSEM | en_US |
| dc.subject | Image Reconstruction | en_US |
| dc.title | Statistically Modified Subset Order Subset Expectation Maximization (SMS-OSEM) SPECT Image Reconstruction | en_US |
| dc.type | Thesis | en_US |