Data Processing Toolbox for PET scanners
Moreno Galera, Amalia (2014)
Moreno Galera, Amalia
2014
Master's Degree Programme in Biomedical Engineering
Tieto- ja sähkötekniikan tiedekunta - Faculty of Computing and Electrical Engineering
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2014-04-09
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tty-201405131156
https://urn.fi/URN:NBN:fi:tty-201405131156
Tiivistelmä
This thesis presents the design and implementation of a Data Processing Toolbox to process and reconstruct the information acquired by different PET scanners. The Toolbox is divided in three processes: generate realistic list-mode files, histogram list-mode data sets and reconstruct sinograms into 3D images.
The first process produces list-mode data from numerical phantoms. This tool is useful in the early stages of developing a PET scanner because permits to process the Toolbox performance when there is not available data. The list-mode file real appearance is due to the bootstrapping and shuffling tools. In addition, the bootstrapping function, which calculates statistically similar sinograms than the acquired, can be used for results realizations and phantom measurements. The second process histograms events from different PET scanners into sinograms. The adaptability to the different scanners is related to the Radon transform dependence with the scanners geometry and format. The third process corrects, rebins and reconstructs sinograms into 3D images. In order to avoid artefacts, different methods have been used to correct the sinograms based on the quantification of events. A rebinning method was selected to work with 3D data due to the computational time decrease and the use of 2D algorithms for the image reconstruction. This method transforms the 3D data to 2D where the information contained in the oblique planes is added to the transaxial planes. The Maximum Likelihood Expectation Maximization (MLEM) iterative algorithm has been used for the reconstruction of the sinograms into the image domain.
The proper Toolbox functionality was demonstrated by numerical simulations and data from two real PET scanners. The numerical phantom and real data are in the two PET data recording options: list-mode file and sinogram data formats. In addition, the evaluation of the Toolbox efficiency is presented studying the parameter dependency with the executing time of the functions.
As a conclusion, this thesis describes a useful Toolbox for a PET scanner development and use. The proposed Toolbox can be used for different scanners due to its adaptability to the scanner geometry. Moreover, due to the Toolbox modularity, the functions can be used independently.
The first process produces list-mode data from numerical phantoms. This tool is useful in the early stages of developing a PET scanner because permits to process the Toolbox performance when there is not available data. The list-mode file real appearance is due to the bootstrapping and shuffling tools. In addition, the bootstrapping function, which calculates statistically similar sinograms than the acquired, can be used for results realizations and phantom measurements. The second process histograms events from different PET scanners into sinograms. The adaptability to the different scanners is related to the Radon transform dependence with the scanners geometry and format. The third process corrects, rebins and reconstructs sinograms into 3D images. In order to avoid artefacts, different methods have been used to correct the sinograms based on the quantification of events. A rebinning method was selected to work with 3D data due to the computational time decrease and the use of 2D algorithms for the image reconstruction. This method transforms the 3D data to 2D where the information contained in the oblique planes is added to the transaxial planes. The Maximum Likelihood Expectation Maximization (MLEM) iterative algorithm has been used for the reconstruction of the sinograms into the image domain.
The proper Toolbox functionality was demonstrated by numerical simulations and data from two real PET scanners. The numerical phantom and real data are in the two PET data recording options: list-mode file and sinogram data formats. In addition, the evaluation of the Toolbox efficiency is presented studying the parameter dependency with the executing time of the functions.
As a conclusion, this thesis describes a useful Toolbox for a PET scanner development and use. The proposed Toolbox can be used for different scanners due to its adaptability to the scanner geometry. Moreover, due to the Toolbox modularity, the functions can be used independently.