Back to search

National Technical University of Athens (NTUA)

Αξιολόγηση και βελτιστοποίηση ιατρικών απεικονιστικών συστημάτων ποζιτρονιακής τομογραφίας με χρήση αλγόριθμων προσομοίωσης Monte Carlo

Abstract

dc:description

Nuclear medical imaging has contributed significantly during the last decades to the development and deployment of advanced technologies that enhance our medical diagnostic and therapeutic schemes for various types of cancer, as well as to a numerous scientific preclinical and clinical studies aiming at the description and deep understanding of the molecular and biochemical background mechanisms underlying these cancer diseases and their response behavior during treatment. This is clearly an inter-disciplinary scientific field which efficiently combines knowledge from the sciences of physics, medicine, informatics and electrical and computer science engineering in order to determine a three-dimensional digital tomographic image of the metabolic activity spatial distribution of the biological tissues and the cancer tumors in particular. The technology of nuclear medical imaging systems is constantly evolving towards the enhancement of the statistical quality and the diagnostic value of medical images, the minimization of the building and operational cost of the imaging systems and the optimization of the data acquisition protocols. These objectives create, by definition, a trade-off between them, implying that the optimal solution could be the determination of a balance point which will ensure the best possible outcome given the current technology and its limitations. In order to implement an inclusive optimization study of the performance of nuclear medical imaging systems, a set of requirements should be satisfied. First of all, a specific evaluation method of these systems should be defined, based on a number of standardized experiments that are going to be conducted for the measurement of a well-defined set of performance parameters. Furthermore, a methodology should be implemented that will be able to determine the independent effect on each of the above performance parameters of the various system (e.g. energy window, dead time response, etc.) and data acquisition parameters (e.g. radiation dose and scan time) examined. The satisfaction of these requirements is not always practical, or even possible, in the case of real clinical systems, because of their design complexity and the multiple inter-relationships between the examined system parameters. However, nowadays, advanced highly specified Monte Carlo (MC) simulation tools have been developed capable of fulfilling the objectives above, since they allow the performance of numerous well defined simulation measurements and the quantification of their independent effect on each of the parameters of the system at a very low cost. Moreover, their implementation on an object oriented environment provides them with the ability to customize or expand their functionalities for every nuclear medical imaging application. GATE, Geant4 Application for Tomography Emission, is a characteristic example of a MC simulator possessing the capabilities above. On the other hand, the high requirements in computational time remain as the principal drawback of MC tools, particularly when the description model of the simulated geometry is very complex and detailed. For this reason, the application of most of the MC algorithms in clinical imaging studies is considered impractical, because those studies involve sources of relatively high radioactivity7 within complex biological tissues resulting in very slow simulations. In the present thesis, a performance optimization study has been developed for a particular class of Geant4 and GATE models (low-energy electromagnetic processes models), allowing the speed-up of the execution by an average factor of 1.5. Additionally, a Linux PC cluster has been built comprised of 17 processing threads, to further enhance the speed-up by a factor of 17. It was achieved by effectively splitting the initial simulation into a number of simulation jobs that were later automatically executed in parallel with a suitable job scheduler. This is the highest theoretically possible speed-up factor that could have been achieved, given the number of processing threads, indicating an ideal scalability7 of the distributed computing platform. Furthermore, an evaluation methodology is presented for the nuclear medical imaging systems, based on the performance parameters of scatter fraction, sensitivity, count rate and spatial resolution. For this purpose, the GATE models of three commercial clinical positron emission tomography (PET) systems have been designed and implemented in GATE: the ECAT EXACT HR+, the Biograph 2 and the Biograph 6 system. The validation of the models was performed by comparing the experimental and the respective simulated measurements, according to the specifications of the NEMA NU2-2001 measurements protocol. Afterwards, the validated models were used together with anthropomorphic voxelized phantom and source geometries, such as the MOBY and NCAT tools, to perform a realistic simulation study aiming at the optimization of the clinical data acquisition protocol in nuclear medicine. In particular, the study involved the optimization of the administered radiation dose and the scan time for different combinations of energy windows, dead time responses, coincidence time windows and sizes of human patient body. The criterion employed for the determination of either the optimal dose for a given scan time, or the optimal scan time for a given dose, has always been the maximization of the Noise Equivalent Count Rate (NECR) of the system, which is mathematically equivalent to the square of the signal-to-noise (SNR) ratio of the projection data and a reliable indicator of the rate in which statistically significant true coincidence events are recorded.

Degree

thesis:*
Grantor dc:publisher
National Technical University of Athens (NTUA)
Year dc:date
2010

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Karakatsanis, Nikolaos
  • Καρακατσάνης, Νικόλαος

Subjects

dc:subject × 24

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/26959

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Karakatsanis, Nikolaos; Καρακατσάνης, Νικόλαος. Αξιολόγηση και βελτιστοποίηση ιατρικών απεικονιστικών συστημάτων ποζιτρονιακής τομογραφίας με χρήση αλγόριθμων προσομοίωσης Monte Carlo. National Technical University of Athens (NTUA), 2010. http://hdl.handle.net/10442/hedi/26959