Advanced

Monte Carlo Simulations of a Clinical PET system using the GATE Software

Bertilsson, Henrik (2009)
Medical Physics Programme
Abstract
Introduction: Monte Carlo simulation is a powerful tool in research in the field of nuclear medicine imaging and there is a wide area of applications. The purpose of this work was (1) to understand the basics of the Monte Carlo simulation software Geant4 Application for Tomographic Emission (GATE); (2) to create a model of Philips Gemini TF PET/CT system; (3) to run a number of simulations to compare the performance of the model to published data for this system; and (4) to further investigate how scattered radiation affects the image quality.Material and Methods: The software used in this work was the GATE version 4.0. A model of Philips Gemini TF PET/CT system was defined. A program written in Fortran language was used for data... (More)
Introduction: Monte Carlo simulation is a powerful tool in research in the field of nuclear medicine imaging and there is a wide area of applications. The purpose of this work was (1) to understand the basics of the Monte Carlo simulation software Geant4 Application for Tomographic Emission (GATE); (2) to create a model of Philips Gemini TF PET/CT system; (3) to run a number of simulations to compare the performance of the model to published data for this system; and (4) to further investigate how scattered radiation affects the image quality.Material and Methods: The software used in this work was the GATE version 4.0. A model of Philips Gemini TF PET/CT system was defined. A program written in Fortran language was used for data processing and image reconstruction was done in a program written in Interactive Data Language (IDL). Simulations of spatial resolution were performed in agreement with NEMA NU-2 specifications and full width at half maximum (FWHM) and full width at tenth maximum (FWTM) were reported. Investigations of how scattered radiation and random events depend on different parameters (for example phantom size, source position and source activity) were done. Sensitivity was measured in accordance with the NEMA NU-2 protocol, and a comparison among three different crystal materials (Lutetium Yttrium Orthosilicate (LYSO), Bismuth Germanate (BGO) and Thallium activated Sodium Iodine (NaI(Tl))) was done. The last step was to simulate a voxelized phantom of a human brain and to obtain images that were reconstructed from total, true, scattered and random events.Results: The transverse spatial resolution was measured to be FWHM/FWTM = 5.0/8.1, 5.4/9.4 and 5.7/8.8 mm for source positions/image directions 1 cm/tangential, 10 cm/tangential and 10 cm/radial. The scatter study shows a linear dependence between scatter fraction and phantom diameter. The random fraction shows also linear dependence on source activity and the random rate shows a clearly quadratic dependence which is expected. The sensitivities for the different materials were 2.6, 9.1 and 1.1 cps/kBq for LYSO, BGO and NaI(Tl), respectively. The result for LYSO does not agree with published data, a deviation likely due to the fact that the material defined in the GATE material database does not have the same density as the real material. Images from the voxelized brain phantom simulation agree with the phantom image but no major difference was seen between images reconstructed from total events compared to those reconstructed from true events.Conclusion: GATE is a powerful tool in research in nuclear medicine imaging. It is easy to define and simulate complex situations but the long simulation time and difficulties in data post processing are this software’s main drawbacks. However its advantages compared to other existing Monte Carlo simulation software for emission tomography makes this software to an important tool that will play a major role in future research within this field. (Less)
Abstract (Swedish)
Monte Carlo-simuleringar har fått en ökad betydelse som forskningsverktyg inom nuklearmedicinsk bildgivning eftersom man på ett mycket kontrollerat sätt kan studera nästan vad man vill. I denna studie har datorsimuleringar av ett PET-system gjorts med ett program som heter GATE.

PET (Positron Emission Tomography) är en nuklearmedicinsk bildgivningsteknik som används mer och mer inom sjukvården som ett viktigt hjälpmedel för att upptäcka cancer. Tekniken bygger på att man injicerar ett radioaktivt läkemedel i kroppen som med dess kemiska egenskaper tas upp i önskvärd vävnad, t.ex. cancertumörer. Det radioaktiva sönderfallet resulterar i fotonstrålning som i sin tur kan detekteras av PET-systemet för att sedan ge digitala bilder som... (More)
Monte Carlo-simuleringar har fått en ökad betydelse som forskningsverktyg inom nuklearmedicinsk bildgivning eftersom man på ett mycket kontrollerat sätt kan studera nästan vad man vill. I denna studie har datorsimuleringar av ett PET-system gjorts med ett program som heter GATE.

PET (Positron Emission Tomography) är en nuklearmedicinsk bildgivningsteknik som används mer och mer inom sjukvården som ett viktigt hjälpmedel för att upptäcka cancer. Tekniken bygger på att man injicerar ett radioaktivt läkemedel i kroppen som med dess kemiska egenskaper tas upp i önskvärd vävnad, t.ex. cancertumörer. Det radioaktiva sönderfallet resulterar i fotonstrålning som i sin tur kan detekteras av PET-systemet för att sedan ge digitala bilder som visar omsättningen av läkemedlet i olika delar av kroppen.

Monte Carlo-tekniken bygger på att man i ett datorprogram efterliknar ett bildsystem (t.ex. ett PET-system) och med hjälp av sannolikhetsteorier och slumptal försöker att förutse vad som skulle ha skett i verkligheten. Namnet har sitt ursprung från kasinospel i Monaco, ett ställe där slump och sannolikhet har stor betydelse. Genom att känna till sannolikheter för hur t.ex. fotonstrålning sprids i olika material, så kan man med hjälp av kraftfulla datorer beräkna hur massor av fotoner skulle ha betett sig i verkligheten. Denna teknik lämpar sig därför väl för PET där miljontals fotoner produceras varje sekund.

I detta arbete har en modell av ett PET-system, som används dagligen på sjukhuset, byggts med hjälp av ett Monte Carlo-program som heter GATE och är anpassat för just nuklearmedicinsk bildgivning. I arbetet simulerades PET-systemets prestanda vilka sedan jämfördes med experimentella mätningar och publicerade data av det verkliga systemet. Dessutom studerades i vilken omfattning spridd strålning försämrar kvaliteten hos de slutliga bilderna. Här visar studien på den verkliga styrkan med Monte Carlo-simuleringar d.v.s. möjligheten att få ut information som är omöjlig vid experimentella mätningar. (Less)
Please use this url to cite or link to this publication:
author
Bertilsson, Henrik
supervisor
organization
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Nukleärmedicin
language
English
id
2157100
date added to LUP
2011-09-13 13:04:15
date last changed
2011-12-08 11:45:16
@misc{2157100,
  abstract     = {Introduction: Monte Carlo simulation is a powerful tool in research in the field of nuclear medicine imaging and there is a wide area of applications. The purpose of this work was (1) to understand the basics of the Monte Carlo simulation software Geant4 Application for Tomographic Emission (GATE); (2) to create a model of Philips Gemini TF PET/CT system; (3) to run a number of simulations to compare the performance of the model to published data for this system; and (4) to further investigate how scattered radiation affects the image quality.Material and Methods: The software used in this work was the GATE version 4.0. A model of Philips Gemini TF PET/CT system was defined. A program written in Fortran language was used for data processing and image reconstruction was done in a program written in Interactive Data Language (IDL). Simulations of spatial resolution were performed in agreement with NEMA NU-2 specifications and full width at half maximum (FWHM) and full width at tenth maximum (FWTM) were reported. Investigations of how scattered radiation and random events depend on different parameters (for example phantom size, source position and source activity) were done. Sensitivity was measured in accordance with the NEMA NU-2 protocol, and a comparison among three different crystal materials (Lutetium Yttrium Orthosilicate (LYSO), Bismuth Germanate (BGO) and Thallium activated Sodium Iodine (NaI(Tl))) was done. The last step was to simulate a voxelized phantom of a human brain and to obtain images that were reconstructed from total, true, scattered and random events.Results: The transverse spatial resolution was measured to be FWHM/FWTM = 5.0/8.1, 5.4/9.4 and 5.7/8.8 mm for source positions/image directions 1 cm/tangential, 10 cm/tangential and 10 cm/radial. The scatter study shows a linear dependence between scatter fraction and phantom diameter. The random fraction shows also linear dependence on source activity and the random rate shows a clearly quadratic dependence which is expected. The sensitivities for the different materials were 2.6, 9.1 and 1.1 cps/kBq for LYSO, BGO and NaI(Tl), respectively. The result for LYSO does not agree with published data, a deviation likely due to the fact that the material defined in the GATE material database does not have the same density as the real material. Images from the voxelized brain phantom simulation agree with the phantom image but no major difference was seen between images reconstructed from total events compared to those reconstructed from true events.Conclusion: GATE is a powerful tool in research in nuclear medicine imaging. It is easy to define and simulate complex situations but the long simulation time and difficulties in data post processing are this software’s main drawbacks. However its advantages compared to other existing Monte Carlo simulation software for emission tomography makes this software to an important tool that will play a major role in future research within this field.},
  author       = {Bertilsson, Henrik},
  keyword      = {Nukleärmedicin},
  language     = {eng},
  note         = {Student Paper},
  title        = {Monte Carlo Simulations of a Clinical PET system using the GATE Software},
  year         = {2009},
}