1. GATE a simulation platform for nuclear medicine based on GEANT4
Application for
Tomographic
Emission
GATE a simulation platform for nuclear medicine based on GEANT4
D. Lazaro, V. Breton
For the OpenGATE Collaboration

2. Imaging in nuclear medicine
Two imaging modalities:
morphological imaging (MRI, scanner, X-rays…)
detailed informations about the patient anatomy
Functional imaging (scintigraphic imaging: SPECT, PET)
non-invasive method: a radioactive drug is injected to the patient and naturally evacuated from the organism
spatial distribution of the radiopharmaceutical in the body
 study of the functional activity of an organ or a specific tissue (e.g tumor)
SPECT
Single photon emitters (99mTc  6 h)
Resolution down to 7 mm
Poor sensitivity
PET
Positron emitters (18FDG  2 h)
Resolution down to 2 mm
Multimodality
Combined imaging modalities (e.g PET/CT)
Resolution improved

3. Why simulations in Nuclear Medicine?
Scanner design
Image Reconstruction
Simulation
Protocol optimization
Data analysis
Algorithm testing
Scatter correction
Quantification recovery

4. Two approaches… General purpose simulation codes (GEANT4, EGS4, MCNP…)
 wide range of physics
 wide community of developers and users
 documentation, maintenance and support
 complexity
 speed
Dedicated simulation codes (PETsim, SimSET, Eidolon,…)
 optimized for nuclear medical imaging applications (geometry, physics...)
 ease of use and fast development
 maintenance, upgrades

5. Simulation requirements
PET/SPECT
dedicated tools
GATE
GEANT4 core
potentialities
Realistic modelisation of PET/SPECT experiments
modelisation of detectors, sources, patient
movement (detector, patient)
time-dependent processes (radioactive decay, movement management, biological kinetics)
Ease-of-use
Fast
Long-term availability, support and training

6. A generic simulation platform for PET/SPECT applications
GATE
A generic simulation platform for PET/SPECT applications
Based on GEANT4
object Oriented Analysis & Design
wide range of physics models
long term availability
upgrades, documentation & support
Specific developments regarding to Nuclear medical imaging needs
material database, sources, readout
time and movement management
Ease-of-use for non C++ programmers
scripting commands to define all paramaters of the simulation (construction of the geometry, specification of the physical processes involved, of the sources...)

7. GATE structure 3 different levels GEANT4 core Developer level
framework and application classes
C++ programming
User level
sequence of scripting commands
geometry construction
physical processes involved
sources (geometry, activity)
movement (type, speed…)
duration of the acquisition
User interface
Application classes
Framework
Geant4

8. Geometry construction by scripting commands
world
A specific mechanism has been developed to help the user construct easily a geometry
scripting commands
geometry = combination of geometric volumes, like ‘russian dolls’
Source
Scanner
Body
Head
Rsector
Crystal
LSO
BGO
D.Strul
Uni Lausanne

9. Geometry examples of GATE applications
Multi-ring PET
D. Strul
IPHE Lausanne
Triple-head gamma camera
S. Staelens
Uni Ghent

10. Source management 15O 11C Multiple sources
controlled by source manager
inserted via scripting
complex geometries: customized GPS
Optimized decay
customized G4 Radioactive Decay Module (RDM)
PET-specific sources
15O
11C

11. Timing Simulation time Time-dependant objects
a clock models the passing of time during experiments
the user defines the experiment timing
Time-dependant objects
updated when time changes
allows programming of movement, tracer kinetics...

12. Physical processes PHOTONS ELECTRONS
Choices of processes via scripting commands:
low energy, standard or inactive
Cut settings
LE photoelectric effect
Standard photoelectric effect
LE Compton scattering
Standard Compton scattering
LE Rayleigh scattering
PHOTONS
ELECTRONS
Standard Gamma conversion
LE Gamma conversion
Standard Ionisation
LE Ionisation
Standard Bremsstrahlung
LE Bremsstrahlung

13. Photons: Standard or low energy processes ?
NaI(Tl) crystal sensitivity : comparison between values computed with different versions of GEANT4 and NIST values
sensitivity (%)
crystal thickness (mm)
crystal thickness (mm)
Relative difference (%)

14. Electrons: Standard or low energy processes ?
energy (keV)
Relative difference (%)
Stopping power in NaI(Tl) in GEANT4 and NIST values
energy (keV)
Relative difference (%)
Stopping power in water in GEANT4 and NIST values

15. Sensitive detector / Digitizer
Hits
Digis
Energy response
Spatial response
Centroid readout
Threshold Electronics
Pre-programmed components
Sensitive detectors
Trajectory analyser
Digitizer
Linear signal processing chain
Modular: set-up via scripting

16. Data output Multiple parallel output channels:
 ROOT (real-time display, storage in ROOT files for further analysis)
 ASCII files
Binary files
Specific scanner formats (e.g Crystal Clear LMF…)
GATE simulation
Sinogram
Reconstructed image

17. The OpenGATE Collaboration
GATE under development since Autumn 2001
Objectives of the collaboration
Develop a reliable generic simulation platform for nuclear medicine
shared development
share results and know-how
allow multiple work axes: development, validation…
Current status: 10 groups
Fields: planar scintigraphy, SPECT, PET, microPET
Organization
Steering committee (one representative / group)
Technical meeting every two months

18. Current efforts of development
Improvement of the detection models
Use of voxelised geometries (sources): under development
Validation work, against experimental data
Small animal imaging gamma camera (LPC – IASA Athens)
Dual-headed gamma camera (University of Ghent)
Strategies to improve computation speed
Parallel computation
Use of computing grids

19. Validation of GATE against experiment
Simulation of a small animal imaging gamma camera
CsI(Tl) crystal array coupled to a PSPMT
Small animal imaging (study of new radiopharmaceuticals)
Lead shielding
PSPMT
LEHR collimator
source
CsI(Tl) crystal array
crystal array

20. GATE deployment on DataGRID
Computation speed becomes (too!) large in case of
voxelised geometry
huge number of events
Potential solution under study at LPC:
deployment of GATE on DataGRID (first tests successful in July 2001)
Input
file
Database
Input file exploding
Output file merging
GRID-GATE output file
GATE output result file
GATE on the GRID

21. For more informations…
GATE web site
Technical Coordinator: Daniel Strul, IPHE Lausanne
Spokesman: Christian Morel, IPHE Lausanne