Validation of GATE Monte Carlo Simulations of The Philips GEMINI TF and TruFlight Select PET CT Scanners Based On NEMA NU2 Standards

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

2012 IEEE Nuclear Science Symposiwn and Medical Imaging Conference Record (NSS/MIC) M10-4

Validation of GATE Monte Carlo Simulations of the


Philips GEMINI TF and TruFlight Select PET/CT
Scanners Based on NEMA NU2 Standards
Andreia Trindade, Pedro Rodrigues, Amy E. Perkins, Michael A. Miller, Manoj Narayanan, Jerome Griesmer,
Chi-Rua Tung, Bin Zhang, Lingxiong Shao, Thomas Laurence, Torsten Solf, Rerfried Wieczorek,

Abstract- The obj ective of this study is to validate the in-house measurements to evaluate it accuracy before further research
GATE simulations of the Philips GEMINI TF and TruFlight and optimizations on the actual systems.
Select PET scanners and evaluate their accuracy for further The purpose of this paper is to validate the Monte Carlo
research and optimization of current and future PET products. models of the Philips GEMINI TF and TruFlight Select PET
GATE results are compared to experimental data obtained
systems [1] implemented in the Geant4 Application for
according to the National Electrical Manufacturers Association
Emission Tomography (GATE) toolkit [2,3]. GATE was used
(NEMA) NU2-2007 standards. A detailed implementation of the
to perform simulations of the National Electrical
geometrical and functional models of the scanners and the NEMA
phantoms was conducted, allowing the evaluation of the simulated Manufacturers Association (NEMA) NU2-2007 protocols [4]
absolute sensitivity, spatial resolution, count rates and the image and the results were compared to experimental data obtained
quality of both systems. All Monte Carlo data production was according to the same standards.
performed according to the NEMA protocols. Simulated data
were converted into the Philips list-mode format and analyzed II. METHODS AND RESULTS
using the same software tools as in the quality control step of the
production line. Good agreement was found between the A detailed implementation of the geometrical and functional
simulated results and the measured data from both scanners. This models of the GEMINI TF and TruFlight Select scanners as
validation study represents an important step towards the use of well as the NEMA phantoms was conducted. The system
these tools as an aid for the optimization of the current covers, shielding layers, patient bed, and crystal modules were
acquisition protocols and the validation of reconstruction and
modeled in GATE macros as shown in Figure 1 for the
data correction techniques.
GEMINI TF scanner as an example.

I. INTRODUCTION

ONTE Carlo simulation codes have became an essential


Mresearch tool in Nuclear Medicine to study the response
of imaging systems such as Positron Emission Tomography I
Bed
(PET) and Single Photon Emission Computed Tomography
Crystals Shielding Covers
(SPECT) scanners. They are mainly used for performance
evaluation of new detectors and design optimization. One of Fig. 1. Implemented model of the GEMINI TF geometry in GATE used to
the advantages of Monte Carlo simulation is the possibility to produce the data for the validation studies. A similar model was implemented
for the TruFlight Select scanner.
change different system parameters and investigate effects of
such modifications, sometimes experimentally impracticable,
All Monte Carlo data production was performed according
on the overall performance of the systems. In addition,
to the NEMA protocols. Simulated data were converted into
simulated data is also being used for the development,
the proprietary Philips list-mode format and analyzed using the
validation and comparative evaluation of image reconstruction
same software tools as in the quality control step of the
techniques and assessment of new correction methods. For any
systems production line. The following sections show the
simulation code it is however important to established a
results obtained for the NEMA sensitivity, spatial resolution,
validation program of the simulations against experimental
count-rate, and IEC image quality performance.

A. Sensitivity
Manuscript received November 16, 2012.
J. Griesmer, Th. Laurence, M. A. Miller, M. Narayanan, A. E. Perkins, L. For the absolute sensItIvIty assessment, the NEMA
Shao, C-H Tung, and B. Zhang are with Philips Healthcare, 595 Miner Road, procedure was simulated with a 700 mm line source uniformly
Cleveland, Ohio 44143, USA 18
filled with 9.25 MBq (0.25 mCi) F. The five concentric
P. Rodrigues, A. Trindade (andreia.trindade@philips.com), and H.
Wieczorek are with Philips Research Europe, High Tech Campus 34, 5656 aluminum sleeves around the source to progressively decrease
AE, Eindhoven, The Netherlands. the attenuation were implemented at the two source locations
T. Soft is with Philips Technologies GmbH, Paulwelstrasse 17D-52074, in the FOY: centered and 10 cm from the center of the
Aachen, Germany.

978-1-4673-2030-6/12/$3l.00 ©2012 IEEE 2546


Authorized licensed use limited to: University of West Attica. Downloaded on December 30,2021 at 17:45:32 UTC from IEEE Xplore. Restrictions apply.
transaxial FOY. A graphical representation of the simulation positions in the FOY, following the NEMA protocol - Figure
setup and the real one is shown in Figure 2. 4.

Fig. 2. Real (left) and simulated (right) setup for the measurement of the Fig. 4. Real (left) and simulated (right) setup for the measurement of the
GEMINI TF sensitivity with the line source and aluminum sleeves (covers are GEMINI TF spatial resolution at different positions in the FOY. A similar
not shown). A similar setup for the simulation of the TruFlight Select system setup for the simulation of the TruFlight Select system was implemented.
was implemented.
Simulated data were reconstructed using Fourier Rebinning
The comparisons of the simulated and measured absolute followed by a 2D FBP reconstruction with the same settings as
sensitivity for both scanners are shown in Figure 3. Deviations for the measured data. Table I and II show the simulated and
are within 1-2%. The experimental sensitivity ratios of the measured spatial resolutions for both scanners. The agreement
GEMINI TF to the TruFlight Select are l.67 (CFOY) and l.63 is within 0.7-9.1%, representing a maximum difference of 400
(10 cm), and the simulated ratio is l.65 at both source microns for the FWHM and 700 microns for the FWTM.
locations.
TABLE I. COMPARISON OF THE S[MULATED AND MEASURED SPATIAL
g 9000 RESOLUTION FOR THE GEMINI TF SYSTEM.
GEMINI TF
� 8800 o Measured Data

o Simulated Data FWHM(mm) Measured Simulated


fr 8600
> Transv.@ I em radius 4.73±0.08 4.37
:� 8400
Transv. Radial @ 10 em radius S.03±0.04 4.91
� 8200

t
QI Transv. Tang. @ 10 em radius 4.98±0.03 4.78
V) 8000

f
Axial 4.S3±0.04 4.41
7800
7600
FWTM(mm) Measured Simulated
7400
Transv.@ I em radius 9.S3±0.08 8.91
7200
Transv. Radial @ 10 em radius 9.76±0.04 9.61
7000
Transv. Tang. @ 10 em radius 9.84±0.03 9.S3
(FOV 10em
Axial 9.S2±0.04 9.46

a- 6000
00 TruFlight Select TABLE II. COMPARISON OF THE SrMULATED AND MEASURED SPATIAL
� 5800 o Measured Data
'" o Simulated Data RESOLUTION FOR THE TRUFLIGHT SELECT SYSTEM.
fr 5600
·f 5400 FWHM(mm) Measured Simulated
:�c 5200
Transv.@ I em radius 4.S1±0.13 4.10
� 5000 Transv. Radial @ 10 em radius 4.S8±0.08 4.27
4800 Transv. Tang. @ 10 em radius 4.86±0.OS 4.S4
4600 Axial 4.31±0.17 4.13
4400

4200 FWTM(mm) Measured Simulated


4000 Transv.@ I em radius 9.08±0.IS 8.42
(FOV 10cm Transv. Radial @ 10 em radius 8.86±0.1O 8.48
Transv. Tang. @ 10 em radius 9.67±0.OS 9.31
Fig. 3. Simulated and measured absolute sensitivities for the Philips Axial 9.01±0.1l 8.79
GEMINI TF (top) and TruFlight Select (bottom) at the two locations of the
line source in the FOY (centered and 10 cm off-center).

C. Count-Rates
B. Spatial Resolution The count rate measurements were simulated by a 70 cm
lS
To assess the spatial resolution of the modeled scanners a uniform line source with 500 MEg (13.5 mCi) F, inserted in
22
3.7 MEg (100 /lCi),1 mm 0 and 1 mm long Na cylinder a scatter phantom with 20 cm 0. A graphical representation of
inside a capillary glass tube was simulated at six different the simulation setup and the real one is shown in Figure 5.

2547
Authorized licensed use limited to: University of West Attica. Downloaded on December 30,2021 at 17:45:32 UTC from IEEE Xplore. Restrictions apply.
GEMINI TF

o
v;-400
e-
:::'350

OSimulated Data • MeasuredData
..
c:::: 300

iE 250 •

cz:: 200

150

100

Fig. 5. Real (left) and simulated (right) setup for the measurement of the
GEMINI TF count-rates with the scatter phantom. A similar setup for the 10 15 20 2S
simulation of the TruFlight Select system was implemented. Activity (kBq/mL)

TruFlight Select
The comparisons of the simulated and measured Trues rate, -;;;-300

o
Co

Randoms rate, and NEe for both the GEMINI TF and �


� 250
TruFlight Select scanners are shown in Figure 6, 7 and 8. i:. o Simulated Data • MeasuredData

GEMINI TF � 200 i • •

v;- 350 ] •
Co a:: 150
�300

0:
100
VI 250

o
,: 50
200 ., .:)

150

100 �IJ�
"
oe
· �tI)tI) __ __-- ____-- ____--
10 15 20
Activity (kBq/mL)
2S

o SimulatedData • Measured Data Fig. 7. Simulated and measured Random coincidences rates as a function
5
� of the source activity. The measured data are from single GEMINI TF (top)
: / ______
and TruFlight Select (bottom) scanners.
10 15 20 25
Activity (kBq/mL)
GEMINI TF
TruFlight Select "VI 120

o
Co

� ce <e °· � o• 0
� 100 � ce
• i :.l O. •
• z oe
8 • 80
8 •
ce
,p . 60 ce
100
.0 ce

� '0

so
o� ��
a SimulatedData • MeasuredData �
OSimulatedData .Measured Data

·
o �--�-- --�-- 2: 1l_�� 10
__

15 20 25
10 15 20 2S Activity (kBq/mL)
Activity (kBq/mL)

TruFlight Select
iii 80
Fig. 6. Simulated and measured True coincidences rates as a function of Co

o

the source activity. The measured data are from single GEMINI TF (top) and �
TruFlight Select (bottom) scanners. � 60
Z O·

A good overall agreement between the simulated and � Ir


'0 •

measurement data can be observed. The simulated maximum O�



NEe and associated activity, as well as the scatter fraction, : v�----- _ ___O 5 ;mU I ated03ta eMeasuredData

were compared. For GEMINI TF, a maximum simulated NEe


of 108.7 kcps for an activity of 16.0 kBq/mL was found in


_ _ __ _
comparison with the typical measured peak NEe value of 110 10 15 20 2S
Activity (kBq/mL)
kcps for an activity of 16 kBq/mL. The simulated scatter
fraction of 30.2% is also in good agreement with the typical
Fig. 8. Simulated and measured NEC rates as a function of the source
measured scatter fraction of 30%. For TruF1ight Select the
activity. The measured data are from single GEMINI TF (top) and TruFlight
maximum simulated NEe of 67.8 kcps for an activity of 16.0 Select (bottom) scanners.
kBq/mL was found. The typical measured peak NEe is 65
kcps for an activity of 16 kBq/mL.

2548
Authorized licensed use limited to: University of West Attica. Downloaded on December 30,2021 at 17:45:32 UTC from IEEE Xplore. Restrictions apply.
An overall good agreement between the simulated and error is 15% for both systems while the simulated ones are
measured data was found for both systems up to the system 11% (GEMINI TF) and 12.3% (TruFlight Select).
cutoffs above 15 kBq/mL which is not included in the
90
simulation model. GEMINI TF
gc 80

0 0
D. Image Quality U 70 j �

� 0
60
The image quality has been simulated according to the
50 0 +
NEMA protocol for a 4: 1 sphere-to-background ratio using • Measured Contrast

40
+ OSimulated Contrast
both the IEC phantom with approximately 52 MBq (1.4 mCi)
1
0
• Measured Background Variance
18 30
+
F and the scatter phantom with a line source of 104 MBq DSimulated Background Variance
18 20 1
(2.8 mCi) F as shown in Figure 9 for the GEMINI TF
10
scanner. 0 0 0 0 0 IJ
10 15 20 25 30 35 40 45 50
Sphere Diameter (mm)

� 1 9
80
TruFlight Select

]
c •
�0
70 •
0
0

60 �

50 • Measured Contrast

40
9 o Simulated Contr3st
• Variance

Measured Background
30
o Simulated Background Variance
20

Fig. 9. Real (left) and simulated (right) setup for the measurement of the
GEMINI TF image quality with the IEC and scatter phantoms. A similar setup
1: 1 5
Cl

10
Cl
15
Cl
20
Cl
25
LI
30 35
LI
40 45 50
for the simulation of the TruFlight Select system was implemented. Sphere Diameter (mm)

Fig. II. Simulated and measured percent contrast and background


The images from the simulated and measured data were
variance for the GEMINI TF (top) and TruFlight Select (bottom) systems.
obtained with the Philips TOF list-mode reconstructor used in
clinical practice. The simulated and measured datasets were
A generally good agreement is observed throughout the
each acquired for 5 minutes and 34 seconds over a single bed
range of spheres sizes. Deviations are slightly higher for
position. The reconstructed images are shown in Figure 10.
GEMINI TF than for TruFlight Select, in particular for the hot
spheres, while the opposite is observed for the cold spheres.

III. DISCUSSION AND CONCLUSIONS

The objective of this work was to validate the GATE


models of the GEMINI TF and TruFlight Select systems
against experimental measurements. The validation was based
on the NEMA NU2-2007 standards for the assessment of PET
system performance. The results of the validation study
indicate that the implemented GATE models can accurately
simulate the main perfonnance parameters of the current
Philips TOF PET systems. It represents the first step in
performing simulation studies for future PET systems in terms
of design and acquisition protocols optimization, data
production for reconstruction validation or development of
new data correction methods.
Fig. 10. Reconstructed images of the NEMA Image Quality phantom from
simulated (right) and measured (left) datasets of GEMINI TF (top) and
TruFlight Select (bottom). REFERENCES

[I] Surti S. et aI., "Performance of Philips GeminiTF PET/CT Scanner with

The quantitative results of the NEMA image quality analysis Special Consideration for its Time-of-Flight Imaging Capabilities", J
Nucl Med. 48(3), 471-80 (2007).
in terms of the contrast and background variability for the hot
[2] Jan S. et aI., "GATE: a simulation toolkit for PET and SPECr", Phys.
(glass spheres with 10, 13, and 17 rum internal diameters) and Med. BioI. 49 4543-4561 (2004).
cold (glass spheres 22 and 37 rum internal diameters) regions [3] Agostinelli S. et aI., "Geant4 - a simulation toolkit", Nucl. Instrum.
Methods Phys. Res. A 506, 250-303 (2003).
are shown in Figure 11. The mean measured relative count
[4] NEMA Standards Publication NU 2-2007 "Performance Measurements
of Positron Emission Tomographs" (2007).

2549
Authorized licensed use limited to: University of West Attica. Downloaded on December 30,2021 at 17:45:32 UTC from IEEE Xplore. Restrictions apply.

You might also like