Content uploaded by Jin Su Kim
All content in this area was uploaded by Jin Su Kim on Apr 12, 2014
Content may be subject to copyright.
IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 60, NO. 2, APRIL 2013 797
Imaging Characteristics of I Between 3D and
2D on Siemens ECAT HR PET Scanner
YoungSubLee, JinSuKim, Hee-JoungKim, Member, IEEE, Sang-Keun Woo, Jong Guk Kim, Ji Ae Park,
Chang Woon Choi, Sang Moo Lim, and Kyeong Min Kim, Member, IEEE
Abstract— I has a complex decay scheme with high gamma
energy and low positron abundance. In this study, comparative
performance measurements of I were performed in terms of
spatial resolution, sensitivity, and image quality. All measurements
were performed using both 2D and3DPETinbothbrainmodeand
whole body mode. The transverse and axial spatial resolutions at 1
cm were 5.56 and 6.07 mm for I, and were 4.58 and 4.77 mm and
for F, respectively. Sensitivities were 0.5 kcps/MBq (2D) and 3.4
kcps/MBq (3D) for I, and 1.8 kcps/MBq (2D) and 9.8 kcps/MBq
(3D) for F. The %contrast of 3D was higher than that of 2D in
I. For I PET imaging, 3D acquisition with brain mode was
highest achievable imaging acquisition mode with ﬁner spatial res-
olution and higher contrast. This result will be useful for IPET
Index Terms—Image quality, reconstruction algorithms, spatial
resolution, whole-body PET.
POSITRON EMISSION TOMOGRAPHY (PET) provides
information on both functional and biochemical processes
in vivo. F-ﬂuorodeoxyglucose (FDG) PET has been routinely
used to diagnose cancer or neurodegenerative diseases. Besides
the F-labeled tracer, many radiopharmaceutical such as C
(half life, 20.4 min), N (half life, 10 min), and O(halflife,
2 min) were also used for PET imaging. Recently, the use of I
has increased because I is a PET tracer that may be used in
personalized dosimetry for I radionuclide –. IPET
can be used to assess the risks and beneﬁts of newly developed
radioisotopes in therapy. Although I has many advantages,
the imaging characteristics of I should be assessed due to the
complex decay scheme of I (602, 723 and 1691 keV) and its
fraction emitted from I was tabulated in Table I.
Manuscript received June 15, 2012; revised November 20, 2012 and January
03, 2013; accepted January 09, 2013. Date of publication February 20, 2013;
date of current version April 10, 2013. This work was supported by grants from
the Nuclear R&D Program (2010-0017587 and 2011-0002286) of the Korea
Science and Engineering Foundation funded by the Ministry of Education, Sci-
ence, and Technology, and supported by grants from the National Research
Foundation (2012013722) of the Korea. (Corresponding author: J. S. Kim.)
Y. S. Lee is with the Korea Institute Radiological and Sciences, Seoul 139-
706, Korea, and also with the Department of Radiological Science, Yonsei Uni-
versity, Wonju 220-710, Korea.
K. M. Kim are with the Korea Institute Radiological and Sciences, Seoul 139-
706, Korea (e-mail: email@example.com).
H.-J. Kim is with the Department of Radiological Science, Yonsei University,
Wonju 220-710, Korea.
Color versions of one or more of the ﬁgures in this paper are available online
Digital Object Identiﬁer 10.1109/TNS.2013.2240012
TAB L E I
THE PHYSICAL CHARACTERISTICS OF IAND F
In I, high energy photons (602, 723 and 1691 keV) are
emitted in a cascade with positrons. Therefore, the PET image
quality is degraded due to cascade photons. Cascade photons
contribute to background image noise in IPET.
Although there were previous reports on the imaging charac-
teristics of I , ,  to the best of our knowledge, there
was no report on comparative imaging studies such as image
qualities of IinSiemensECATHR scanner using brain
mode and whole body mode and with 2D and 3D acquisition
The goal of the present study was the comparison of imaging
characteristics between 3D and 2D in I PET in terms of
image quality to determine the highest achievable PET acquisi-
tion mode for I PET. All experiments were performed using
the brain and whole body modes. The brain mode was designed
for brain PET imaging in which high resolution is provided and
the pixel dimensions is brain mode meet the National Electrical
Manufacturers Association (NEMA) guidelines. In whole body
PET, the pixel dimension was greater than the NEMA require-
ment , .
Regarding internal dosimetry for I PET , whole body
PET scan is needed. In contrast, high resolution PET scan would
be suitable when regional PET for thyroid scan was performed.
Therefore, assessment of imaging characteristics of I for both
brain mode and whole body mode would be needed to interpret
In addition, we measured the spatial resolution and sensi-
tivity of I PET, and then compared with that of F, because
F could be regarded as a gold-standard for the comparison of
II. MATERIALS AND METHODS
To assess the imaging characteristics of Iand F, th e s p a -
tial resolution and image quality were measured on an ECAT
0018-9499/$31.00 © 2013 IEEE
798 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 60, NO. 2, APRIL 2013
HR scanner partly according to the NEMA NU2-2007 stan-
A. System Description
ASiemensECATHR PET scanner was used in this study.
The ECAT HR scanner has retractable septa and can be op-
erated using 2D and 3D PET acquisition modes. The coinci-
dence time window was set to 12 ns, and the energy resolution
was 26%. PET data were acquired within an energy window of
350–650 keV for the brain and whole body modes.
B. Spatial Resolution
The spatial resolution of the system was deﬁned as the ability
to distinguish between two points on an image. To compare
the spatial resolution between Iand F, a point source
(diameter, 1.1 mm) was made. The total activity was
MBq for each acquisition. To measure the spatial resolution
of I, a tissue-equivalent material was wrapped to the point
source to avoid positron escape. Tissue-equivalent material
composed of slabs of parafﬁnwax,ricebagsﬁlled with soda,
gauze coated with petrolatum, and synthetic-based substances.
This tissue-equivalent material was called as Bolus. Point
source was wraped with Bolus (thickness: 5 mm and length:
10 cm) to avoid positron escape. The point source was located
at the center of the axial ﬁeld of view (FOV) and offset 1/4
axial FOV from the center. The point source was positioned at
three locations in the transaxial plane as follows: cm,
cm; cm, cm; and cm,
cm. Therefore, data were acquired at 6 positions using brain
mode and whole body mode, respectively. At each position,
at least one hundred thousand counts were acquired to ensure
enough counts. The images were reconstructed in both brain
and whole body mode using the 2D acquisition mode. For
reconstruction, ﬁltered back projection (FBP) with ramp ﬁlter
was used according to NEMA NU2-2007 standards . The
pixel size of the reconstructed image was 0.51 0.51 mm for
the brain mode and 3.96 3.96 mm for the whole body mode.
The spatial resolution was calculated for each point source
position as full width at half maximum (FWHM) and full width
at tenth maximum (FWTM) of the point spread function deter-
mined in all 3 directions. Radial and tangential resolutions for
each radial position (1 and 10 cm) were averaged for both axial
positions according to NEMA NU2-2007 standards .
The sensitivity is expressed as counts per second where true
coincidence events are detected for a given source strength
(count/sec/MBq). The sensitivity was measured using a NEMA
sensitivity phantom , a mm plastic tubing (inner
diameter, ID: 0.8 mm; outer diameter, OD: 3 mm; length, 70
mm) ﬁlled with Iand F surrounded 5 concentric aluminum
sleeves (ID: 0.39, 0.7, 1.02, 1.34, and 1.66 cm), the scanned
center of FOV and a 10-cm offset from the center. The sensi-
tivity phantom was scanned in 2D and 3D acquisition modes,
and the PET acquisition time was 600 s. Initial activities for
Iand F were 4.625 MBq and 4.107 MBq, respectively.
The total system sensitivity was calculated by dividing the
Fig. 1. Determined of by ﬁtting natural logarithm of the measured counting
rates as function of the sleeve thickness for Iand F.
total count rate in the absence of attenuating material
at corresponding activity. was determined by ﬁtting the
natural logarithm of the measured counting rates as a function
of the sleeve thickness (Fig. 1).
D. Image Quality
To assess the image quality of Iand F, the N E M A I n -
ternational Electrotechnical Commission (IEC) body phantom
was used . The NEMA IEC body phantom consists of 6 hot
spheres (ID, 10, 13, 17, 22, 28, and 37 mm) that were ﬁlled
with Ior F solution . The activity concentration in the
background was 5.3 kBq/cc and the activity concentration in the
spheres was 4 times that of the background activity for both I
and F. The energy window was 350–650 keV.
PET data were acquired for 320 s for both Iand F. The
number of coincidences used to reconstruct the image was 6
Mcounts (for 2D, F), 29 Mcounts (for 3D, F), 2 Mcounts
(for 2D, I), and 16 Mcounts (for 3D, I).
Although the larger 2 spheres (ID, 28 and 37 mm) were ﬁlled
with nonradioactive water according to NEMA NU2-2007,
in this study all spheres were ﬁlled with radioactive water to
assess the %contrast in the largest 2 spheres. The phantom was
positioned at the axial and transaxial center of the scanner FOV.
PET data were acquired using 2D and 3D modes, as well as
both the brain and whole body modes. Transmission PET was
acquired for 105 sec using Ge source (activity: MBq).
Data were corrected for random coincidence, normalization,
dead-time loss, scatter, and attenuation. The OSEM algorithm
(iteration, 2; subset: 16), with Hanning ﬁlter and default clinical
setting, was used for reconstruction. Pixel size was 0.51 0.51
mm for the brain mode and 3.96 3.96 mm for the whole
body mode. To assess the image quality, %contrast, and percent
background variability (%BV) were measured.
Percent Contrast (%Contrast): To calculate %contrast, the
transverse image of the IEC phantom was used in the anal-
ysis. Regions of interest (ROIs) were drawn on each sphere.
The size of ROIs was equal to the inner diameter of the sphere.
Twelve ROIs were drawn in the background using the same size
of spheres ROI in the IEC phantom. The ROIs were drawn on
slices on the center slice. Twelve ROIs were drawn on each
LEE et al.: IMAGING CHARACTERISTICS OF I BETWEEN 3D AND 2D ON SIEMENS ECAT HR PET SCANNER 799
TAB L E I I
COMPARISON OF SPATI A L RESOLUTION BETWEEN IAND F
slice, where a total of 60 background ROIs were drawn on 5
slices. The %contrast for sphere j was calculated as follows:
where is the average counts in ROI for sphere j, is
the average of the background ROI counts for sphere, a is the
activity concentration in the spheres, and is the activity con-
centration in the background.
Percent Background Variability (%BV): The %BV for
sphere j was calculated by
where was the standard deviation of the background ROI
counts for sphere j. SD was calculated as follows:
In this study, the imaging characteristics of Iand Fwere
measured in terms of spatial resolution, sensitivity and image
quality according to NEMA NU2-2007 standards. Both 2D and
3D acquisition modes were used in the measurements.
A. Spatial Resolution
The transverse and axial resolution at an offset of 1 cm from
the center was 5.56 cm and 6.07 cm (brain mode) and 9.74 cm
and 10.61 cm (whole body mode) for I, and 4.58 cm and 4.77
cm (brain mode) and 8.55 cm and 9.12 cm (whole body mode)
for F. The spatial resolution of Iwaslowerthanthatof F.
The spatial resolutions of Iand Finbrainmodeandwhole
body modes are shown in Table II.
Fig. 2. Images of IEC body phantom 2D Brain mode (A), 2D Whole body
mode (B), 3D Brain mode (C), and 3D Whole body mode (D) for I and 2D
Brain mode (E), 2D Whole body mode (F), 3D Brain mode (G), and 3D Whole
body mode (H) for F.
Sensitivities were 0.5 kcps/MBq (2D) and 3.4 kcps/MBq
(3D) for I, and 1.8 kcps/MBq (2D) and 9.8 kcps/MBq (3D)
for F. The branching ratios were not corrected.
C. Image Quality
Fig. 2 shows the PET image for the assessment of image
quality for both Iand F.
Percent Contrast (%Contrast): The %contrasts of Iand
F in brain and whole body mode are Fig. 3(A) and (B). %con-
trast of 3D was higher than that of 2D in I. %contrast of I
was lower than that of F.
Percent Background Variability (%BV): The %BVs of I
and F in brain and whole body modes are shown in Fig. 3(C)
and (D). The %BV of 2D was higher than that of 3D in I.
The %BV of I was higher than that of F. % BV of I-124
was worsened by 7% for 2D brain, 15% for 2D WB, 13% for
3D brain, and 21% for 3D WB.
In this study, we assessed the spatial resolution, sensitivity
and image quality of Iand F. A comparative study was
also performed for both brain and whole body modes and for
2D and 3D acquisitions.
A. Spatial Resolution
We assessed the effects of pixel size and the positron range on
spatial resolution. First, we used brain and whole body modes to
observe the effect of pixel size. Pixel size was 0.51 0.51 mm
in brain mode and 3.96 3.96 mm in whole body mode. In the
NEMA NU2-2007 guidelines , the pixel size should be
of the expected spatial resolution. Because the expected spatial
resolution was 4–5 mm in ECAT HR , the pixel size in whole
body mode was too coarse to measure the spatial resolution in
the ECAT HR scanner. Table 3 shows the result of the spatial
resolution at each condition. We found that the spatial resolu-
tion was 9.74–10.91 mm for I and 8.5–9.9 mm for Fin
800 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 60, NO. 2, APRIL 2013
Fig. 3. Image qualities of IEC body phantom using Iversus
F, % contrast of I (A), % contrast of F (B), % BV of
the whole body mode and 5.56–6.72 mm for I and 4.58–5.18
mm for F in the brain mode. For an actual PET scan such as
pre-therapeutic I PET dosimetry, the spatial resolution of the
whole body mode PET scan would be more informative than the
result according to NEMA guidelines. Second, regarding the ef-
fect of the positron range, we compared the spatial resolutions
of Iand F. The transverse spatial resolution at an offset
of 1 cm was 5.56 mm for I and 4.58 mm for F, respec-
tively. Worse resolution was observed with I compared to
F, which is attributed to the longer positron range of I.
Theoretically, the transverse resolution corrected for source
dimension for both Iand FonECATHR scanner can be
calculated with the following equation , ;
Where is the dimension of the crystal element(s) and
is the ring diameter, p is the positron range (
mm for Iand F) , and b is the block factor (assumed to
zero) . The theoretical spatial resolution value is 5.10 mm
for I and 3.94 mm for F. In our measurements, the spa-
tial resolution with source dimension correction was 5.46 mm
for I and 4.46 mm for F. Our spatial resolution measure-
ments compared well with the predicted values. Small differ-
ences are attributed to the presence of block factor in the actual
The FWHM/FWTM ratio was 0.54 for I and 0.62 for F,
indicating a larger tail for I. In I, the coincidence between
the annihilation photons and correlated single cascade photons
with energy above the lower discriminator value of the energy
window will contribute to a uniform background in the recon-
structed image. The larger tail is probably due to the uniform
background in I.
The sensitivity is affected by many factors such as axial FOV
and the image acquisition method. With regard to the sensitivity
comparison between Iand F, the sensitivity of Fwas3.6
times higher than that of I in 2D and 2.8 times higher in 3D.
The smaller difference of sensitivity in 3D was due to greater
inclusion of correlated single cascade photons in 3D PET. In
2D PET, septa preclude the inclusion of corrected single cascade
photons. With respect to the sensitivity comparison between
2D and 3D, the 3D sensitivity was 5.4 times higher than that
of 2D for F and 6.8 times for I. This tendency was similar
to previous results. In previous studies, the 3D sensitivity was
4–6 times higher than the 2D sensitivity. The larger sensitivity
difference for I was due to greater inclusion of correlated
single cascade photons.
C. Image Quality
In this paper, the image quality of Iand F was assessed in
terms of %contrast, and %BV. All experiments were performed
in 2D and 3D, and brain and whole body modes.
Percent Contrast (%Contrast): The %contrast of Iwas
lower than that of F. The lower %contrast was probably due
to higher positron range of I. The %contrast of 3D was higher
than that of 2D for both Iand F. This tendency was also
found in a previous study . In addition, the difference be-
tween 2D and 3D was larger in Ithanin F. In I, %
contrast of 2D was worsened by 41% for brain mode and 29%
LEE et al.: IMAGING CHARACTERISTICS OF I BETWEEN 3D AND 2D ON SIEMENS ECAT HR PET SCANNER 801
for whole body mode compare to that of 3D. In F, the differ-
ence between 2D and 3D was 1.5% for brain mode and 10% for
whole body mode. For I, the difference was more signiﬁcant
in brain mode.
The %contrast in the brain mode was higher than that in
whole body mode.
We co ul d ﬁnd that %contrast of 3D was higher than that of 2D
and %contrast of brain mode was higher than that of whole body
mode. The higher %contrast in 3D or brain mode was due to
better spatial resolution. In this study, the spatial resolution was
measured in 2D mode. According to the Adam et al.,’s result,
the spatial resolution at center was 4.5 mm for 2D and 4.3 mm
for 3D. The spatial resolution at center in 3D was higher than
that of 2D .
Percent Background Variability (%BV): Regarding the com-
parison between 2D and 3D, the %BV of 3D was equivalent to
2D for both Iand F. The %BV in the brain mode was ap-
proximately higher than that in the whole body mode. The %BV
of I was higher than that of F.
In this study, we compared imaging characteristics of I
and F. At center of FOV, the spatial resolution of Iwas
worsened by 19% for brain mode and 13% for whole body mode
compared to that of F. In addition, the spatial resolution of I
at whole body mode was worsened by 54% compare to that of
We compared the image characteristics such as % contrast,
% background variability of Iand F using various image
acquisition conditions including 2D and 3D, and the brain and
whole body modes. For I PET imaging, 3D acquisition with
brain mode was highest achievable imaging acquisition mode
with ﬁner spatial resolution and higher contrast. This result will
be useful for I PET imaging.
 M. Lubberink, A. van Schie, H. W. de Jong,G.A.vanDongen,andG.
J. Teule, “Acquisition settings for PET of I administered simulta-
neously with therapeutic amounts of I,” J. Nucl. Med., vol. 47, pp.
R. Bares, W. Burchert, and A. Bockisch, “ I PET dosimetry in dif-
ferentiated thyroid cancer: Recovery coefﬁcient in 2D and 3D modes
for PET/CT systems,” Eur. J. Nucl. Med. Mol. Imaging., vol. 35, pp.
 R. A. Gregory, C. A. Hooker, M. Partridge, and G. D. Flux, “Optimiza-
tion and assessment of quantitative I imaging on a Philips Gemini
dual GS PET/CT system,” Eur.J.Nucl.Med.Mol.Imaging.,vol.36,
pp. 1037–1048, 2009.
 S. Surti, R. Scheuermann, and J. Karp, “Correction technique for
cascade gammas in I imaging on a fully-3D, time-of-ﬂight PET
scanner,” IEEE Trans. Nucl. Sci., vol. 56, no. 3, pt. 1, pp. 653–660,
U. Haberkorn, J. Doll, F. Oberdorfer, and W. J. Lorenz, “Performance
evaluation of a whole-body PET scanner using the NEMA protocol,”
J. Nucl. Med., vol. 38, pp. 1614–1623, 1997.
 National Electrical Manufacturers Association, Rosslyn, VA, USA,
“NEMA standards publication NU2-2007: Performance measurements
of positron emiision tomographs,” 2007.
 W. Jentzen, “Experimental investigation of factors affecting the ab-
solute recovery coefﬁcients in I PET lesion imaging,” Phys. Med.
Biol., vol. 55, pp. 2365–2398, 2010.
 S.Robinson,P.J.Julyan,D.L.Hastings, and J. Zweit, “Performance of
a block detector PET scanner in imaging non-pure positron emitters-
modelling and experimental validation with I,” Phys. Med. Biol.,
vol. 49, pp. 5505–5528, 2004.
 W. W. Moses and S. E. Derenzo, “Empirical observation of resolu-
tion degradation on positron emission tomographs utilizing block de-
tectors,” J. Nucl. Med., vol. 34, p. 101, 1993.
D. H. Moon, “Performance measurement of the microPETFocus 120
scanner,” J. Nucl. Med., vol. 48, pp. 1527–1535, 2007.
W. J. Oyen, M. Gotthardt, and E. P. Visser, “Image-quality assessment
for several positron emitters using the NEMA NU 4-2008 standards in
the siemens inveon small-animal PET scanner,” J. Nucl. Med.,vol.51,
pp. 610–617, 2010.
 N. Tomic, C. J. Thomson, and M. E. Casey, “Investigating of the block
effect on spatial resolution in PET detectors,” IEEE Trans. Nucl. Sci.,
vol. 52, pp. 599–605, 2005.
 V. Bettinardi, P. Mancosu, M. Danna, G. Giovacchini, C. Landoni,
M. Picchio, M. C. Gilardi, A. Savi, I. Castiglioni, M. Lecchi, and F.
Fazio, “Two-dimensional vs three-dimensional imaging in whole body
oncologic PET/CT: A discovery-STE phantom and patient study,” Q.
J. Nucl. Med. Mol. Imaging, vol. 51, pp. 214–223, 2007.
 L. E. Adam, J. Zaers, H. Ostertag, H. Trojan, M. E. Bellemann, and G.
Brix, “Performance evaluation of the whole-body PET scanner ECAT
EXACT HR following the IEC standard,” IEEE Trans. Nucl. Sci.,
vol. 44, no. 3, pt. 2, pp. 1172–1179, 1997.