ORIGINAL ARTICLE

Comparative analysis of cadmium-zincum-

telluride cameras dedicated to myocardial
perfusion SPECT: A phantom study
Orazio Zoccarato, PhD,a Domenico Lizio, MSc,b Annarita Savi, PhD,c
Luca Indovina, PhD,d Camilla Scabbio, MSc,e Lucia Leva, MD,b Angelo del Sole,
MD,e Claudio Marcassa, MD,a Roberta Matheoud, PhD,b Michela Lecchi, PhD,e
and Marco Brambilla, PhDb,f
a
Unit of Nuclear Medicine and Department of Cardiology, S. Maugeri Foundation, IRCCS,
Scientific Institute of Veruno, Veruno, NO, Italy
b
Departments of Medical Physics and Nuclear Medicine, University Hospital ‘Maggiore della
Carità’, Novara, Italy
c
Department of Nuclear Medicine, Hospital San Raffaele, Milan, Italy
d
Department of Medical Physics, Polyclinic Agostino Gemelli, Rome, Italy
e
Department of Health Sciences, University of Milan and Nuclear Medicine Unit, San Paolo
Hospital, Milan, Italy
f
Medical Physics Department, University Hospital of Novara, Novara, Italy

Received Feb 5, 2015; accepted Jun 2, 2015

doi:10.1007/s12350-015-0203-7

Background. This investigation used image data generated by an anthropomorphic

phantom with a cardiac insert for a comparison between two solid state cameras: D-SPECT and
D530c.
Methods. For each camera, two sets (with and without a simulated transmural defect (TD))
of scans were acquired starting from the in vivo standard count statistics in the left ventricle
(LV). Other two acquisitions corresponding to 150% and 50% of the reference count statistics
were acquired. Five performance indices related to spatial resolution, contrast, and contrast-to-
noise ratio (CNR) were analyzed.
Results. D-SPECT showed a lower LV wall thickness and an inferior sharpness than
D530c. No significant differences were found in terms of contrast between LV wall and the
inner cavity, TD contrast or CNR. No significant differences were observed in CNR when
moving from the reference level of count statistics down to 50% or up to 150% of the counts
acquired on the LV.
Conclusions. Our results show that D-SPECT and D530c have different performances. The
lack of differences in the image performance indices along the range of count statistics explored,
indicates that there is the possibility for a further reduction in the injected activity and/or the
acquisition time, for both systems. (J Nucl Cardiol 2015)
Key Words: SPECT Æ Cardiac Æ Cadmium-zincum-telluride

Reprint requests: Marco Brambilla, PhD, Medical Physics Department,

University Hospital of Novara, Novara, Italy; marco.brambilla@
maggioreosp.novara.it
1071-3581/$34.00
Copyright Ó 2015 American Society of Nuclear Cardiology.
 Zoccarato et al
Performance of CZT Cameras in cardiac SPECT
Abbreviations
TD Transmural defect
LV Left ventricle
AC Attenuation correction
CNR Contrast-to-noise ratio
CZT Cadmium-zinc-telluride
D530c Discovery 530c
ROI Region of interest
FWHM Full width at half maximum
CIC Percent contrast between LV wall and
inner cavity
CTD Percent TD contrast in LV wall
See related editorial, doi:10.1007/s12350-
015-0216-2.
INTRODUCTION
Recently two direct-conversion cadmium-zinc-tel-
luride (CZT) detector cameras have become available: the
D-SPECT (Spectrum Dynamics, Caesarea Israel) and
Discovery 530c (D530c, GE Healthcare, Haifa Israel).
These systems enhance the sensitivity of count
detection enabling the reduction of scan times and/or
tracer activity using very different technologies: 9
mobile columns of CZT detectors with wide parallel-
hole collimators for D-SPECT and 19 fixed CZT
detectors with multi-pinhole collimators for the D530c.
Several studies have been published since CZT
cameras were implemented in the clinical practice. The
major aim of these studies was to evaluate the physical
characteristics,1,2 the performances in comparison to
conventional Anger cameras,3,4 their potential to reduce
scan times and/or tracer activity,5,6 and/or their effec-
tiveness in clinical use.7,8
The superior performances reported for these two
CZT systems in comparison to conventional Anger
cameras, although based on identical detector material,
are strictly dependent on the intrinsically different
technology and acquisition techniques which are, in
clinical conditions, potentially capable of producing
significantly different results.
The only published study comparing different CZT
cameras showed different performances depending on the
figure of merit considered.9 However, it was restricted to
only a single point of the count statistic (500 kilocounts
over the left ventricle) and used a simplified phantom,
which is not expected to accurately mimic the conditions
encountered during clinical acquisitions.
The aim of this study was to characterize the
relative performances of the above-mentioned CZT
Journal of Nuclear CardiologyÒ
cameras over a wide range of acquisition count statistics,
using an anthropomorphic phantom and selected phys-
ical figures of merit for the image evaluation. The
application of attenuation correction was also studied
with the D530c.
MATERIALS AND METHODS
Gamma cameras
Two nuclear medicine departments (referred later as
‘centers’) participated to the study allowing the use of the
following gamma cameras: D-SPECT (Center N.1) and D530c
(Center N.2). The scanners characteristics and the acquisition
and reconstruction protocols are detailed in Table 1.
Both D-SPECT and D530c use identical square CZT
modules. Each module consists of a 16x16 pixelated CZT crystal
backed by a proprietary application-specific integrated circuit.
The D-SPECT system uses 9 rotating detector column
blocks of pixelated CZT detectors (pixel size, 2.5 9 2.5 mm)
associated with a wide-angle square-hole tungsten collimator.
A total of 120 projections are recorded by each block.
D-SPECT requires a brief prescan, thereafter the technologist
draws a region of interest (ROI) to define the left ventricle
(LV) for cardiocentric detector column movements. Utilizing
count rate information from the prescan, D-SPECT adjusts
scan time in 1 second increments to attain a predetermined
number of counts. A specific algorithm of iterative recon-
struction is used to compensate for the collimator-related loss
in spatial resolution.10
D530c employs an L-shaped array of 19 stationary
pinhole units, which are placed over an arc of 180°, orientated
to focus on the heart area and arranged in three rows. Each unit
consists of a tungsten-tipped lead-bodied pinhole collimator
with an effective aperture of 5.1 mm, projecting onto 4 CZT
modules of 32x32 CZT pixels (pixel size, 2.5 9 2.5 mm)
arranged in a 2 9 2 square.2 Nineteen projections are thus
recorded, and an iterative reconstruction allows modeling for
collimator geometry. Attenuation correction for the D530c
studies was performed by means of a computed tomography
scan of the phantom.
After phantom acquisitions and image reconstruction,
image quality evaluation and statistical analysis were per-
formed at a core laboratory.
Anthropomorphic Phantom Preparation
An anthropomorphic phantom of the chest, with inserts
simulating lungs, liver, spine, LV wall, LV chamber, and a
transmural defect (TD), was used (Torso PhantomTM and
Cardiac InsertTM, Data Spectrum Corporation, Hillsborough,
NC,USA) (Figure 1). The ratio of radioactivity concentrations
for the different inserts in the phantom study was derived from
a previous publication.11 The TD insert (45° 9 2 cm) was
filled with non-radioactive water and located in mid-septal
position in the LV wall. The lung inserts, filled with
Journal of Nuclear CardiologyÒ Zoccarato et al
Performance of CZT Cameras in cardiac SPECT

Table 1. Details of gamma cameras, acquisition, and reconstruction protocols used for the phantom
studies

D-SPECT Discovery 530c#

Acquisition
Collimator Wide-angle parallel hole Multi-pinhole
Energy window 140 keV ± 10% 140 keV ± 10%
Number of projections 120 (99 blocks) 19
Detector angle between consecutive projections 0.4°-7°* Fixed detector
Reconstruction
Method Iterative 3D Iterative 3D
Number of iterations 9 subsets (4 ? 3) 9 32 60 9 1
Iteration filter Kernel (0.125 mm)
Post reconstruction filter Normalization filter§ Butterworth (order 7,
cutoff 0.37 cm-1)
#
Offline CT was employed to acquire the attenuation map used to correct the emission data.
* Angles are approximately 0.4° for projections passing through cardiac region (as defined by ROI positioning on the left
ventricle on prescan images) and up to 7° for other projections.
§
Proprietary infomations.

Phantom Acquisitions and Reconstruction

In each center, two sets (with and without TD) of scans
were acquired starting from the in vivo standard count
statistics of 1 9 106 counts in the LV. Then, the other two
acquisitions corresponding to 150% and 50% of the reference
count statistics were acquired, changing in each step only the
acquisition time. Images were reconstructed using all the
reconstruction parameters currently recommended for clinical
routine (Table 2). Images were reconstructed with and without
the attenuation correction (AC) for the D530c. No AC is
currently available on the D-SPECT.

Image Quality Evaluation

Figure 1. The anthropomorphic torso phantom used for the
study.
TM
Styrofoam beads and non-radioactive water, were used to
simulate lung tissue attenuation density. The other phantom
chambers were filled with 99mTc solutions of different
radioactivity concentrations: 158 ± 5 kBqmL-1 for LV wall,
95 ± 3 kBqmL-1 for liver, and 8.1 ± 0.6 kBqmL-1 for chest
and LV chamber. Actual activity concentrations in the phan-
tom’s chambers were planned using activity meters with an
accuracy within 5%, as measured during routine quality
controls. All compartment contents were well mixed to avoid
artifacts caused by inhomogeneous activity. The same phan-
tom was used in all centers. The accurate positioning of the
phantom was secured by a flat portion of the phantom which
allowed a reproducible positioning on the imaging bed and by
marking the position of the first acquisition.
After reconstruction, all image sets were sent to a core
laboratory for processing using the following five physical
figures of merit, associated with image quality in cardiac SPECT:
(a) LV wall thickness;
(b) Sharpness index;
(c) Contrast between LV wall and inner chamber;
(d) Contrast-to-noise ratio;
(e) TD contrast with LV wall;
*** The first four indexes were evaluated using the image
sets without TD insert in the LV wall. For each figure of merit,
each measurement was repeated at least fourfold in order to
provide an estimate of the measurement errors. The figures of
merit were evaluated using ImageJ software.12
LV wall thickness. The LV wall thickness was
evaluated as ‘‘full width at half maximum’’ (FWHM) of the
myocardial signal in the mid-ventricular section of the LV
short axis.11 The FWHM was calculated convolving the central
count profile, obtained using ImageJ software, with a
 Zoccarato et al Journal of Nuclear CardiologyÒ
Performance of CZT Cameras in cardiac SPECT

Table 2. Results for the five physical indexes used in this study to evaluate image quality in relation to
scanner/software combinations and count statistics

Thicknessof LV Sharpness
Main effect Description wall (mm) index (cm21) CIC (%) CNR CTD (%)
Scanner/ D-SPECT 13.6 ± 0.7 0.56 ± 0.02 80.9 ± 3.2 4.4 ± 0.9 59.3 ± 4.6
software Discovery 530c 19.6 ± 1.3 0.66 ± 0.04 83.9 ± 5.8 4.5 ± 0.7 55.8 ± 5.9
Discovery 530c-AC 19.6 ± 1.1 0.69 ± 0.01 84.5 ± 3.5 4.8 ± 0.8 56.1 ± 2.8
ANOVA: P value <0.0001* <0.0001* 0.25 0.41 0.11
Study count 1.5 Mcounts 17.6 ± 2.9 0.63. ± 0.06 78.5 ± 5.7 5.0 ± 0.7 57.9 ± 4.4
statistics 1.0 Mcounts 17.7 ± 3.1 0.63. ± 0.07 79.6 ± 5.2 4.5 ± 0.7 57.7 ± 4.8
0.5 Mcounts 17.5 ± 3.1 0.64. ± 0.05 80.5 ± 5.1 4.2 ± 0.9 56.1 ± 5.3
ANOVA: P value 0.67 0.62 0.55 0.018* 0.57

LV, left ventricular; IC, inner chamber; TD, transmural defect; AC, attenuation correction.
* Statistically significant.

theoretical Gaussian one. The true wall thickness of the LV
phantom insert is 10 mm.
Sharpness index. A sharpness index was deter-
mined on the horizontal profile of the mid-ventricular short-
axis slice. This index was computed as the maximal slope of
the decrease in myocardial counts (cm-1) on the epicardial
border of the lateral wall and after exclusion of the low-count
part of the profile (activities \ 25% of the maximal value).9
Contrast between LV wall and inner
chamber. Two ROIs were drawn on the mid-ventricular
short-axis images in order to estimate gLV and gIC as the mean
concentration activity per voxel for LV wall and inner cavity,
respectively. The percent contrast between LV wall and inner
cavity (CIC) was defined as the ratio between measured
contrast (CM) and the true one (CT) as:
CM
CIC ð%Þ ¼  100;
CT
where CM and CT were calculated using:
g  gIC
CM ¼ LV
gLV þ gIC
CLV  CIC
CT ¼
CLV þ CIC
and CLV and CIC are the true activity concentrations in
the inserts simulating the LV and the inner chamber,
respectively. The ideal value for CIC is 100% for all
center acquisitions.11 Figure 2A provides a schematic
representation of the ROI dimensioning and placement
for the calculation of CIC (%).
Contrast-to-noise ratio. This ratio was deter-
mined on the mid-ventricular short-axis slice of each SPECT
recording. Mean myocardial counts and corresponding SD
were determined on a ring-shaped ROI encompassing the
endocardial and epicardial borders. Mean background counts
and corresponding SD were determined on a half-moon-shaped
ROI, at 15 mm from the lateral wall (Figure 2B). Contrast-to-
noise ratio was estimated using the following formula:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
CNR ¼ ðMMy  MBG Þ= SD2My þ SD2BG ;
where MMy is mean myocardial counts, MBG is mean
background counts, SDMy is mean myocardial SD, and
SDBG is mean background SD.9
TD contrast in LV wall. TD contrast in LV wall
was evaluated both in LV short-axis and in horizontal long-
axis images (Figure 2C). ROIs were drawn on the central
short-axis and horizontal long-axis images of the phantom
acquisitions with TD defect in order to estimate gLV and gIC as
the mean counts per voxel for the LV wall and TD,
respectively. The TD contrast in LV wall (CTD) was defined
as the percent difference between gLV and gIC as:
gLV  gTD
CTD ð%Þ ¼  100
gLV
The ideal CTD (%) is 100% for all central acquisitions.11
Statistical Analysis
The impact of the scanner/software combinations and of
the different levels of count statistics on each of the image
quality indexes, was assessed by a two-way factorial
ANOVA. Scanner/software combinations and the relative
level of count statistics were considered as independent
variables (factors) and image quality indexes as dependent
variables.
A post hoc test (Scheffe F test) was performed to identify
the main sources of variability. If a significant F value was
found for one independent variable, then this was referred as a
Journal of Nuclear CardiologyÒ
Figure 2. Example of the ROI dimension and position used
for the evaluation of the image quality indexes: A CIC; B CNR;
C CTD.
main effect. When a main effect was found, then a post hoc test
(Scheffé test) was performed to compare the dependent
variable with the levels of the factor, thus identifying the
Zoccarato et al
Performance of CZT Cameras in cardiac SPECT
main sources of variability. Analysis was performed with
Statistica version 6.0 (StatSoft Inc., Tulsa, OK, USA) using a
two-sided type I error rate of P = .05. Values are reported as
observed means ± standard deviations and shown as least
square averages ± standard errors of the means.
RESULTS
Phantom and Image Quality Evaluation
An example of the reconstructed image sets (best
short-axis slices for TD phantom) for the two cameras
and a relative level of count statistics of 100% and of
50%, is reported in Figure 3. The detailed comparison
between the statistical results for the five physical
indexes used in this study to evaluate image quality in
relation to scanner/software combination and relative
level of count statistics are reported in Table 2, and
summarized below.
LV wall thickness. The scanner/software com-
bination (F = 376; P \ .0001) was a main effect with a
statistically significant impact on the thickness degra-
dation of the LV wall. Post hoc test of the different
scanner/software combinations showed a significant
increase in FWHM values from D-SPECT to D530c
(13.6 ± 0.7 vs 19.6 ± 1.3 mm; P \ .0001). No signifi-
cant differences were observed between D530c and
D530c_AC (19.2 ± 1.1 mm; P = .99). The relative
level of count statistics was not a main effect
(F = 0.4; P = .67) (Figure 4).
Figure 3. The transaxial slices of the TD phantom acquired
with the D_SPECT (left column), the D530c (central column),
and D530C_AC (right column) for count statistics of 1.5,
(upper row), 1 (central row), and 0.5 Mcounts (lower row), are
shown. The display image zoom has been adjusted to take in
account for the different pixel size.
 Zoccarato et al
Performance of CZT Cameras in cardiac SPECT
Figure 4. LV wall thickness, expressed as FWHM (mm), as a
function of scanner/software combination and count statistics.
Points represent least square averages; vertical bars represent
95% confidence intervals.
Sharpness index. The scanner/software combi-
nation (F = 64; P \ .0001) was a main effect with a
statistically significant impact on the sharpness index.
Post hoc test of the different scanner/software combi-
nations showed a significant increase in sharpness values
from D-SPECT to D530c (0.56 ± 0.02 vs
0.66 ± 0.04 cm-1; P \ .0001). No significant differ-
ences were observed between D530c and D530c_AC
(0.69 ± 0.01 cm-1; P = .13). The relative level of count
statistics was not a main effect (F = 0.5; P = .621)
(Figure 5).
Contrast between LV wall and inner
chamber. Neither the scanner/software combination
(F = 1.5; P = .25) nor the count statistics (F = 0.55;
P = 0.59) were main effects with a statistically
Figure 5. Sharpness index as a function of scanner/software
combination and count statistics. Points represent least square
averages; vertical bars represent 95% confidence intervals.
Journal of Nuclear CardiologyÒ
significant impact on the LV wall/inner chamber con-
trast (Figure 6).
Contrast-to-noise ratio. The scanner/software
combination was not a main effect (F = 0.92; P = .41).
The count statistics (F = 4.4; P = .018) was a main
effect with a statistically significant impact on the CNR.
Post hoc count tests showed a significant increase in the
CNR values from 0.5 to 1.5 Mcounts (4.2 ± 0.9 vs
5.0 ± 0.7; P = .020). No significant differences were
observed in CNR between 0.5 and 1 Mcounts (4.2 ± 0.9
vs 4.5 ± 0.7; P = .19) or between 1 and 1.5 Mcounts
(4.5 ± 0.7 vs 5.0 ± 0.7; P = .57) (Figure 7).
TD contrast in LV wall. Neither the scan-
ner/software combination (F = 2.29; P = .11;
Figure 7A) nor the count statistics (F = 0.58;
P = .57) were main effects with a statistically signifi-
cant impact on the TD contrast in LV wall (Figure 8).
DISCUSSION
In the present study, head-to head comparisons were
conducted between the D-SPECT and D530c cameras,
including D530c acquisitions in conjunction with an
offline AC, by analyzing a number of physical figures of
merit related to the quality of myocardial SPECT
images, namely LV wall thickness, sharpness of the
myocardial contour, contrast between the LV wall and
the inner chamber, contrast-to-noise ratio and transmural
defect contrast in the LV wall. This is the first study, to
the best of our knowledge, investigating the behavior of
these performance parameters over a wide interval of
count statistics, related to clinical studies.
Different from previous studies, that used a simpli-
fied cardiac insert, in the present study the parameters
Figure 6. LV wall/inner chamber contrast (CIC (%)) as a
function of scanner/software combination and count statistics.
Points represent least square averages; vertical bars represent
95% confidence intervals.
Journal of Nuclear CardiologyÒ
Figure 7. Contrast-to-noise ratio (CNR) as a function of
scanner/software combination and count statistics. Points
represent least square averages; vertical bars represent 95%
confidence intervals.
Figure 8. Transmural defect contrast (CTD) as a function of
scanner/software combination and count statistics. Points
represent least square averages; vertical bars represent 95%
confidence intervals.
were determined on an anthropomorphic torso phantom
with a cardiac insert, in order to mimic the conditions of
attenuation, scatter, and partial volume effects that can
be encountered in clinical practice.
Wall Thickness
According to our results, the wall thickness mea-
sured on the D530c was of the same order of magnitude
(20 mm) of that measured on conventional gamma
cameras using OSEM reconstruction11 or equipped with
confocal collimators with dedicated reconstruction soft-
ware.13 Moreover, the same conventional gamma
cameras equipped with software for resolution recovery
Zoccarato et al
Performance of CZT Cameras in cardiac SPECT
and/or noise suppression demonstrated values of wall
thickness (16-19 mm) even lower. Liu et al14 also
reported thicker myocardial walls using the D530c and
an anthropomorphic phantom (21-26 mm) in compar-
ison to a conventional Anger camera (16-18 mm). The
use of AC on D530c studies did not result in significant
differences in wall thickness; this was an anticipated
result, since AC is not expected to impact on spatial
resolution. At odds with the previous data, lower values
of wall thickness (13.6 mm) were documented with
D-SPECT, in comparison to both D530c and conven-
tional cameras. This is clearly visible from Figure 3.
Other authors9 have previously demonstrated that the
central spatial resolution is slightly better with the
D530c (6.7 mm) than with D-SPECT (8.6 mm). Thus,
the lower LV wall thickness exhibited by D-SPECT can
be explained only by the reconstruction protocol. The
default algorithm D-SPECT uses is entirely based on
OSEM reconstruction. An initial standard OSEM image
(generally 3 iterations) is first reconstructed on which
the myocardial contours are then estimated to generate a
uniform model of the myocardium. This model is then
used as initial guess (prior) for additional iterations
(generally 4 iterations). Thus, the D-SPECT image is the
optimal result reached by the OSEM iterative recon-
struction using the myocardial contours as initial guess.
Other manufacturers use a FBP image as initial guess for
the iterative reconstructions. It is worth saying that
D-SPECT is neither projecting counts, nor placing the
reconstructed pixels of the myocardium into the best
position. In a similar way, the ‘‘motion-frozen’’
approach proposed by Slomka et al, based on three-
dimensional LV contours identified on the images of the
individual time phases, has demonstrated to improve LV
wall thickness.15
Sharpness Index
The sharpness profile is a parameter influenced by
spatial resolution.16 Imbert et al9 demonstrated values
for the sharpness index of myocardial contours on
human images using the D530c (1.02 cm-1) higher than
those obtained with the D-SPECT (0.92 cm-1). Our
results support this trend, showing significantly superior
values for the sharpness index measured on phantom
images for the D530c (0.66 cm-1) with respect to
D-SPECT (0.56 cm-1) and are in quantitative agree-
ment, as for the D530c, with the results reported by Liu
et al using the same phantom.14 This can be accounted
mainly because of the difference in voxel size used in
image reconstruction, since we can expect higher
degradation of the sharpness index with higher voxel
size. As a matter of fact, the voxel size was 4.92 mm in
D-SPECT studies and 4.0 mm in D530c studies.
 Zoccarato et al
Performance of CZT Cameras in cardiac SPECT
Image Contrast
No statistically significant differences were found in
the contrast between the myocardial wall and the LV
inner cavity among the D-SPECT (81%), the D530c
(84%), and the D530c_AC (85%). Noteworthy, the
values of CIC (%) using CZT cameras were higher than
those observed using conventional gamma cameras
(46%-48%) even when the latter were equipped with
software for resolution recovery (64%) and scatter
corrections were also applied (73%).11
A slight trend toward better values for CTD (%) for
D-SPECT (59.3%) than for D530c (55.8%) can be
observed, as reported in Figure 8, although statisti-
cally not significant. Comparable values of CTD (%)
(58%-60%) were previously reported using a conven-
tional gamma camera with confocal collimators and
without AC in Caobelli et al.13
Contrast-to-Noise Ratio
The CNR values obtained were 4.4, 4.5, and 4.8 for
the D-SPECT, D530c, and D530_AC, respectively;
these values are in substantial agreement with those
reported by Imbert et al for the D-SPECT (4.1) and for
the D530c (4.6).9 However, CNR values in our study are
higher than those reported by Liu et al for the D530c
(*3.8).14
Effect of Count Statistics
For none of the above performance parameters, the
level of the count statistics resulted in a significant factor
(Figures 4, 5, 6, and 8). Hence, in the range of 50%-
150% of the reference count statistic of 1 Million counts
over the LV, the values of wall thickness, sharpness, CIC
(%), and CTD (%) were not affected by the level of
counts acquired. On one side, this suggests that increas-
ing the count statistics above the level recommended by
the manufactures and currently used in the two partic-
ipating in the present study does not bring any additional
benefits in terms of resolution and/or contrast. On the
other side, there is still a possibility for the optimization
of acquisition protocols in order to further reduce the
emission scan duration and/or the injected activity of
radiopharmaceutical.
The only performance parameter which seems
somewhat influenced by the level of count statistics is
the CNR (Figure 7). However, the only comparison that
reached a statistical significance was that between 0.5
Mc and 1.5 Mc, while no statistical significant differ-
ences were observed moving from the reference level of
count statistics down to 50% or up to 150% of the counts
acquired on the LV.
Journal of Nuclear CardiologyÒ
NEW KNOWLEDGE GAINED
Some differences between the D-SPECT and D530c
camera performances are documented, mainly in the LV
wall thickness; however, perfusion defect detection was
similar using both cameras. The most important finding
is that similar performances of both cameras were
obtained with lower and higher acquired counts. Thus,
increased tracer doses are not needed for improved
image quality. Additional decrease of tracer dose or
additional decrease of imaging time may be possible
using CZT cameras.
CONCLUSIONS
The two analyzed SPECT systems, although based on
the same detectors (semiconductor CZT detectors), are
inherently different in both the geometry of data acqui-
sition as well as in the image reconstruction algorithms.
In clinical conditions and using the reconstruction
parameters currently recommended for the clinical
routine, our results show that these different character-
istics lead to significantly different results, mainly in the
LV wall thickness. No significant differences were
found in the contrast between the myocardial wall and
the inner cavity or the contrast on the transmural defect.
The application of the AC to the D530c system, did
not improve any of the figures of merit considered.
However, these results were obtained with a phantom
whose dimensions reproduce a patient with normal
weight, and do not necessarily apply to overweight or
obese patients.
The lack of significant differences in the considered
parameters along the range of count statistics explored,
combined with the increased sensitivity of these systems
(fourfold for the D530c to sevenfold for the D-SPECT)
in comparison to a conventional Anger camera,9,14
seems to indicate that there is still a possibility for a
further reduction in the injected activity and/or the
acquisition time, in order to improve patients’ comfort
and contain both patients’ and operators’ radiation
exposure.
Disclosure
The authors have indicated that they have no financial
conflict of interest.
References
1. Erlandsson K, Kacperski K, Van Gramberg D, Hutton BF. Per-
formance evaluation of D-SPECT: A novel SPECT system for
nuclear cardiology. Phys Med Biol 2009;54:2635-49.
Journal of Nuclear CardiologyÒ
2. Bocher M, Blevis IM, Tsukerman L, Shrem Y, Kovalski G,
Volokh L. A fast cardiac gamma camera with dynamic SPECT
capabilities: Design, system validation and future potential. Eur J
Nucl Med Mol Imaging 2010;37:1887-902.
3. Sharir T, Ben-Haim S, Merzon K, Prochorov V, Dickman D, Ben-
Haim S, et al. High-speed myocardial perfusion imaging initial
clinical comparison with conventional dual detector anger camera
imaging. JACC Cardiovasc Imaging 2008;1:156-63.
4. Esteves FP, Raggi P, Folks RD, Keidar Z, Askew JW, Rispler S,
et al. Novel solid-state detector dedicated cardiac camera for fast
myocardial perfusion imaging: Multicenter comparison with
standard dual detector cameras. J Nucl Cardiol 2009;16:927-34.
5. Gambhir SS, Berman DS, Ziffer J, Nagler M, Sandler M, Patton J,
et al. A novel high-sensitivity rapid acquisition single-photon
cardiac imaging camera. J Nucl Med 2009;50:635-43.
6. Herzog BA, Buechel RR, Katz R, Brueckner M, Husmann L,
Burger IA, et al. Nuclear myocardial perfusion imaging with a
cadmium-zinc-telluride detector technique: Optimized protocol for
scan time reduction. J Nucl Med 2010;51:46-51.
7. Duvall WL, Sweeny JM, Croft LB, Barghash MH, Kulkarni NK,
Guma KA, et al. Comparison of high efficiency CZT SPECT MPI
to coronary angiography. J Nucl Cardiol 2011;18:595-604.
8. Buechel RR, Herzog BA, Husmann L, Burger IA, Pazhenkottil
AP, Treyer V, et al. Ultrafast nuclear myocardial perfusion
imaging on a new gamma camera with semiconductor detector
technique: First clinical validation. Eur J Nucl Med Mol Imaging
2010;37:773-8.
9. Imbert L, Poussier S, Franken PR, Songy B, Verger A, Morel O,
et al. Compared performances of high-sensitivity cameras dedicated
Zoccarato et al
Performance of CZT Cameras in cardiac SPECT
to myocardial perfusion SPECT: A comprehensive analysis of
phantom and human images. J Nucl Med 2012;53:1898-903.
10. Patton JA, Slomka PJ, Germano G, Berman DS. Recent technologic
advances in nuclear cardiology. J Nucl Cardiol 2007;14:501-13.
11. Zoccarato O, Scabbio C, De Ponti E, Matheoud R, Leva L,
Morzenti S, et al. Comparative analysis of iterative reconstruction
algorithms with resolution recovery for cardiac SPECT studies. A
multi-center phantom study. J Nucl Cardiol 2014;21:135-48.
12. Rasband WS. ImageJ. Bethesda, MD: U.S. National Institutes of
Health; 1997-2012. http://imagej.nih.gov/ij/. Accessed 23 Jan
2015.
13. Caobelli F, Kaiser SR, Thackeray JT, Bengel FM, Chieregato M,
Soffientini A, et al. IQ SPECT allows a significant reduction in
administered dose and acquisition time for myocardial perfusion
imaging: Evidence from a phantom study. J Nucl Med
2014;55:2064-70.
14. Liu CJ, Cheng JS, Chen YC, Huang YH, Yen RF. A performance
comparison of novel cadmium-zinc-telluride camera and conven-
tional SPECT/CT using anthropomorphic torso phantom and water
bags to simulate soft tissue and breast attenuation. Ann Nucl Med
2015. doi:10.1007/s12149-015-0952-z.
15. Slomka PJ, Nishina H, Berman DS, Kang X, Akincioglu C,
Friedman JD, et al. ‘‘Motion-Frozen’’ display and quantification
of myocardial perfusion. J Nucl Med 2004;45:1128-34.
16. Shea SM, Kroeker RM, Deshpande V, Laub G, Zheng J, Finn JP,
et al. Coronary artery imaging: 3D segmented k-space data
acquisition with multiple breath-holds and real-time slab follow-
ing. J Magn Reson Imaging 2001;13:301-7.