Dual-source cardiac computed tomography: image quality and dose considerations.

Abstract

Computed tomography (CT) imaging of the heart, most prominently coronary CT angiography, is currently subject to intense interest and is increasingly incorporated into clinical decision-making. In spite of tremendous progress in CT technology over the past decade, the limited temporal resolution has remained one of the most severe problems, especially for cardiac imaging. The novel design concept of dual-source CT (DSCT) allows for an effective scan time of 83 ms independent of heart rate. While large trials are still missing, initial studies have shown improved image quality, especially for visualizing the coronary arteries and detecting coronary artery stenoses. Further investigations have shown that routine beta blockade to lower the heart rate is not necessary to reliably achieve diagnostic image quality. Other applications that may particularly benefit from increased temporal resolution are the analysis of ventricular function and of the cardiac valves. Dose issues which are of interest for cardiac CT in general are discussed in some detail, including a quantitative analysis of dose values and three-dimensional dose distributions. Various strategies to lower radiation exposure are available today, and DSCT offers specific potential for this.

Full text

Page 1

Eur Radiol (2008) 18: 1188–1198
DOI 10.1007/s00330-008-0883-3
COMPUTER TOMOGRAPHY
Stephan Achenbach
Katharina Anders
Willi A. Kalender
Received: 10 September 2007
Revised: 10 December 2007
Accepted: 11 January 2008
Published online: 26 February 2008
# European Society of Radiology 2008
Dual-source cardiac computed tomography:
image quality and dose considerations
Abstract Computed tomography
(CT) imaging of the heart, most
prominently coronary CT angiogra-
phy, is currently subject to intense
interest and is increasingly incorpo-
rated into clinical decision-making. In
spite of tremendous progress in CT
technology over the past decade, the
limited temporal resolution has re-
mained one of the most severe prob-
lems, especially for cardiac imaging.
The novel design concept of dual-
source CT (DSCT) allows for an
effective scan time of 83 ms indepen-
dent of heart rate. While largetrials are
still missing, initial studies have
shown improved image quality, espe-
cially for visualizing the coronary
arteries and detecting coronary artery
stenoses. Further investigations have
shown that routine beta blockade to
lower the heart rate is not necessary to
reliably achieve diagnostic image
quality. Other applications that may
particularly benefit from increased
temporal resolution are the analysis of
ventricular function and of the cardiac
valves. Dose issues which are of
interest for cardiac CT in general are
discussed in some detail, including a
quantitative analysis of dose values
and three-dimensional dose distribu-
tions. Various strategies to lower
radiation exposure are available today,
and DSCToffers specific potential for
this.
Keywords Computed tomography.
Dual-source CT.Heart.Coronary
arteries.Coronary angiography.
Dose
Introduction
Rapid motion of the heart, the small dimensions of cardiac
structures—especially the coronary vessels—and the need
to synchronize image acquisition or image reconstruction to
the cardiac cycle have constituted tremendous obstacles
to the use of computed tomography for cardiac imaging. In
the late 1990s, however, the design of multidetector
computed tomography (CT) systems and the development
of ECG-correlated partial scan image reconstruction
algorithms allowed first attempts at cardiac and coronary
imaging [1–4]. From the very beginning, coronary artery
visualization constituted of the most prominent areas of
interest and “coronary CTangiography” has been the major
driving force for further technical development and clinical
evaluation. The first multidetector CT systems, with four
detector rows, provided a gantry rotation time of
approximately 0.5 s, and imaging of the heart could be
completed in about 30–40 s. Initial reports on visualization
of the coronary artery lumen by four-slice CT after
intravenous injection of contrast agent were published in
the year 2000 [4, 5]. However, because of limitations in
temporal and spatial resolution, the obtained data sets were
frequently of insufficient quality for interpretation: in the
initial studies, up to 30% of arteries were classified as
“unevaluable” [6, 7]. In addition, the requirement of a 30-s
breath-hold was an obstacle to broad clinical use.
Rapid improvements in CT technology occurred in the
subsequent years, and they rapidly led to improved
temporal and spatial resolution as well as to reduced
overall acquisition time [8]. Sixteen-slice CT was intro-
duced around 2002 and 64-slice scanners with gantry
S. Achenbach
Department of Cardiology,
University Erlangen-Nuernberg,
Ulmenweg 18,
91054 Erlangen, Germany
K. Anders
Institute of Diagnostic Radiology,
University Erlangen-Nuernberg,
Maximiliansplatz 1,
91054 Erlangen, Germany
W. A. Kalender (*)
Institute of Medical Physics,
University Erlangen-Nuernberg,
Henkestr. 91,
91052 Erlangen, Germany
e-mail: willi.kalender@imp.uni-
erlangen.de
Tel.: +49-9131-8522310
Fax: +49-9131-8522824

Page 2

rotation times of 330–420 ms, introduced in 2004, are
currently considered “state-of-the-art” equipment for CT
visualization of the coronary arteries. These systems
provide a slice collimation of 0.5–0.625 mm and acquisi-
tion of a data set for coronary visualization can be
accomplished in a breath-hold of 6–12 s duration [9].
Coronary artery visualization by 64-slice CT
A number of studies were able to demonstrate relatively
high accuracy for the detection of hemodynamically
relevant coronary artery stenoses by 64-slice CT. Ob-
viously, the accuracy for stenosis detection will depend on
the operators’ expertise, on the scanner technology that is
used, and also on the prevalence of disease in the patient
population that was studied. For 64-slice CT, sensitivity
values ranging from 73% to 99% and specificity values
between 93 and 97% were obtained in patients referred for
a first diagnostic coronary angiogram [9]. A recent meta-
analysis showed a clear increase in diagnostic accuracy as
technology improved from four-slice to 16-slice and 64-
slice CT, with a pooled sensitivity of 95% and specificity of
93% for 64-slice CT on a per-vessel basis. In a per-patient
analysis, pooled sensitivity for 64-slice CT was 99% and
specificity was 93% [10]. Importantly, patients in all
published studies were somewhat preselected (e.g., exclu-
sion of all patients with renal failure or arrhythmias,
exclusion of unstable patients and patients with acute chest
pain). In addition, interpretation was usually limited to
coronary segments with a diameter of at least 1.5–2.0 mm.
The available data demonstrate that with adequate data
acquisition and reconstruction, as well as experience in
interpretation, high sensitivity and specificity values for the
detection of hemodynamically relevant stenoses can be
achieved by 64-slice CT. All the same, up to 12% of
coronary artery segments had to be excluded from these
analyses because they were deemed to be “unevaluable”
[11]. While sometimes a consequence of massive calcifi-
cation, motion artifacts are another frequent reason for
impaired evaluability in 64-slice CT. The temporal reso-
lution of 64-slice CT is still not sufficient for imaging of
unselected patients. In fact, heart rate is a major predictor of
image quality [12–16] and usually, heart rates of less than
65 bpm or, optimally, less than 60 bpm are suggested in
order to achieve predictably good image quality. This often
requires premedication with beta blockers or other heart-
rate lowering agents and next to concerns of side effects
and logistic challenges, some patients cannot achieve the
target heart rates in spite of pretreatment. Finally, heart rate
variability has been identified as another parameter that
negatively influences image quality [12] in 64-slice CT.
For this reason, improvements in temporal resolution have
been considered very desirable in cardiac CT imaging and,
especially, coronary CT angiography.
Dual-source computed tomography (DSCT)
DSCT was introduced in late 2005 [17]. The system at
hand, the SOMATOM Definition (Siemens Medical
Solutions, Forchheim, Germany) contains two sets of X-
ray tube and detector, which are arranged in a single gantry
at 90° offset. Since data acquired over 180° are required to
reconstruct a single cross-sectional image (an assumption
which holds true for the center of rotation, slightly more
data are needed for off-center regions), a one-quarter
rotation of the gantry is sufficient to collect the data
necessary for one image when the two tubes and detectors
of the DSCT system are operated simultaneously. With a
gantry rotation time of 330 ms, DSCT therefore provides
an effective scan time of 83 ms in the center of rotation.
Notably, this can be achieved without multisegment
Fig. 1 Visualization of the coronary arteries by DSCT. a Here, the
left anterior descending coronary artery (large arrow) which
contains a proximal calcification (small arrow) is visualized in a
curved multiplanar reconstruction. b Three-dimensional visualiza-
tion of the heart and coronary arteries (large arrow left anterior
descending coronary artery, small arrows right coronary artery)
1189

Page 3

reconstruction; we use single-segment reconstruction (with
all data originating from the same cardiac cycle) in all
cases. This allows use of larger pitch values CT with less
overlap compared with single-source and reduced dose.
Artifacts that may occur when data collected during more
than one heart beat are “averaged” for image reconstruction
are avoided. This approach also assures that temporal
resolution is entirely independent of heart rate. Accord-
ingly, we observed high image quality of DSCT for
coronary artery visualization from the beginning of our
DSCT work in 2005 [18] (Fig. 1).
In addition, DSCToffers the possibility for simultaneous
data acquisition with different X-ray energies, which may
permit improved tissue differentiation [17, 19, 20]. One
tube is operated at 80 kV tube voltage, while the other tube
is operated at 140 kV. Thereby the increased temporal
resolution compared with single-source CT (SSCT) is
sacrificed for this mode of operation. Except for phantom
studies, which indicated potential for plaque tissue differ-
entiation [20], the potential of dual-energy imaging for
cardiac applications has not yet been thoroughly explored.
Potential limitations include high image noise, especially
for images acquired at 80 kV, and the lack of thoroughly
validated software to achieve complex tasks, such as
removal of certain plaque components.
Typical image acquisition parameters for DSCT coro-
nary angiography are listed in Table 1. The typical scan
duration is 7–10 s and between 50 and 80 ml of contrast
agent are usually injected at a flow rate of 5 ml/s [18, 21,
22]. Images are most frequently reconstructed using
0.75-mm slice thickness, a slice increment of 0.4–0.5 mm,
and the dedicated kernel “B26f,” which uses a three-
dimensional (3D) noise reduction algorithm [23].
Coronary CT angiography by DSCT
Based on a number of publications, it has become evident
that the image quality of DSCT coronary angiography is
less dependent on heart rate than for 64-slice CTand allows
coronary CT angiography data sets of fully diagnostic
quality to be obtained, even at higher heart rates. We [18]
published an initial series of 14 patients in whom DSCT
coronary angiography data sets were obtained without the
use of beta blockers for lowering the heart rate. With a
mean heart rate of 71 bpm, 98% of all coronary segments
Table 1 Typical data acquisi-
tion parameters for DSCT
coronary angiography
Premedication
Rotation time
Total scan time
Slice width
Collimation
Pitch
Nitrates s.l.
330 ms
7–10 s
0.6 mm (64 overlapping slices)
19.2 mm
0.20
0.22
0.28
0.33
0.39
0.44
0.50
Tube current modulation should always be used; typically, dose
reduction factors of 0.4–0.8 can be achieved
120 kV
100 kV for patients <85 kg
80 kV for very slim patients
e.g., 400+400 mA
e.g., 267 mAs (= 800×0.33); the resulting value is relevant for
image quality considerations
e.g., 534 mAs (= 267×0.6/0.3); in this example, a pitch of 0.3
and an ECG pulsing efficiency factor of 0.6 were assumed; the resulting
value is relevant for dose considerations
50–80 ml at 5 ml/s (consider 6 ml/s in patients >100 kg)
Test bolus or bolus tracking
Heart rate <50 bpm
Heart rate 50–59 bpm
Heart rate 60–69 bpm
Heart rate 70–79 bpm
Heart rate 80–89 bpm
Heart rate 90–99 bpm
Heart rate ≥100 bpm
“ECG pulsing”
Tube voltage
Tube current
mAs value per rotation
Effective mAs value
Contrast agent
Contrast timing
1190

Page 4

were visualized free of motion artifact. Most frequently, an
image reconstruction window starting at 70% of the R-to-
R-interval provided optimal image quality. Similarly,
Johnson et al. [21] published early results of 24 patients
with heart rates between 44 and 92 bpm and did not find
significiant impairment of image quality at high heart rates.
They reported that image reconstruction windows around
70% of the cardiac cycle are usually optimal for low heart
rates, and end-systolic time instants are often better for
heart rates above 75 bpm.
This was investigated in more detail by Leschka et al.
[22] in a recently published series of 60 patients, with heart
rates ranging from 35 to 117 bpm (mean: 69 bpm). In this
series, diagnostic image quality was obtained in 98% of all
coronary segments. The authors thoroughly analyzed the
relationship between heart rate and the distribution of the
“optimal” time windows for reconstructing the data set
within the cardiac cycle. While 70% of the R-R interval
was the most frequent “best time instant” for reconstruc-
tion, that time instant was found between 60% and 70% for
heart rates of less than 60 bpm, between 60 and 80% for
heart rates between 60 and 70 bpm, between 55 and 80%
for heart rates between 70 and 80 bpm, and between 30%
and 80% for heart rates of more than 80 bpm. These results
have implications for tube current modulation, a strategy
used to limit radiation exposure: the time window of full
tube current can be kept rather short for low heart rates and
should be of larger duration for high heart rates. In a further
study which encompassed 80 patients, the same authors
were able to show that in patients with a heart rate below 65
bpm, all coronary segments were visualized with diagnos-
tic image quality, while in patients with a heart rate of 65
bpm or greater, 98% of coronary segments were visualized
with diagnostic image quality [24].
Advantages of DSCT in challenging situations
While image quality and diagnostic accuracy of 64-slice
SSCT coronary angiography was uniformly found to be
high in patients with low heart rates [9], patients with high
heart rates were often considered problematic, which was a
certain limitation to clinical application. The high temporal
resolution of DSCT provides improved image quality in
patients with high heart rates as outlined above (Fig. 2).
This may be especially valuable in patients with acute
coronary syndromes, in whom there may be no time for
heart-rate lowering medication [25]. In addition, other
situations in which image quality has been somewhat
impaired in 16- and 64-slice CT might profit from the
higher temporal resolution of DSCT, since artifacts are
often aggravated by limited temporal resolution. Examples
include the setting of severe coronary calcification—
motion artifacts and subsequent “blurring” of calcium
have been identified as a frequent reason for false-positive
findings in coronary CTangiography [26]—and potentially
also the analysis of coronary artery stents, even though no
data on the accuracy for detection of in-stent restenosis by
DSCT are currently available (Fig. 3).
Scheffel et al. [27] investigated the accuracy of DSCT
for the detection of coronary artery stenoses in a group of
patients with high pretest likelihood of disease. The often
severe atherosclerosis in these patients makes stenosis
detection by coronary CT angiography more challenging,
and with previous scanner generations the accuracy for
stenosis detection in high-risk patient groups was consis-
tently lower than in patients with lower prevalence of
disease [26, 28]. In their report, Scheffel et al. [27]
analyzed a group of 30 patients with a mean age of 63±
11 years and an average heart rate of 70±14 bpm. No beta
blockers were given in preparation for the scan. They
reported a sensitivity of 96%, specificity of 98%, positive
predictive value of 86% and negative predictive value of
Fig. 2 Patient with a heart rate of 125 bpm as a consequence of
pericardial effusion. a Transaxial image as acquired by contrast
enhanced DSCT (arrow cross-section of right coronary artery,
asterisks pericardial effusion). b Maximum Intensity projection
(MIP) or the right coronary artery (arrows)
1191

Page 5

99% for the detection of coronary artery stenoses on a per-
segment basis. This indicates that even in patient groups
with challenging anatomy, DSCT delivers high image
quality and diagnostic accuracy.
Irregular heart rates are also challenging for SSCT,
especially if multi-segment reconstruction algorithms are
used in order to improve temporal resolution. DSCT, with a
high temporal resolution independent of heart rate, appears
to be less susceptible to artifacts by irregular heart rates,
and even the successful imaging of patients in atrial
fibrillation has been anecdotally reported (see Fig. 4) [29].
However, no sufficiently large trials about the accuracy of
DSCT coronary angiography in patients with atrial fibril-
lation or other arrhythmias are currently available and it
remains to be determined whether such promising initial
results will be viable in a clinical context.
One of the most challenging applications of coronary CT
angiography is the visualization and analysis of nonsteno-
tic coronary atherosclerotic plaque (Fig. 5). The small
dimensions of coronary atherosclerotic plaques, along with
the lack of strong contrast between noncalcified plaque
components and the perivascular connective tissue, require
optimal image quality in order to allow detection and,
possibly, quantification and characterization of plaques. In
a phantom study that analyzed the visibility of plaque
components at varying heart rates, Reimann et al. were able
to demonstrate a significant advantage of DSCT over 64-
slice CT for the visualization of coronary atherosclerotic
plaque [30].
Of note, the two tubes of the DSCT scanner can be
combined in order to decrease image noise; as a trade-off,
the high temporal resolution is lost. This option may be
Fig. 4 Visualization of the right
coronary artery in a patient with
atrial fibrillation. a Multiplanar
reconstruction of the right coro-
nary artery (arrows). b ECG
trace. The gray rectangles in-
dicate the time windows used
for image reconstruction. They
are positioned 100 ms before the
peak of the R-wave
Fig. 3 DSCT visualization of a
coronary artery stent (arrows in
b) implanted in the left anterior
descending coronary artery. b A
magnified image segment of a
1192