Open Journal of Me di cal Imaging, 2011, 1, 21-25
doi:10.4236/ojmi.2011.12004 Published Online December 2011 (
Copyright © 2011 SciRes. OJMI
Adenosine Stress Perfusion Cardiac MRI: Improving
Image Quality Using a 32-Channel Surface Coil
Thomas R. Burchell, Redha Boubertakh, Saidi Mohiddin, Marc E. Miquel,
Mark A. Westwood, Anthony Mathur, L. Ceri Davies
The London Chest Hospital, London, UK
Received September 14, 2011; revised November 8, 2011; accepted N ovember 18, 2011
Purpose: Adenosine stress CMR is commonly used to assess myocardial ischaemia. Obtaining high quality
images requires maximising signal to noise ratio (SNR) over a large double-oblique field of view (FOV)
whilst minimising artefacts. A 32-channel surface coil may provide a higher SNR over a larger FOV com-
pared to standard coils, possibly leading to improved image quality. Materials and Methods: 50 adenosine
perfusion CMR scans were performed on a Philips Achieva CV 1.5T, with either a 5 or 32-channel coil (25
patients each) using standardised acquisition protocols. 3 short axis slices were acquired per cardiac cycle
and the resulting cine images were scored by two blinded CMR specialists on a quality scale of 1 to 5. Phan-
tom studies were performed using similar acquisition parameters and the SNR was calculated and compared
across a range of acceleration factors. Results: The mean patient age was 62 ± 11 years and 50% of patients
were male. The image quality scores were higher using the 32-channel coil (mean 3.8 ± 0.7 vs 3.2 ± 0.9 p =
0.002). The average phantom SNR was greater for the 32-element coil across the range of acceleration fac-
tors measured (103 vs 86 p = <0.001). Conclusions: The 32-channel coil produces significantly higher qual-
ity images and a higher SNR than the 5-channel coil in routine perfusion CMR.
Keywords: Cardiovascular MRI, Myocardial Perfusion, 32-Channel Coil, Adenosine
1. Introduction
Perfusion Imaging
First pass stress perfusion cardiac magnetic resonance
imaging (Stress Perfusion CMR) has been comprehen-
sively shown to be a safe, reliable and reproducible
method of identifying areas of inducible myocardial is-
chaemia without the use of ionising radiation [1-5]. It has
also been shown to predict cardiovascular morbidity and
mortality in patients with known coronary disease [6].
Acquiring high quality diagnostic images in a routine
clinical setting can be technically challenging due to the
limitations of the current hardware and acquisition pro-
tocols. These limitations can lead to a low in-plan e reso-
lution, resulting in dark rim Gibbs artefact [7], non-uni-
form sensitivity over the FOV, poor SNR and long ac-
quisition times.
Various techniques have been used in an attempt to
overcome these limitations:
A combination of highly accelerated parallel imaging
and temporal encoding (e.g. k-t SENSE) can be used to
increase spatial resolution [8,9] however, using temporal
information requires post processing and is prone to
movement artefacts, which cannot be identified until
after the perfusion scan has been completed. Higher field
strengths to increase contrast to noise ratio (CNR) have
shown promise using 3T but CNR inhomogeneities occur
across the left ventricle, particularly in the inferior wall
[10]. Also 3T magnets are not yet widely available in the
clinical setting.
Increasing the number of surface coils has the poten-
tial advantages of increasing the SNR [11-13] and pro-
vides a better field of view coverage, allowing higher
acceleration factors and hence higher spatial resolution
or faster acquisition times [14].
We assessed the difference in visual quality and SNR
between dedicated cardiac 32-element and standard 5-
element surface coils in routine clinical myocardial per-
fusion imaging using an identical acquisition proto col for
both coils.
2. Materials and Methods
All scans were performed using a Philips Achieva CV
1.5T MR scanner providing a gradient strength of 33
mT/m and maximum slew rate of 180 mT/m/ms (Philips
Medical Systems, Best, The Netherlands). The scanner
was equipped with 32 independent receiver channels. The
scans were performed with either a standard 5-element
(Philips) or 32-element phased-array surface coil (In vivo
Corporation, Gainesville, FL, USA) (Figure 1).
The 32-element coil consists of two parts, an anterior
(flexible) and posterior (fixed), each with 16 elements,
which are 9.1 × 10.6 cm in size and are overlapped to pre-
vent mutual inductance. Both arrays are arranged in a 4 ×
4 honeycomb pattern and measure 30.3 × 36.4 cm overall,
being widest in the left-right direction. The standard
5-element phased-array co il also consists of 2 parts, with
a fixed posterior 3-element and flexible anterior 2-
element coil. The anterior circular coils are 200 mm in
diameter and overlap, the posterior array is made of three
rectangular coils measuring 138 × 200 mm (Figure 1).
2.1. Declaration of Helsinki
This study was conducted in accordance with the prince-
ples of the “Declaration of Helsinki” (as amended in
Tokyo, Venice, Johannesburg, and Edinburgh) and within
the local laws and regulations.
2.2. Patient Studies
2.2.1. Patient s
50 consecutive patients (25 with each coil) through the
routine clinical service underwent perfusion CMR scans.
They were instructed to refrain from any caffeine con-
taining products for >1 2 hours prior to the scan. Both the
Figure 1. 5 and 32-element receiver coils. A schematic draw-
ing of the receiver coils. Pane A shows the posterior section
of the 32-element phased array coil, demonstrating the 16
hexagonal overlapping coil elements. The anterior part uses
a further 16 elements in the same configuration (not shown
for clarity). Pane B shows the anterior 2 circular elements
and the posterior 3 rectangular elements of the 5-element
acquisition protocols and the image manipulation/analyses
were identical for both surface coils).
2.2.2. Stress P rotocol
Intravenous aden osine was infused at 140 µg/Kg/min for
3 minutes with continuous heart rate and blood pressure
recording every minute. Following this, an intravenous
bolus of 0.05 mmol/kg of Gadoteric acid (Dotarem, Guer-
bet, France) was administered via an antecubital fossa
vein on the contralateral arm to the adenosine, using a
power injector (Spectris Solaris EP, Medrad, Indianola,
PA, USA) at a rate of 4 ml/sec with an immediate 0.9%
saline flush of 25 ml at a rate of 4 ml/sec.
3 short axis slices, each of 10 mm thickness, were ac-
quired per cardiac cycle, at the basal, mid papillary and
apical levels of the left ventricle, with the patient free-
breathing throughout the acquisition. We used a single
shot prospectively gated, balanced steady state free pre-
cession sequence (TR 2.6 ms TE 1.3 ms, Flip angle 50˚)
and typical voxel size of 2.8 × 2.9 × 10 mm with a ma-
trix size of 111 - 131 × 256 - 320, with a field of view
(FOV) between 343 mm and 410 mm to minimise alias-
ing artefacts. A saturation pre-pulse was applied prior to
data acquisition. A 90 degrees preparation pulse was
used with a 100 ms recovery time. This saturation pulse
was repeated before the acquisition of every slice within
a cardic cycle. Parallel imaging was used with a sensi-
tiveity encoding (SENSE) factor of 2.1 - 2.5.
2.2.3. Image Ma nipul a ti o n and Analysis
The 3 stress cine images were saved as lossless avi vid-
eos: Codec: Microsoft MPEG-4 V2, Resolution: 1250 ×
925, Colour Depth: 16 million, Frame rate: 13fps.
The videos were then reviewed independently, by two
experienced cardiac MRI specialists (who were blinded
to the choice of coil) in a randomised and anonymised
sequence, using Windows Media Player (Microsoft, Se-
attle, USA). Each video was assessed for image quality,
noise and artefact and given an overall visual quality
score using a 5 point scale, where 1 = Non Diagnostic; 2
= Poor; 3 = Adequate; 4 = Good; 5 = Excellent.
2.3. Phantom Study
Measurements were performed for both coils, using a
rectangular phantom filled with a CuSO4/NaCl solution
(770 mg CuSO4 + 2000 mg NaCl/1000 ml of water) and
with dimensions of 260 mm × 240 mm × 370 mm (L ×
W × H), in a sagittal oblique orientation, using a non
balanced gradient echo sequence, with the same scan
parameters used for human studies with the exception of
a saturation pulse. Reconstruction filters were disabled to
assess the non-smoothed reconstructed images. A range
Copyright © 2011 SciRes. OJMI
of SENSE factors were acquired: 1.0, 1.5, 2.0, 2.3, 2.5,
2.7, 3.0, 3.5 and 4.0, with the majority between 2 and 3
to mirror clinical practice. Measurements were repeated
100 times for each sequence to allow for a temporal
mean to be calculated. The noise standard deviation was
assessed on magnitude images using the multiple acqui-
sition method [13]. To compensate for the Rayleigh
noise distribution in modulus images, a correction factor
of 0.655 was used to correct the underestimated noise
standard deviation. The mean signal (MS) and the stan-
dard deviation (SD) were measured. From this data, the
signal to noise ratio (SNR) was calculated as previously
described [15]. A central circular region of interest,
measuring 110 mm, covering approximately 20% of the
total phantom area was used for analysis, in a location
similar to the heart size and position (see Figure 2).
2.4. Statistics
The image quality scores for the 5 and 32-channel coils
were compared using the Mann-Whitney test for both
observers individually and also with their scores combined.
Intra-observer variability was assessed by comparing the
different observer scores, using Bland Altman analysis.
Signal to noise values were compared using a Student’s
t-test. All p-values are two tailed. Statistical analysis was
performed using SPSS Version 17.0 (SPSS Inc, Chicago,
3. Results
3.1. Patient Studies
Patient characteristics are listed in Table 1. The mean
Figure 2. Phantom measurements. Sagittal oblique phan-
tom cross section. The phantom’s central area (white cir-
cular ROI) is used for signal to noise and standard devia-
tion measurements for both 5 and 32 channel coils.
patient age was 62 ± 11 years (range 36 to 82 years), and 25
(50%) were male. There was an improvement in image
quality score using the 32-channel coil compared to the
5-channel coil for observer 1 (mean score 4.1 ± 0.7 vs. 3.5
± 1 p = 0.04), observer 2 (mean score 3.4 ± 0.7 vs. 3.0 ±
0.6 p = 0.02) and with both observers’ scores combined
(mean score 3.8 ± 0.7 vs. 3.2 ± 0.9 p = 0.002), which was
highly statistically significant. The mean difference in
scores between observer 1 and 2 was 0.6 ± 1.7 (Fig ure 3).
3.2. Phantom Study
Th e min imu m, max imu m an d mea n S NR v alue s fo r bo th
coils are listed in Table 2. The average SNR was greater
for the 32-element coil than the 5-element coil across the
entire range of acceleration factors measured from 1 to 4
(103 vs 86 p = <0.001) (Figure 4).
4. Discussion
There has been a recent and rapid increase in the demand
for myocardial perfusion assessment, especially for pla-
nning reperfusion strategies [16,17].
CMR perfusion imaging has been shown to be
non-inferior to SPECT in demonstrating myocardial is-
chaemia [18] and has the added advantages of providing
validated information about both left ventricular function
and viability as well as being free of ionising radiation.
CMR perfusion, however is still constrained by tech-
nical limitations. Each image is acquired over a relatively
short period of the cardiac cycle as a single shot, yielding
a low SNR, resulting in a relatively low resolutio n.
Ideally, a high quality perfusion study depends on a
number of factors, including; high image resolution, ade-
quate cardiac coverage, a high signal to noise ratio and
minimal artefact.
This study has demonstrated that the 32-channel coil
provides superior observed image quality for qualitative
myocardial perfusion scanning. This is due mainly to
increased SNR and the larger FOV coverage afforded by
this design of phased array coil.
The SNR was shown (in a phantom) to be significantly
higher for the 32-channel coil over a wide range of ac-
Table 1. Baseline characteristics of the study population.
Characteristics5-channel 32-channel p-value
Age 61.5 ± 10.4 62.3 ± 11.4 0.80
Gender 11 male (44%) 14 male (56%)0.57
Weight in kg 79.5 ± 13.9 72.2 ± 9.6 0.035
Peak HR
(during stress) 89.7 ± 20.9 92.0 ± 24.3 0.71
Copyright © 2011 SciRes. OJMI
Table 2. Signal to noise ratios.
SENSE Mean Max MinMean Max Min
1.0 140.4 217.0 96.0151.8 233.7109.6
1.5 112.4 176.7 75.7128.2 182.887.8
2.0 98.6 160.4 60.8108.1 171.171.7
2.3 87.1 131.1 54.9100.5 150.573.8
2.5 86.5 140.9 53.398.3 145.966.7
2.7 79.0 145.7 49.294.7 140.364.8
3.0 72.4 99.9 45.7 91.7 133.461.0
3.5 51.9 85.4 25.3 76.8 117.239.6
4.0 42.5 72.3 20.7 77.7 120.541.4
Signal to noise ratios for each coil over a range of acceleration factors.
Results are expressed as mean, maximum and minimum values within the
central region of interest for each coil and each acceleration factor. Note:
there is a gr eater number of measu remen ts between acc elerat ion facto rs 2 to
3 to mirror standard clinical practice.
celeration factors, most importantly in the range of
common clinical practice (R = 2 - 3).
This increased SNR should allow the use of higher
acceleration factors whilst minimising the noise degrada-
tion seen with the use of standard coils. Speeding up the
acquisition in this way would facilitate either an increase
in cardiac coverage (more slices) or a higher image
resolution, reducing dark rim artefact [9], which is an
area of potential further investigation.
Routine use of the 32-channel coil could provide im-
ages that are easier and faster to interpret by the physic-
cian and may improve workflow through the MRI de-
Figure 3. Intra-observer variability. Bland Altman plot sh-
owing variance from the mean for the 2 observer’s scores.
The 0.95 confidence intervals (±1.96 standard deviations)
are shown as dashed lines.
Figure 4. SNR maps. Signal to noise for the 5 channel coil
(top row) and 32 channel coil (bottom) for 3 SENSE accel-
eration factors. The phase encoding direction is AP.
partment. By extension of the increase in SNR and image
quality, they may prove to have a higher sensitivity and
specificity, however this is beyond the scope of this
Hardware limitations:
The 32-channel coil howev er is not without limitations:
As it is still a new technology this coil is no t currently in
widespread use. It is considerably larger in all dimen-
sions than the standard coil and weighs approximately
1.2 kg more, leading to an effective reduction in bore
size and an increased sensation of confinement. The
subsequent increased coverage of the coil may however,
reduce the dependence on coil positioning, especially
when imaging in the double oblique plane.
Study Limi ta tions:
The study used unpaired patient groups, as it was an
observation of a real world clinical service. As a cones-
quence, there is a small difference in weights between
the groups. Although significantly in creased body weight
may impact on image quality, the difference in groups
was less than 10%. The phantom data provides a paired
comparison between the 2 coil types. The clinical data
was included as an indicator of the clinical utility of in-
creased image quality.
Image quality is a subjective measurement, being an
assessment of a combination of noise, artefact, resolution
and contrast, although in our observations the variation
between observer scores was within the 95% confidence
We have not indicated whether the improved image
quality translates into improved identification of perfu-
sion defects, which will require further investigation and
this is clearly a significant weakness. Assessment of test
sensitivity and specificity would require an invasive
coronary angiogram in all patients, which would not be
normally indicated in the presence of a normal functional
Copyright © 2011 SciRes. OJMI
Copyright © 2011 SciRes. OJMI
In conclusion, the 32-channel phased array coil pro-
duces both a high er SNR and significantly higher quality
images than the standard 5-channel coil in routine stress
CMR despite using near identical acquisition protocols
for both coi l s.
5. References
[1] N. Al-Saadi, E. Nagel, M. Gross, et al., “Noninvasive
Detection of Myocardial Ischemia from Perfusion Re-
serve Based on Cardiovascular Magnetic Resonance,”
Circulation, Vol. 101, No. 12, 2000, pp. 1379-1383.
[2] T. Giang, D. Nanz, R. Coulden, et al., “Detection of
Coronary Artery Disease by Magnetic Resonance Myo-
cardial Perfusion Imaging with Various Contrast Medium
Doses: First European Multi-Centre Experience,” Euro-
pean Heart Journal, Vol. 25, No. 18, 2004, pp. 1657-
1665. doi:10.1016/j.ehj.2004.06.037
[3] E. Nagel, C. Klein, I. Paetsch, et al., “Magnetic Reso-
nance Perfusion Measurements for the Noninvasive De-
tection of Coronary Artery Disease,” Circulation, Vol.
108, No. 4, 2003, pp. 432-437.
[4] J. Schwitter, D. Nanz, S. Kneifel, et al., “Assessment of
Myocardial Perfusion in Coronary Artery Disease by
Magnetic Resonance a Comparison with Positron Emis-
sion Tomography and Coronary Angiography,” Circula-
tion, Vol. 103, No. 18, 2001, pp. 2230-2235.
[5] N. Wilke, M. Jerosch-Herold, A. Zenovich and A. Still-
man, “Magnetic Resonance First-Pass Myocardial Perfu-
sion Imaging: Clinical Validation and Future Applica-
tions,” Journal of Magnetic Resonance Imaging, Vol. 10,
No. 5, 1999, pp. 676-685.
[6] C. Jahnke, E. Nagel, R. Gebker, et al., “Prognostic Value
of Cardiac Magnetic Resonance Stress Tests: Adenosine
Stress Perfusion and Dobutamine Stress Wall Motion
Imaging,” Circulation, Vol. 115, No. 13, 2007, pp. 1769-
1776. doi:10.1161/CIRCULATIONAHA.106.652016
[7] E. Di Bella, D. Parker and A. Sinusas, “On the Dark Rim
Artifact in Dynamic Contrast-Enhanced MRI Myocardial
Perfusion Studies,” Magnetic Resonance in Medicine,
Vol. 54, No. 5, 2005, pp. 1295-1299.
[8] T. Niendorf and D. Sodickson, “Highly Accelerated Car-
diovascular MR Imaging Using Many Channel Technol-
ogy: Concepts and Clinical Applications,” European Ra-
diology, Vol. 18, No. 1, 2008, pp. 87-102.
[9] S. Plein, S. Kozerke, D. Suerder, et al., “High Spatial
Resolution Myocardial Perfusion Cardiac Magnetic
Resonance for the Detection of Coronary Artery Dis-
ease,” European Heart Journal, Vol. 29, No. 17, 2008,
pp. 2148- 2155.
[10] P. Araoz, J. Glockner, K. McGee, et al., “3 Tesla MR
Imaging Provides Improved Contrast in First-Pass Myo-
cardial Perfusion Imaging over a Range of Gadolinium
Doses,” Journal of Cardiovascular Magnetic Resonance,
Vol. 7, No. 3, 2005, pp. 559-564.
[11] C. J. Hardy, H. E. Cline, R. O. Giaquinto, et al., “32-
Element Receiver-Coil Array for Cardiac Imaging,” Mag-
netic Resonance in Medicine, Vol. 55, No. 5, 2006, pp.
1142-1149. doi:10.1002/mrm.20870
[12] Y. Zhu, C. J. Hardy, et al., “Highly Parallel Volumetric
Imaging with a 32-Element RF Coil Array,” Magnetic
Resonance in Medicine, Vol. 52, No. 4, 2004, pp. 869-
877. doi:10.1002/mrm.20209
[13] S. Reeder, B. Wintersperger, O. Dietrich, et al., “Practical
Approaches to the Evaluation of Signal-to-Noise Ratio
Performance with Parallel Imaging: Application with
Cardiac Imaging and a 32-Channel Cardiac Coil,” Mag-
netic Resonance in Medicine, Vol. 54, No. 3, 2005, pp.
748-754. doi:10.1002/mrm.20636
[14] M. Ohliger and D. Sodickson, “An Introduction to Coil
Array Design for Parallel MRI,” NMR in Biomedicine,
Vol. 19, No. 3, 2006, pp. 300-315. doi:10.1002/nbm.1046
[15] M. Fenchel, V. Deshpande, K. Nael, et al., “Cardiac Cine
Imaging at 3 Tesla: Initial Experience With a 32-Element
Body-Array Coil,” Investigative Radiology, Vol. 41, No.
8, 2006, pp. 601-608.
[16] L. Shaw, D. Berman, D. Maron, et al., “Optimal Medical
Therapy with or without Percutaneous Coronary Inter-
vention to Reduce Ischemic Burden: Results from the
Clinical Outcomes Utilizing Revascularization and Ag-
gressive Drug Evaluation (COURAGE) Trial Nuclear
substudy,” Circulation, Vol. 117, No. 10, 2008, pp. 1283-
1291. doi:10.1161/CIRCULATIONAHA.107.743963
[17] P. Tonino, B. De Bruyne, N. Pijls, et al., “Fractional
Flow Reserve versus Angiography for Guiding Percuta-
neous Coronary Intervention,” The New England Journal
of Medicine, Vol. 360, No. 3, 2009, pp. 213-224.
[18] J Schwitter, C. Wacker, A. van Rossum, et al., “MR-
IMPACT: Comparison of Perfusion-Cardiac Magnetic
Resonance with Single-Photon Emission Computed To-
mography for the Detection of Coronary Artery Disease
in a Multicentre, Multivendor, Randomized Trial,” Euro-
pean Heart Journal, Vol. 29, No. 4, 2008, pp. 480-489.