Assessment of Regional Myocardial Oxygenation
Changes in the Presence of Coronary Artery
Stenosis With Balanced SSFP Imaging at 3.0T:
Theory and Experimental Evaluation in Canines
Rohan Dharmakumar, PhD,1*Jain Mangalathu Arumana, Dipl Ing,1Richard Tang, MD,1
Kathleen Harris, BA,1Zhouli Zhang, MD, PhD,1and Debiao Li, PhD1,2
Purpose: To examine the dependence of steady-state free-
precession (SSFP) -based myocardial blood-oxygen-level-
dependent (BOLD) contrast on field strength using theoret-
ical and experimental models.
Materials and Methods: Numerical simulations using a
two-pool exchange model and a surgically prepared dog
model were used to assess the SSFP-based myocardial BOLD
signal changes at 1.5T and 3.0T. Experimental studies were
performed in eight canines with pharmacological vasodilation
under various levels of left circumflex coronary artery steno-
sis. Experimentally obtained BOLD signal changes were cor-
related against microsphere-based true flow changes.
Results: Theoretical results showed that, at 3.0T, relative
to 1.5T, a threefold increase in oxygen sensitivity can be
expected. Experimental studies in canines showed near
similar results—a 2.5 ? 0.2-fold increase in BOLD sensi-
tivity at 3.0T relative to 1.5T (P ? 0.05). Based on the scatter
gram of BOLD data and microsphere data, it was found that
the minimum regional flow difference that can be detected
with SSFP-based myocardial BOLD imaging at 1.5T and
3.0T were 2.9 and 1.6, respectively (P ? 0.05).
Conclusion: This study demonstrated that SSFP-based
myocardial BOLD sensitivity is substantially greater at
3.0T compared with 1.5T. The findings here suggest that
SSFP-based myocardial BOLD imaging at 3.0T may have
the necessary sensitivity to detect the clinically required
minimum flow difference of 2.0.
Key Words: BOLD MRI; 3T; SSFP; oxygen; blood; coronary
J. Magn. Reson. Imaging 2008;27:1037–1045.
© 2008 Wiley-Liss, Inc.
CORONARY ARTERY DISEASE (CAD) is the most com-
mon form of heart disease and is the primary cause of
death among men and women in the United States (1).
Typically, CAD leads to a reduction in the size of the
coronary artery lumen resulting in reduced oxygen sup-
ply to the heart muscle. Accurate early detection of flow
deficits may permit interventional revascularization
procedures to re-establish flow to the hypoperfused re-
gions (2). The absence of revascularization increases
the risk of sudden cardiac death (3).
Significant research efforts have been devoted to the
development of noninvasive methods that can identify
the presence of CAD on the basis of functional status of
the myocardium, but establishment of such methods
remains challenging. Current approaches include sin-
gle-photon emission computed tomography (SPECT)
and positron emission tomography (PET; 4,5). SPECT
imaging is the technique most widely used for detecting
both metabolic activity and perfusion. However, SPECT
studies are limited by low spatial resolution and expo-
sure to ionizing radiation. PET is a promising method
for detecting regional myocardial blood flow differences
but is not widely available and is also limited by low
spatial resolution and/or potentially harmful ionizing
Assessment of myocardial perfusion reserve changes
secondary to coronary artery disease (6), in the pres-
ence of vasodilatory agents, have been successfully
demonstrated using the first-pass MRI method (7,8).
This method relies on bolus injection(s) of gadolinium-
based contrast medium and is performed using rapid
imaging techniques with multislice capabilities. Al-
though the advantages of first-pass method are well
known, there are also pragmatic limitations with the
method. First, given the need to capture the first pas-
sage of the contrast medium at relatively high temporal
resolution, compromises are often made in myocardial
coverage and/or optimal spatial resolution. Second, the
susceptibility differences between the contrast medium
and myocardium, reduced spatial resolution, and/or
cardiac motion are known to result in dark “rimming”
artifacts that mimic perfusion deficits. Finally, gadolin-
ium conjugates have been implicated in the cause of
nephrogenic systemic fibrosis (NSF) or nephrogenic fi-
brosing dermopathy (NFD) in patients with renal insuf-
ficiency (9,10). Because a significant number of pa-
tients with renal disease also suffer from coronary
artery disease (11,12), a safe diagnostic method that
1Department of Radiology, Northwestern University, Chicago, Illinois.
2Department of Biomedical Engineering, Northwestern University, Chi-
Contract grant sponsor: the National Institutes of Health Research;
Contract grant number: HL57484; Contract grant sponsor: American
Heart Association; Contract grant number: SDG 0735099N.
*Address reprint requests to: R.D., Department of Radiology, North-
western University, 737 N Michigan Avenue, Suite 1616, Chicago, IL
USA, 60611. E-mail: email@example.com
Received October 31, 2007; Accepted January 29, 2008.
Published online in Wiley InterScience (www.interscience.wiley.
JOURNAL OF MAGNETIC RESONANCE IMAGING 27:1037–1045 (2008)
© 2008 Wiley-Liss, Inc.
can be more widely used for assessing the functional
status of CAD is necessary.
A more direct method for evaluating the functional
status of the myocardium may be possible with blood-
oxygen-level-dependent (BOLD) imaging. BOLD imag-
ing detects changes in oxygen saturation of the hemo-
globin molecule (13). In particular, the differential
changes in oxygen saturation results in alterations in
magnetic susceptibility of the hemoglobin molecule.
The oxygen saturation of the hemoglobin molecules is
indirectly related to the bulk magnetic susceptibility
shift between the red blood cells (RBCs) and plasma
(14). The changes in bulk magnetic susceptibility, in
addition to altering the magnetic field in the intravas-
cular space (inside vessels, i.e., blood), also affects the
extravascular (outside vessels) space (15,16). These
magnetic field perturbations influence the phase evolu-
tion of water protons in the tissue and lead to MR signal
changes that are a direct reflection of blood oxygen
The potential benefits of BOLD MRI for detecting
global or regional myocardial oxygenation changes due
to CAD were demonstrated at least a decade ago
(17,18). Several studies have demonstrated the feasibil-
ity of using the BOLD effects to assess myocardial blood
oxygenation secondary to flow changes in both animals
and humans (19–25). Existing BOLD methods primar-
ily rely on deriving oxygen-sensitive contrast with phar-
macological stress agents by inducing coronary artery
vasodilation that increase baseline coronary venous
%O2from 20–30% to 70–80% in healthy tissue (26).
However, in the absence of collateralization, %O2in the
myocardial territories supplied by the stenotic arteries
remains relatively unaltered between basal and vasodi-
latory states. This permits the identification of myocar-
dial regions with reduced perfusion reserve as hypoin-
tense regions in myocardial BOLD MR images.
The most commonly used myocardial BOLD imaging
technique relies on detecting changes in %O2by means
of T2* methods (17–19,21,22). While T2* studies have
demonstrated the feasibility of detecting myocardial ox-
ygenation changes, large magnetic susceptibility arti-
facts, particularly from the heart–lung interface (27),
have significantly compromised the image quality. More
robust T2-prepared methods have helped to improve
image quality through minimizing the influence of sus-
ceptibility artifacts from the lungs (23,24,28,29). How-
ever, long data acquisition times, cardiac and respira-
acquisition have been significant obstacles. More re-
cently, an oxygen-sensitive balanced SSFP imaging
method has been proposed to overcome many of the
limitations (30,31). The capacity of SSFP imaging to
generate cardiac phase-resolved regional myocardial
BOLD signal changes under pharmacological stress
has also been demonstrated (25).
In many organ systems, the advantage of using high-
field MRI for enhancing the sensitivity of BOLD imaging
is well established (32,33). However, similar studies
demonstrating the advantages of high-field imaging for
the assessment of myocardial oxygenation have not yet
been performed. This work examines the benefits and
the feasibility of extending the SSFP-based myocardial
BOLD imaging methods at 1.5T to 3.0T. First, using
simulations, this work examines the dependence of
strength at 1.5T and 3T. This theoretical study is fol-
lowed by controlled experimental studies using a ca-
nine model with the goal of validating the theoretical
findings and demonstrating the possibility of advancing
the SSFP-based myocardial BOLD imaging from 1.5T to
3.0T. Experimental results will be used to evaluate
whether, any potential increase in oxygen sensitivity at
3.0T will permit the detection of clinically useful twofold
regional differences in the vasodilated flow between
myocardial territories supplied by stenotic and healthy
coronary vessels (34).
MATERIALS AND METHODS
A two-pool exchange model described by Dharmaku-
mar (31) was adopted to evaluate the myocardial signal
response under various conditions at 1.5T and 3.0T.
The model accounts for relaxation constants, spin-res-
idence times, blood hematocrit, field strength, relative
blood volume, changes in intravascular oxygenation
due to pharmacological stress resulting in variations in
transverse relaxation constants in blood and frequency
variation between the intravascular and extravascular
pools. The two-pool model allows one to simplify the
tissue into intravascular and extravascular compart-
ments because the spin exchange within each compart-
ment is fast. As modeled, the intravascular compart-
ment accounts for red blood cells, plasma, and other
cell types present in blood, while the extravascular
compartment accounts for the interstitium and the in-
tracellular space outside the vasculature.
The time evolution of the orthogonal components of
the magnetization vector from the two different com-
partments was computed by solving the set of coupled
ordinary differential equations representing the two-
pool exchange model subject to repeated radiofre-
quency (RF) pulses (31). The RF pulses were assumed to
be instantaneous and were applied in a phase-cycled
pattern with TR? 5.2 ms, flip angle ? 60°. Two hundred
RF pulses were applied to transition the initial magne-
tization into steady state. The steady-state transverse
magnetizations from the different compartments were
added together as a vector sum. The magnitude of this
vector sum represented the intensity of the SSFP signal.
Effects of intravascular oxygenation changes were
modeled as changes in T2of blood and frequency shifts
between the intra- and extravascular compartments.
The dependence of T2of blood (T2b) on intravascular
oxygen saturation (%O2) was modeled as:
1/T2b? 1/T20? K?1 ? %O2/100?2, 
where T20is the T2of fully oxygenated blood and K is a
parameter that is determined by scan parameters and
physical properties of blood (30). The frequency shifts
between the intravascular and extravascular pools (??)
as a function of %O2was computed from:
1038Dharmakumar et al.
?? ? ??0?1 ? %O2/100?, 
where ??0is the peak frequency shift (200 rad/s at 1.5T
and 400 rad/s at 3.0T) between the intravascular and
extravascular pools when the blood is fully deoxygen-
ated (31). The relaxation constants, relative myocardial
blood volume under baseline conditions, T20, and K
values used to solve the model are collected in Table 1.
The K values corresponding to TR? 5.2 ms at 1.5T and
3.0T were obtained by means of a linear interpolation
between previously reported (30) K values at TR? 4 ms
and 8 ms at a flip angle of 60°.
Simulations were performed to compute the myocar-
dial signal intensities at 1.5T and 3.0T in regions sup-
plied by healthy and stenotic arteries under conditions
mimicking the effects of pharmacological stress agents
such as adenosine or dipyridamole. Consistent with
previous reports, simulations were performed assum-
ing the myocardial %O2of 80% in healthy myocardium
and 30% in affected myocardium (26). Calculations also
assumed that under vasodilation, the relative myocar-
dial blood volume increases by 50% in healthy zones
(26), but no change takes place in regions supplied by
the stenotic artery. From the resulting signal values,
theoretical oxygen contrast (OCT) was computed as:
OCT? 100% ? ?Sh? Ss?/Sh, 
where Shand Ssare the corresponding simulated signal
magnitudes obtained from healthy and stenotic myo-
cardial territories, respectively, under the influence of
vasodilatory agents. The theoretical field strength de-
pendence between 1.5T and 3.0T was computed as:
Field Dependence ?FDT? ? OCT?3.0T?/OCT?1.5T?
Surgical Preparation of Animals
Seven, 20–25 kg, mongrel dogs were operated on and
studied by use of procedures and protocols approved by
our Institutional Animal Care and Use Committee.
Dogs were given a preanesthetic tranquilizer (Innovar, 1
ml IM) and then anesthetized using Propofol (5.0–7.5
mg/kg IV). The animals were intubated, ventilated, and
placed on gas anesthesia (2.0–2.5% isoflurane and
100% oxygen). Under a sterile technique, a left lateral
thoracotomy was performed between the eighth and
ninth ribs to expose the heart. The pericardium was
opened and sewed into a cradle, and the left circumflex
coronary artery (LCX) was identified. An external hy-
draulic occluder was placed around the LCX 1.0–1.5
cm from the bifurcation of the left main coronary artery
for inducing reversible stenoses within the LCX. A
Doppler ultrasound flow probe (Crystal Biotech, North-
borough, MA) was placed circumferentially around the
LCX 2–3 cm downstream from the occluder to measure
LCX blood velocity. Catheters were surgically implanted
into the left and right atria and were used for the deliv-
ery of normal saline (0.9%), adenosine, and fluorescent
microspheres during MR experiments. A catheter was
also implanted in the aorta for direct blood flow mea-
surements to be used with microsphere analysis. At the
end of the surgical procedure, the chest was closed and
the animals were allowed to recover for 7 days before
MRI scans were performed.
Animal Preparation for Imaging Studies
Before imaging, dogs were examined for reactive hyper-
emia response to determine the effectiveness of adeno-
sine in vasodilating the coronary tree and to rule out
any pre-existing stenosis of the LCX. Relative changes
in Doppler flow velocity in the LCX artery between rest
and stress conditions were used to assess the extent of
reactive hyperemia. An increase in blood flow velocity of
at least 50% from resting conditions was considered
acceptable. In two animals, before the second MR
study, reactive hyperemia was below acceptable levels
and a second MR study was not completed. Each dog
was sedated, intubated, and then placed on a ventilator
with gas anesthesia (Isoflurane 2.0–2.5% and 100%
oxygen) for the remainder of the study. Systemic O2
saturation, end-tidal PCO2, and heart rate were moni-
tored continuously throughout the imaging study. A
power injector was connected to the right atrial catheter
for delivery of adenosine.
The feasibility of SSFP-based myocardial BOLD imag-
ing at 1.5T using a TR of 6.3 ms has been recently
demonstrated in canines (25). However, the image qual-
ity issues with SSFP imaging, particularly at 3.0T and
high TRare well known (35). To address this issue, a set
of baseline studies were performed in noninstrumented
and instrumented animals (n ? 4) to assess the highest
TRthat can be reliably achieved without being impaired
by banding artifacts in the mid-ventricular regions of
the myocardium. Animals were sedated, ventilated, and
were positioned in feet-first right-anterior oblique posi-
tion. A flexible, phased-array surface coil was placed
over the chest used for signal reception. After this,
scout images were obtained to localize the heart. When
Physical Parameters Used in the Simulations Used
T2Eis the T2of the extravascular space. T1Eand T1Iare the T1of extravascular and intravascular spaces, respectively. RBV is the relative
blood volume fraction in myocardial tissue. Values were obtained from the following references:a(30);b(43);c(31); andd(16).
SSFP-Based Myocardial BOLD Imaging at 3.0T1039
shimmed and center frequency was scouted, myocar-
dial images free of banding artifacts were consistently
obtained at TE/TR? 2.6/5.2 ms and flip angle of 60°.
These scan parameters were subsequently used in all
imaging studies aimed at assessing myocardial oxygen-
ation changes in the presence of coronary artery steno-
sis. With this set of parameters, at least 25 min of
prescanning (shimming and center frequency adjust-
ments) was required to acquire a mid-ventricular short
axis image of the myocardium, while significantly less
time (under 5 min) was required to achieve the same
task at 1.5T.
Assessment of Myocardial Oxygenation and
All SSFP BOLD images were acquired in a short-axis
slice located at the level of the mid-left ventricle, clearly
delineating anterior and posterior papillary muscles,
within one breath hold (12–18 s). All imaging studies
were performed using a Siemens Sonata 1.5T and Tim
Trio 3.0T scanners (Erlangen, Germany). At each of the
field strengths, two to three SSFP BOLD images (one
under vasodilation and one to three stenosis levels un-
der vasodilation) were acquired in the cine mode. Effort
was always made to study at least two stenosis levels,
however, in some (n ? 2) animals, increasing the ste-
nosis extent led to premature ventricular complexes
(PVCs), which in one case resulted in fatal ventricular
arrhythmia. In other animals, the heart rate remained
approximately constant during all scans under vasodi-
lation. Vasodilation of the coronary arterial tree was
achieved by means of constant adenosine infusion
(0.14 mg/min) into the right atrial catheter. During the
imaging studies, flow velocity changes within the LCX
were assessed with Doppler measurements. The LCX
artery lumen diameter was controlled using the hy-
draulic occluder. Mild and severe stenoses were defined
as those necessary to reduce LCX flow velocity by
roughly 50% and 80%. Polystyrene fluorescent micro-
spheres (3 ? 106, Molecular Probes, Eugene, OR) were
injected through the left atrial catheter to obtain the
true regional myocardial perfusion under various ex-
perimental conditions. Each experimental condition
was assessed with a unique color of microsphere. Mi-
crospheres were injected at prestenosis with adenosine
and at the different stenosis levels to evaluate perfusion
changes. Between each scan, the occluder was deflated
and the flow was allowed to recover. If PVCs were ob-
served in the presence of stenosis, animals were al-
lowed to recover, and when possible, a reduced stenosis
level was used in subsequent studies.
Typically, two MRI studies were performed in each
animal; one at 1.5T and another at 3.0T. The order of
the studies was randomized. To match the imaging
slices between the two studies, myocardial landmarks
from the scout scans from the initial studies were used.
A total of 11 MR studies (5 at 1.5T and 6 at 3.0T) were
performed in 7 dogs with at least 72 h of rest between
studies. The scan parameters were: field of view ? 130
mm ? 260 mm, segments (lines) per cardiac phase ?
5–7, 10–12 phases/heart beat, imaging matrix ? 144 ?
192, slice thickness ? 6 mm, TE/TR? 2.6/5.2 ms, flip
angle ? 60°, readout bandwidth ? 345 Hz/pixel, RF
pulses were phase cycled, number of averages ? 2, and
one whole heart beat of dummy RF pulses were applied
before data acquisition. Approximately, this translated
to one full cine acquisition taking 20–30 s (40 to 60
heartbeats) with a temporal resolution of 40–50 ms per
First-pass Perfusion and Myocardial Viability
Each study was terminated with a first-pass contrast-
enhanced MRI perfusion exam  at the severe stenosis
state to provide visual comparisons to the BOLD im-
ages. Gd-DTPA (0.075 mmol/kg, Magnevist, Berlex
Laboratories, Inc., Montville, NJ) was injected intrave-
nously followed by a 10-cc saline flush. During the
infusion, a 30-s breath-held saturation-recovery turbo
FLASH sequence (36) was prescribed over the mid-ven-
tricular imaging slice used for BOLD imaging studies.
The scan parameters at 1.5T and 3.0T were as follows:
imaging matrix ? 144 ? 192, flip angle ? 10°, slice
thickness ? 8 mm, TE/TR? 1.2/2.5 ms, and readout
bandwidth ? 1000 Hz/pixel. At the end of the study,
delayed-enhancement images were acquired to ascer-
tain whether infarction had occurred during the inflic-
tion of LCX stenosis. An additional dose of Magnevist
was delivered, followed by a 10-cc saline flush, so that
the full dose of Magnevist, before the viability scans was
0.15 mmol/kg. Viability scans were performed over the
mid-ventricular slice of interest and two adjacent slices
with an inversion-recovery FLASH sequence (37) with
the following scan parameters: TE/TR? 1.11 / 360 ms;
TI? 210 ms; flip angle ? 25°; slice thickness ? 6 mm;
and the imaging matrix and field-of-view were the same
as those used for the first-pass perfusion scans.
Animal Euthanasia, Tissue Sectioning, and
After the final MRI study, animals were euthanized by
means of an intravenous delivery of pentobarbital so-
dium at a dose of 0.2 cc/kg. To independently charac-
terize the myocardial flow changes throughout the myo-
microsphere analysis was performed in a manner sim-
ilar to previous studies (24,25). The left ventricle was
dissected free from the thoracic cavity and cut into four
circumferential (short axis) rings. To match the slices to
mid-ventricular short axis BOLD images, the most api-
cal and basal slices were discarded. The remaining two
rings, including the papillary muscles, were each sec-
tioned into eight equal circumferential sectors for stan-
dard microsphere analysis (38), from which the micro-
circulatory perfusion values were computed. Perfusion
values in the respective sectors from the slices were
averaged. Perfusion changes in the LCX territory, rela-
tive to the left anterior descending coronary artery
(LAD) territory, in each animal and field strength were
Relative Microsphere-Based Regional Flow Difference
? ? ?QLCX
1040Dharmakumar et al.
mean flow in the LAD and LCX territories, respectively,
and the superscripts represent the stenosis state of the
LCX; prestenosis (“p”) and stenosis (“s”).
MR Image Analysis
SSFP BOLD images were exported from the MRI console
(Leonardo, Siemens Medical Solutions, Erlangen, Ger-
many) in the Digital Imaging and Communications in
Medicine format (DICOM) and loaded into Image J (ver-
sion 1.36b, National Institutes of Health, Bethesda,
MA) installed on a 1.7 GHz personal computer (Pentium
4, Intel, USA). Inner and outer contours were manually
drawn around the end systolic, short axis, myocardial
images and were sectioned into eight segments corre-
sponding to microsphere measurements. End systolic
images were chosen from the cine image set to have a
large myocardial surface area for accurate data analy-
sis. Signal intensities were measured from each myo-
cardial image segment corresponding to the LCX and
LAD territories. The mean signal intensities from each
of these regions were then used to compute the BOLD
signal changes between the LCX and LAD territories as:
Relative Regional BOLD Contrast (?Sr)
?(?Sr) ? ??SLCX
? ? ?SLCX
where SLADand SLCXrepresent the mean BOLD signal
intensities from the LAD and LCX territories, respec-
tively, and the superscripts represent the stenosis state
of the LCX; prestenosis (“p”) and stenosis (“s”). Using
?Qrand ?Sr, experimental oxygen contrast (OCE), at
each stenosis level and field strength, was computed as:
The experimental Field Dependence (FDE) between 1.5T
and 3.0T was computed in a similar manner using Eq.
, with the exception of OCTreplaced with OCE. In
addition to evaluating BOLD contrast changes as a
function of field strength, it is useful to compute con-
strengths as well. To be able to compare contrast-to-
noise changes at the two different field strengths with-
out bias in flow-related changes between experiments,
flow difference (?Qr) normalized CNR (CNR), was com-
CNR ? ?SLCX
?/??n? ?Qr?, 
at 1.5T and 3.0T, where SLCX, ?Qr, and superscripts are
as defined above, and ?nis the standard deviation of the
image noise (air).
A linear mixed effect regression model was used to as-
sess the relationship between microsphere-based per-
fusion (?Qr) and BOLD MRI data (?Sr). Power analysis
was performed on the data to assess the power in pre-
dicting the observed correlation coefficient for a P ?
0.05. One-sample, one-tailed, Student’s t-test was used
on OCEto test whether flow-normalized BOLD contrast
changes at 3.0T was greater than that at 1.5T. All sta-
tistical analyses were performed on Microcal origin 7.0
(Northampton, MA) and a P value of less than 0.05 was
used to assess the statistical significance of results.
Theoretical oxygen contrast values, OCT, corresponding
to 1.5T and 3.0T are collected in Figure 1. Results show
that there is a significant increase in OCTas the field
strength is doubled from 1.5T to 3.0T. In particular, the
theoretical Field Dependence (FDT) ? 3.0.
Every viability scan, performed before terminating the
imaging studies, was negative. This permitted us to
analyze all 23 stress-stenosis studies (10 at 1.5T and 13
at 3.0T) that we collected from the seven animals.
Figure 2 shows myocardial images obtained at 1.5T
(unprimed labels) and 3.0T (primed labels). Figure
shows a set of typical short axis SSFP-based cardiac
BOLD images obtained under adenosine infusion dur-
ing prestenosis (A, A?) and severe stenosis (B, B?) con-
ditions. Note that the prestenosis BOLD images show
signal homogeneity throughout the myocardium. How-
ever, under severe stenosis conditions, appreciable sig-
nal deficits are evident in the LCX territories (smaller
arc subtended by arrows), while, no visible changes in
signal intensities were observed in other parts of the
myocardium. Figure 2 also shows a set of first-pass
perfusion images (C, C?) and the spatial maps of Micro-
sphere-Based Regional Flow Difference between prest-
enosis and severe stenosis states (D, D?). Note the close
correspondence between the BOLD images, first-pass
perfusion images, and microsphere-based flow differ-
Figure 3 shows scatter plots with line of best fit be-
tween Relative Microsphere-Based Regional Flow Dif-
ference (?Qr) and Relative Regional BOLD Contrast
Figure 1. Theoretically expected dependence of OCT(Eq. )
on external, static magnetic field strength. As the field strength
is doubled from 1.5T to 3.0T, OCTincreases by a factor of 3.
SSFP-Based Myocardial BOLD Imaging at 3.0T1041
(?Sr) at 1.5T and 3.0T. Linear regression analysis
yielded a correlation coefficient (r ? 0.7) between mi-
crosphere data and MR data with P ? 0.01 at both field
strengths with the following linear equations of fit: y ?
0.052 x 0.017 (at 1.5T) and y ? 0.332 x 0.128 (at 3.0T),
where y and x represent ?Sr and ?Qr, respectively.
Power analysis revealed that we can be 90% certain that
the observed correlation coefficient is 0.7 with P ? 0.01.
Upper and lower 95% prediction bands were also com-
puted and it was found that the future events will be
within approximately ? 0.034 and ? 0.120 of the mean
?Srat 1.5T and 3.0T, respectively.
Figure 4 shows a plot with mean OCEvalues that were
measured at 1.5T and 3.0T. Results show that there is
a statistically significant difference between OCE at
1.5T and 3.0T. In particular, the FDEwas 2.5, indicat-
ing that, relative to 1.5T, at 3.0T substantial increase in
myocardial oxygen sensitivity is possible.
Figure 5 shows a plot of flow-normalized average CNR
values that were measured at 1.5T and 3.0T. Similar to
OCE, statistically significant increases in CNR values
were also observed at 3.0T in relation to 1.5T. Approx-
imately a fourfold increase in flow-normalized CNR is
evident at 3.0T compared with that at 1.5T.
SSFP imaging can identify changes in blood oxygen
saturation levels. Driven by this observation (30), sev-
eral studies have shown that SSFP-based oxygen-sen-
sitive imaging can be used for non–contrast-enhanced
angiography, assessment of brain activation, and in the
evaluations of systemic hypoxia. More recently, SSFP-
based cine BOLD imaging has also been used for eval-
uating myocardial oxygenation changes in canines with
coronary stenosis at 1.5T. The extension of this method
to 3.0T is highly desirable given the sensitivity of SSFP
BOLD imaging is expected to be strongly dependent on
magnetic field strength. Furthermore, while the depen-
dence of field strength on BOLD sensitivity has been
studied in a number of different tissues, to date, there
are no similar studies of myocardial BOLD imaging.
The current work explored the benefits and the pos-
sibility of extending SSFP-based myocardial BOLD im-
aging to 3.0T with numerical simulations and con-
Numerical studies showed that at 3.0T, compared with
1.5T, a threefold increase in myocardial BOLD sensitiv-
ity is possible. Image quality assessments on healthy
and instrumented animals with respect to scan param-
eters (TRand flip angle) provided an optimum set of
scan parameters for acquiring images free of banding
artifacts at 1.5T and 3.0T. Experimental studies evalu-
ating myocardial oxygenation in the presence of selec-
tive coronary artery stenosis and pharmacological va-
sodilation showed that, at 3.0T, compared with 1.5T,
Figure 2. Typical short axis MR images obtained at 1.5T (top row, A–C) and at 3T (bottom row A?–C?). Images A and A? are
steady-state free-precession (SSFP) images obtained without stenosis, images B and B? are SSFP images at systole under severe
stenosis of the left circumflex coronary artery (LCX), and images C and C? are the corresponding first-pass images acquired
under stenosis of similar extent as in B and B?. Images D and D? represent the spatial map (scale provided by the gray-scale bar)
of Microsphere-Based Regional Flow Difference between prestenosis and severe stenosis during 1.5T and 3.0T studies, respec-
tively. The arrows subtend the suspected regions (LCX territory) where the perfusion deficits are expected to develop as a result
of LCX stenosis in dogs. Note the discriminating signal loss in these regions in images B and B? and the close correspondence
between the first-pass perfusion (C,C?) and microsphere-based flow difference maps (D,D?).
Figure 3. Scatter plot showing the correlation between Re-
Contrast (?Sr, Eq. ) and Regional Microsphere-based Flow
Difference (?Qr, Eq. ) at 1.5T (A) and 3.0T (B). There is a
strong correlation between microsphere and BOLD MRI data;
r ? 0.7 at both 1.5T and 3.0T (P ? 0.01). Fit equations: y ?
0.052 x ? 0.017 (at 1.5T) and y ? 0.332 x ?0.128 (at 3.0T),
where y and x represent ?Srand ?Qr, respectively. Note that
the slope of the line of best fit has a significantly higher slope
at 3.0T than at 1.5T, confirming the theoretical predictions
that the BOLD sensitivity at 3.0T is higher than at 1.5T. The
plot also shows the upper and lower 95% prediction bands for
predicting future observations.
1042Dharmakumar et al.
SSFP-based myocardial BOLD contrast increases by a
factor of 2.5. The relative increase in SSFP-based myo-
cardial BOLD contrast between 1.5T and 3.0T was con-
firmed independent of the flow variations between the
studies based on microsphere flow analysis. At both
field strengths, statistically significant correlation be-
tween microsphere-based perfusion (?Qr) and BOLD
contrast (?Sr) was also observed. The extent of this
correlation was similar to previous studies (24,25,28).
Theory Versus Experiment
Although there is a strong agreement between numer-
ical simulations and experimental data with respect to
the dependency of myocardial oxygen sensitivity and
field strength, the data are not in complete agreement.
In particular, the theory predicts that when the field
strength is doubled from 1.5T to 3.0T, FDTis 3.0, while
experimental results show a reduced FDEof 2.5 ? 0.2.
There may be several factors that have contributed to
this modest difference. First, in our simulations, we
assumed that blood hematocrit is 0.4. However, after
surgery, blood hematocrit can vary between 0.3 and 0.4
in animals. Second, B1 inhomogeneities at 3.0T are
more severe than at 1.5T (35). This finding suggests
that, at 3.0T, it is likely that our intended flip angle of
60° may have been reduced. Because SSFP BOLD con-
trast is directly related to flip angle (30,39), this may
have also reduced the OCEat 3.0T. Third, an assump-
tion implicit in the FDEcomputation is that the micro-
and SSFP signals are linearly related at 1.5T and 3.0T.
However, the SSFP signal intensities with respect to
blood oxygen saturation, at least in whole blood, is
nonlinear with the extent of the nonlinearity directly
related to field strength and TR(30). In addition, note
that the OCTvalues were computed assuming complete
stenosis, whereas the OCEare computed from a range
of stenosis levels. If the nonlinearity between oxygen-
ation and SSFP signal intensity in whole blood is ex-
tended to microvasculature, small changes in oxygen-
ation between healthy and affected regions may reduce
OCErelative to OCT, particularly at 3.0T. Reduced oxy-
genation differences between myocardial territories
supplied by stenotic and nonstenotic vessels may occur
in the presence of inadvertent baseline stenosis or re-
duced vasodilatory response of the microvasculature
due to the well known rapid breakdown of adenosine in
canines. This potential reduction in OCEat 3.0T may
provide a partial explanation for the reduced FDEcom-
pared with FDT. Fourth, although every effort was made
to match the imaging slices between 1.5T and 3.0T,
there may have been slight deviations. Finally, an in-
complete knowledge of relaxation constants and other
simulations parameters, such as myocardial blood vol-
ume, and animal-to-animal variations may also ac-
count for the differences between theoretical and exper-
To optimize the scan parameters against imaging arti-
facts at 3.0T, based on baseline scans, we chose TR?
5.2 ms and flip angle ? 60°. These parameters were also
used for the 1.5T studies to directly compare the field
strength effects. However, previous SSFP-based myo-
cardial BOLD imaging studies at 1.5T used an opti-
mized TRand flip angle of 6.3 ms and 90°, respectively.
Given that both TRand flip angle were lower in our
studies, the oxygen sensitivity may have been compro-
mised at 1.5T. To account for the difference in BOLD
sensitivity between the two different sets of parameters,
we computed the OCTbetween the two different condi-
tions. Our results showed that the reduced TRand flip
angle in the current 1.5T studies may have reduced the
OCTby approximately 20%. Given that the increase in
BOLD sensitivity at 3.0T relative to 1.5T (with lower TR
and flip angles) is 2.5-fold, even if the sensitivity com-
parisons were to be made using the higher TRand flip
angle at 1.5T, the sensitivity at 3.0T is expected to be far
larger than at 1.5T.
One of the assumptions of our theoretical model is
that the oxygen sensitivity may be adequately assessed
based on intravascular T2changes and exchange ef-
fects. In reality, however, perivascular gradients set up
by the susceptibility shift between the intravascular
Figure 4. Experimentally observed dependence of myocardial
OCE(Eq. ) on external, static magnetic field strength. There
is a statistically greater OCEavailable at 3.0T compared with
1.5T (P ? 0.01). Specifically, when the field strength is doubled
from 1.5T to 3.0T, OCEincreases by a factor of 2.5.
Figure 5. Experimentally observed dependence of flow nor-
malized CNR (CNR, Eq. ) on external, static magnetic field
strength. There is significantly greater CNR available at 3.0T
compared with 1.5T ((P ? 0.01) for evaluating myocardial ox-
SSFP-Based Myocardial BOLD Imaging at 3.0T1043
and extravascular effects may introduce greater T2-
weighting through diffusion-mediated signal losses.
This explanation is particularly relevant in the heart
because most of the myocardial blood volume, in excess
of 90%, is found in the capillaries (15,16,40–42). Be-
cause diffusion mediated signal changes in T2-weighted
imaging is strongly influenced by the capillaries, and
the expected spin-echo behavior for SSFP acquisitions
with TE? TR/2, diffusion effects may be significant.
However, given the close correlation between experi-
ment and theory, we anticipate that the contribution to
the BOLD contrast from the perivascular gradients to
be moderate to small within the TRvalues used in this
study. Further studies are necessary to evaluate the
role of diffusion effects on SSFP-based myocardial
In interpreting the myocardial BOLD sensitivity at 1.5T
and 3.0T, it is important to evaluate whether the in-
crease in the BOLD sensitivity may be useful in provid-
ing a means for detecting clinically significant flow dif-
ferences between healthy and affected regions. To
assess this, it is necessary to take into account not only
the sensitivities, but also the variations (scatter) in the
relation between ?Srand ?Qrat 1.5T and 3.0T. The
measures of sensitivity and corresponding variations
can be obtained as the slopes of the lines of best fit and
variation from the lines of best fit to the 95% prediction
limits of the scatter plots (Fig. 3), respectively. When
these values are taken together, it is possible to show
that one can detect 2.9- and 1.6-fold regional flow dif-
ferences between healthy and affected regions of the
myocardium at 1.5T and 3.0T, respectively. Provided
the scatter remains the same, for the optimized param-
eter set of TRof 6.3 ms and flip angle of 90° (25), based
on simulations suggesting an increase in sensitivity of
20%, one may be able to detect a 2.2-fold regional flow
difference. These findings strengthen the case for using
3.0T over 1.5T for SSFP-based myocardial BOLD imag-
ing studies because 3.0T has the potential to permit the
detection of clinically useful minimum regional flow
difference of twofold; a demand that may not be met at
Although our studies have demonstrated that signif-
icant increase in myocardial oxygen sensitivity may be
obtained at 3.0T, our experience with 3.0T SSFP myo-
cardial BOLD imaging suggests that imaging artifacts
are more difficult to overcome at 3.0T than at 1.5T. This
finding is not surprising, because the effects of B0in-
homogeneities on SSFP imaging, particularly for car-
diac applications at 3.0T, is well known (35). It has been
previously shown that the B0field variations at 1.5T in
the human thoracic cavity, due to susceptibility shifts
induced by the air in the lungs, can result in peak
center frequency shifts of 70–100 Hz (27). At 3.0T, we
expect that these shifts to approximately double. This
suggests that, without local shimming, banding arti-
facts (dark bands) may be present in the images even at
TRof 2.5 ms with RF phase-cycling. To overcome much
of this artifact at 3.0T and to be able to generate myo-
cardial images free of banding artifacts, several steps
were undertaken in this study. The first of these is the
use of an optimized flip angle (60°) that minimizes the
influence of off-resonance frequency shifts on SSFP sig-
nal intensity as demonstrated by Dharmakumar. Sec-
ond, we repeatedly performed local shimming and cen-
ter frequency scouting to overcome the banding
artifacts. In comparison with 1.5T, at 3.0T, significantly
more time, in excess of 25 min, was spent on shimming
and frequency scouting. At TRof 5.2 ms, we were able to
consistently obtain mid-ventricular short axis SSFP im-
ages free of banding artifacts. However, it is unclear
whether these artifact reduction strategies will success-
fully translate for clinical studies. We anticipate that,
for clinical studies, more efficient shimming methods
may be necessary to take advantage of the increased
BOLD sensitivity available at 3.0T.
In conclusion, guided by previous findings of SSFP-
based oxygen sensitivity on field strength, we examined
the dependence of field strength on myocardial BOLD
sensitivity using theoretical and experimental models.
Both theory and experiments demonstrated that, for a
fixed set of TR and flip angle, increasing the field
strength from 1.5T to 3.0T led to a nearly threefold
increase in SSFP-based oxygen sensitivity. These re-
sults show that the use of SSFP-based myocardial im-
aging can provide enhanced sensitivity for detecting
myocardial oxygenation abnormalities originating from
the presence of coronary artery stenosis in canines. In
particular, our results showed that SSFP-based 3.0T
BOLD imaging has the sensitivity to detect clinically
useful flow differences over 1.5T. We anticipate that
clinical translation of SSFP-based myocardial BOLD
imaging at 3.0T may require more time-efficient and
robust methods to overcome magnetic field inhomoge-
1. Rosamond W, Flegal K, Friday G, et al. Heart disease and stroke
statistics—2007 update: a report from the American Heart Associ-
ation Statistics Committee and Stroke Statistics Subcommittee.
2. Pagley PR, Beller GA, Watson DD, Gimple LW, Ragosta M. Improved
outcome after coronary bypass surgery in patients with ischemic
cardiomyopathy and residual myocardial viability. Circulation
3. Mickleborough LL, Maruyama H, Takagi Y, Mohamed S, Sun Z,
Ebisuzaki L. Results of revascularization in patients with severe left
ventricular dysfunction. Circulation 1995;92(Suppl):II73–II79.
4. Vanoverschelde JL, Gerber B, Pasquet A, Melin JA. Nuclear and
echocardiographic imaging for prediction of reversible left ventric-
ular ischemic dysfunction after coronary revascularization: current
status and future directions. J Cardiovasc Pharmacol 1996;
5. Clark AN, Beller GA. The present role of nuclear cardiology in
clinical practice. Q J Nucl Med Mol Imaging 2005;49:43–58.
6. Klocke FJ. Measurements of coronary flow reserve: defining patho-
physiology versus making decisions about patient care. Circulation
7. Wilke N, Simm C, Zhang J, et al. Contrast-enhanced first pass
myocardial perfusion imaging: correlation between myocardial
blood flow in dogs at rest and during hyperemia. Magn Reson Med
8. Arai AE. Magnetic resonance first-pass myocardial perfusion imag-
ing. Top Magn Reson Imaging 2000;11:383–398.
9. Marckmann P, Skov L, Rossen K, et al. Nephrogenic systemic fibro-
sis: suspected causative role of gadodiamide used for contrast-
enhanced magnetic resonance imaging. J Am Soc Nephrol 2006;
1044Dharmakumar et al.
10. Kuo PH, Kanal E, Abu-Alfa AK, Cowper SE. Gadolinium-based MR Download full-text
contrast agents and nephrogenic systemic fibrosis. Radiology
11. Brown JH, Hunt LP, Vites NP, Short CD, Gokal R, Mallick NP.
Comparative mortality from cardiovascular disease in patients with
chronic renal failure. Nephrol Dial Transplant 1994;9:1136–1142.
12. Gradaus F, Ivens K, Peters AJ, et al. Angiographic progression of
coronary artery disease in patients with end-stage renal disease.
Nephrol Dial Transplant 2001;16:1198–1202.
13. Thulborn KR, Waterton JC, Matthews PM, Radda GK. Oxygenation
dependence of the transverse relaxation time of water protons in
whole blood at high field. Biochim Biophys Acta 1982;714:265–
14. Wright GA, Hu BS, Macovski A. 1991 I.I. Rabi Award. Estimating
oxygen saturation of blood in vivo with MR imaging at 1.5 T. J Magn
Reson Imaging 1991;1:275–283.
15. Boxerman JL, Hamberg LM, Rosen BR, Weisskoff RM. MR contrast
due to intravascular magnetic susceptibility perturbations. Magn
Reson Med 1995;34:555–566.
16. Bauer WR, Nadler W, Bock M, et al. The relationship between the
BOLD-induced T2and T2*: a theoretical approach for the vascula-
ture of myocardium. Magn Reson Med 1999;42:1004–1010.
17. Atalay MK, Forder JR, Chacko VP, Kawamoto S, Zerhouni EA.
Oxygenation in the rabbit myocardium: assessment with suscep-
tibility-dependent MR imaging. Radiology 1993;189:759–764.
18. Li D, Dhawale P, Rubin PJ, Haacke EM, Gropler RJ. Myocardial
signal response to dipyridamole and dobutamine: demonstration of
the BOLD effect using a double-echo gradient-echo sequence.
Magn Reson Med 1996;36:16–20.
19. Beache GM, Herzka DA, Boxerman JL, et al. Attenuated myocardial
vasodilator response in patients with hypertensive hypertrophy
revealed by oxygenation-dependent magnetic resonance imaging.
20. Foltz WD, Huang H, Fort S, Wright GA. Vasodilator response as-
sessment in porcine myocardium with magnetic resonance relax-
ometry. Circulation 2002;106:2714–2719.
21. Wacker CM, Hartlep AW, Pfleger S, Schad LR, Ertl G, Bauer WR.
Susceptibility-sensitive magnetic resonance imaging detects hu-
man myocardium supplied by a stenotic coronary artery without a
contrast agent. J Am Coll Cardiol 2003;41:834–840.
22. Friedrich MG, Niendorf T, Schulz-Menger J, Gross CM, Dietz R. Blood
oxygen level-dependent magnetic resonance imaging in patients with
stress-induced angina. Circulation 2003;108:2219–2223.
23. Fieno DS, Shea SM, Li Y, Harris KR, Finn JP, Li D. Myocardial
perfusion imaging based on the blood oxygen level-dependent effect
using T2-prepared steady-state free-precession magnetic reso-
nance imaging. Circulation 2004;110:1284–1290.
24. Shea SM, Fieno DS, Schirf BE, et al. T2-prepared steady-state free
precession blood oxygen level-dependent MR imaging of myocardial
perfusion in a dog stenosis model. Radiology 2005;236:503–509.
25. Dharmakumar R, Arumana JM, Larson AC, Chung Y, Wright GA, Li
D. Cardiac phase-resolved blood oxygen-sensitive steady-state free
precession MRI for evaluating the functional significance of coro-
nary artery stenosis. Invest Radiol 2007;42:180–188.
26. Li D, Oellerich WF, Gropler RJ. Magnetic resonance assessment of
myocardial oxygenation. In: Manning W, Pennell DJ, editors. Car-
diovascular magnetic resonance. Philadelphia, PA: Churchill Liv-
ingstone; 2001. p 447–454.
27. Reeder SB, Faranesh AZ, Boxerman JL, McVeigh ER. In vivo mea-
surement of T2* and field inhomogeneity maps in the human heart
at 1.5 T. Magn Reson Med 1998;39:988–998.
28. Wright KB, Klocke FJ, Deshpande VS, et al. Assessment of regional
differences in myocardial blood flow using T2-weighted 3D BOLD
imaging. Magn Reson Med 2001;46:573–578.
29. Foltz WD, Al-Kwifi O, Sussman MS, Stainsby JA, Wright GA. Op-
timized spiral imaging for measurement of myocardial T2relax-
ation. Magn Reson Med 2003;49:1089–1097.
30. Dharmakumar R, Hong J, Brittain JH, Plewes DB, Wright GA.
Oxygen-sensitive contrast in blood for steady-state free precession
imaging. Magn Reson Med 2005;53:574–583.
31. Dharmakumar R, Qi X, Hong J, Wright GA. Detecting microcircu-
latory changes in blood oxygen state with steady-state free preces-
sion imaging. Magn Reson Med 2006;55:1372–1380.
32. Yang Y, Wen H, Mattay VS, Balaban RS, Frank JA, Duyn JH.
Comparison of 3D BOLD functional MRI with spiral acquisition at
1.5 and 4.0 T. Neuroimage 1999;9:446–451.
33. Duong TQ, Yacoub E, Adriany G, Hu X, Ugurbil K, Kim SG. Micro-
vascular BOLD contribution at 4 and 7 T in the human brain:
gradient-echo and spin-echo fMRI with suppression of blood ef-
fects. Magn Reson Med 2003;49:1019–1027.
34. Klocke FJ, Simonetti OP, Judd RM, et al. Limits of detection of
regional differences in vasodilated flow in viable myocardium by
first-pass magnetic resonance perfusion imaging. Circulation
35. Schar M, Kozerke S, Fischer SE, Boesiger P. Cardiac SSFP imaging
at 3 Tesla. Magn Reson Med 2004;51:799–806.
36. Lauerma K, Virtanen KS, Sipila LM, Hekali P, Aronen HJ. Multislice
MRI in assessment of myocardial perfusion in patients with single-
vessel proximal left anterior descending coronary artery disease be-
fore and after revascularization. Circulation 1997;96:2859–2867.
37. Kim RJ, Fieno DS, Parrish TB, et al. Relationship of MRI delayed
contrast enhancement to irreversible injury, infarct age, and con-
tractile function. Circulation 1999;100:1992–2002.
38. Glenny RW, Bernard S, Brinkley M. Validation of fluorescent-la-
beled microspheres for measurement of regional organ perfusion.
J Appl Physiol 1993;74:2585–2597.
39. Mangalathu Arumana J, Li D, Dharmakumar R. The influence of
imaging parameters on SSFP-based microcirculatory Blood-Oxy-
gen-Level Dependent (BOLD) contrast: theory and experiment.
ISMRM 2007; Berlin, Germany. p 3790.
40. Bauer WR, Nadler W, Bock M, et al. Theory of the BOLD effect in the
capillary region: an analytical approach for the determination of T2
in the capillary network of myocardium. Magn Reson Med 1999;
41. Kennan RP, Zhong J, Gore JC. Intravascular susceptibility con-
trast mechanisms in tissues. Magn Reson Med 1994;31:9–21.
42. Kassab GS, Fung YC. Topology and dimensions of pig coronary
capillary network. Am J Physiol 1994;267(Pt 2):H319–325.
43. Donahue KM, Weisskoff RM, Chesler DA, et al. Improving MR
quantification of regional blood volume with intravascular T1con-
trast agents: accuracy, precision, and water exchange. Magn Reson
SSFP-Based Myocardial BOLD Imaging at 3.0T1045