In recent years, system advances, the routine use of intravascular gadolinium and the development of navigator echoes for magnetic resonance angiography (MRA) have significantly improved image quality, permitting exquisite detailing of the thoracic vasculature. This article describes basic MRA techniques and the recent advancements that permit diagnostic quality thoracic MRA.
Magnetic resonance angiography (MRA) is an established technique
for the evaluation of the cerebral vasculature and is rapidly
gaining acceptance as an alternative to peripheral angiography.
Historically, however, MRA of the thorax has been limited by the
requisite long acquisition times and multiple motion vectors
resulting in severe image degradation. In recent years, advanced
magnetic resonance (MR) systems that are capable of sub-second
image acquisitions, the routine use of intravascular gadolinium
and, in the case of coronary artery imaging, the development of
"navigator echoes" have significantly improved image quality,
permitting exquisite detailing of the thoracic vasculature.1-3 This
article describes the basic MRA techniques and the recent
technologic advancements that permit diagnostic quality thoracic
Basic MRA techniques
The basic concept of MRA is to suppress the signal emanating
from stationary tissues in the body and enhance the signal from
flowing blood. There are two techniques commonly used clinically:
phase-contrast and time-of-flight MRA. Each of these techniques can
be acquired as a series of continuous two-dimensional (2D) slices
or a single three-dimensional (3D) volume. The advantages and
disadvantages of each technique relative to thoracic MRA is
Phase-contrast (PC) MRA utilizes two opposing flow-encoding
gradients to produce a phase shift in moving protons. The phase
shift describes a vector with both velocity and directional
information. From the diagram in figure 1 we can see that flow
information can only be evaluated in protons moving along the axis
of the flow-encoding gradients. To describe flow in all directions,
at least four pairs of flow-encoding gradients must be applied
across the tissues.l Because of the need for additional gradients,
PC-MRA is quite time consuming. In addition, this technique is
exquisitely sensitive to motion, a significant disadvantage in the
dynamic environment of the chest. Cardiac and respiratory gating
can be utilized to reduce the associated motion artifacts from
physiologic thoracic activity, but this is at the expense of
additional acquisition times. Although impressive images
quantitating flow through the coronary arteries have been
demonstrated,4 the requisite long acquisition times limit the
routine use of PC MRA in the chest.
Time-of-flight (TOF) MRA
The object of TOF-MRA is to suppress the T1 signal from
background tissues and enhance the T1 signal emanating from flowing
blood in each image. This task is accomplished through saturation
of the stationary background spins.
Saturation-When hydrogen protons in a large magnetic field are
exposed to a resonant radiofrequency, the individual protons jump
to a higher energy state against the main magnetic field. The
result is a gradual loss of longitudinal magnetization (figures
2A-C). When the radiofrequency is switched off, the protons
gradually return to the lower energy state, the longitudinal
magnetization recovers, and the excess energy is given off in the
form of heat (figure 2D). This is known as T1 relaxation.
The T1 relaxation is a finite interval that differs for various
compounds and tissues; it is the basis for contrast on a
T1-weighted image. Some tissues recover very fast (short T1) and
others take a little longer (long T1). The percent recovery just
before the next radio pulse will determine the amount of T1 signal
for each compound. By selecting a short repetition time (TR), we
can maximize the contrast between tissues (figures 3A, B). However,
if we fire a series of rapid radio pulses, one right after another,
none of the tissues will have time to recover and no T1 signal will
In TOF-MRA, the T1 signal from a slice or slab of tissue is
saturated with a series of rapid radio pulses (figure 3C). As such,
no appreciable signal emanates from the stationary protons in that
slice. However, blood flowing into this area does not experience
the saturation pulses and enters the slab fully relaxed, with a
maximal longitudinal magnetization. After the next radio pulse, the
background tissues remain saturated while the fresh spins in the
flowing blood produce a high T1 signal. This is the basis for
vascular contrast in TOF-MRA.
One of the limitations with TOF-MRA is the saturation effects
the blood experiences as it flows through the imaged tissue volume.
This is more problematic with thicker tissue volumes and slow blood
flow, areas where the blood experiences multiple saturation pulses
for an extended period of time, resulting in severe signal drop-off
at the distal end of the volume.
2D vs 3D acquisition
One way to limit the saturation effects of TOF MRA is to acquire
a series of thin, individual slices through the anatomic area of
interest and stack the individual sections to produce a volumetric
image of the vasculature. This is referred to as a 2D acquisition.
Because the individual slices are thin, the blood doesn't
experience multiple saturation pulses as it traverses the imaged
tissue slice. However, disadvantages of 2D acquisitions include
inter-slice motion which results in severe post-processing image
degradation, and the relative poor quality multiplanar reformations
due to the finite thickness of each slice.1,2 This is especially
problematic in the chest, where the intricate architecture of the
thoracic vasculature may warrant the use of complex reformations to
delineate the tortuous course of the coronary and pulmonary
In a 2D acquisition, the thickness of each slice is limited by
the strength of the gradient oriented along the slice selection
axis. The steeper the gradient, the thinner the tissue section
excited for a given radiofrequency (rf) pulse. In a 3D acquisition,
an entire volume of tissue is initially excited with a single rf
pulse. The individual slices in that volume are then determined by
multiple phase encoding steps applied along the slice selection
direction. Because this process is relatively independent of the
gradient strength, very thin sections (<1 mm) can be acquired.
This technique allows for detailed postprocessing reconstructions
of the complex vascular anatomy present in the chest. The
trade-off, of course, is the severe signal drop-off at the distal
end of the tissue slab, resulting from the multiple saturation
pulses the blood experiences as it traverses the imaged volume.
However, the recent introduction of a number of new imaging
techniques, combined with the latest advanced scanner hardware, has
effectively eliminated the in-plane saturation effects and motion
artifacts, making 3D-TOF the optimal technique for imaging the
thoracic vasculature.6 The remainder of this article describes
these advances and their effect on image quality.
A scanner's speed is directly related to the maximum amplitude
and the rise time of the gradient subsystems. The latest generation
of MR units are equipped with high powered gradient coils that
permit fast scanning techniques which can markedly reduce
acquisition times and permit single breath-hold evaluations of the
chest. At our institution, a 3D-TOF volumetric image of the
thoracic vasculature can be obtained in 19 seconds. In addition,
the improved frequency and phase resolution between adjacent spins
afforded by the powerful gradients in conjunction with a shorter
rise time permits a faster readout following a radio pulse and,
thus, significantly reduces the minimum echo time (TE). The effect
is an improved signal-to-noise ratio resulting from decreased
intravoxel dephasing of spins (figure 4).
In general, gadolinium increases the T1 signal of the
vasculature by markedly decreasing the T1 relaxation time of the
blood. In one study, administration of an IV dose of 0.3 mmol/kg
Gd-DTPA was found to reduce the T1 relaxation of blood from 1200 ms
to 300 ms.7 In other words, it doesn't take as long for the
longitudinal magnetization to recover following an excitation.
With regards to our 3D-TOF acquisitions, the short repetition
time of the saturation pulse suppresses the T1 signal from all the
tissues in the imaged slice. However, the gadolinium in the
vasculature markedly shortens T1 relaxation and allows the protons
within the blood to at least partially recover their longitudinal
magnetization before the next radio pulse. Thus, a higher T1 signal
is maintained within the vasculature relative to the stationary
background tissues despite the series of saturation pulses. In
essence, vascular enhancement is no longer flow dependent.8,9
At our institution, all MRA images of the thoracic vasculature
are obtained with gadolinium-DTPA contrast. A total of 0.2 mmol/kg
body weight is administered intravenously at 2 cc/second. Imaging
begins immediately with a series of 3 to 5, 19-second 3D-TOF
breath-hold sequences and a 3-second breath between each series.
Depending on the desired vascular phase, optimal contrast
enhancement usually is achieved by the third or fourth series
Respiratory and cardiac gating
The standard 3D, contrast-enhanced MRA images of the thoracic
aorta and pulmonary vasculature are acquired over multiple cardiac
cycles. Therefore, the final images are actually composites of the
systolic and diastolic phases of the heart. Alternatively, cardiac
activity can be electronically monitored and the interval between
contractions (R-R interval) subdivided into predetermined "bins."
During image acquisition, the acquired data is placed in the bin
that corresponds to the current position along the R-R interval.
Therefore, each bin contains the image data for a single phase of
the cardiac cycle. This is the definition of cardiac gating. When
the images from each bin are played in a cine loop, a dynamic
evaluation of heart wall motion and vascular pulsations can be
appreciated. A similar technique can be used to gate respiratory
Evaluation of the coronary arteries warrants special
consideration. The relatively small diameter of these vessels
compared to the excursion distance associated with a normal cardiac
contraction requires accurate cardiac gating to "freeze" wall
motion during data acquisition. Because image data can only be
collected during a finite portion of the R-R interval, a 3D volume
can not be completed during a single breath-hold. As such,
respiratory gating must be used in conjunction with cardiac gating
to accurately image the proximal coronary arteries.3,5,10,11
In the past, volumetric images of the proximal coronary arteries
could be obtained by segmenting the acquisition into a series of
suspended respirations.12 A mechanical device was used to measure
the excursion of the chest wall and would then provide feedback to
the patient through audible or visual ques. These ques would help
the patient reproduce the degree of inspiration after each
breath.13-16 This, of course, would require full patient
cooperation. A novel technique for gating respiratory motion
without the need of external mechanical devices or significant
patient cooperation is the use of "navigator echoes" to monitor the
superior-inferior (SI) position of the hemidiaphragm.3
Kinematic studies of heart motion during respiration have
demonstrated that cardiac displacement is mainly in an SI
direction. In addition, there is essentially a linear relationship
between the diastolic position of the heart and the SI position of
the hemidiaphragm.17 The navigator echo is simply a thin,
longitudinally oriented volume that intermittently monitors the SI
position of the hemidiaphragm and determines the interval of
acceptable data acquisition. Thus, patient cooperation is no longer
essential for effective coronary imaging (figure 6).3
The advantages of thoracic MRA over standard angiographic
techniques in the chest include its ability to acquire volumetric
3D images of the vessels noninvasively without ionizing radiation
or iodinated intravenous contrast. A review of current and
preliminary clinical applications relative to specific sections of
the thoracic vasculature is provided below.
Pulmonary embolism-Pulmonary embolism is one of the major causes
of mortality in the hospitalized patient.18 The standard diagnostic
techniques, V/Q scan and angiography, are both time consuming and
invasive. Recent advances in CT angiography show promise in the
evaluation of central and paracentral pulmonary emboli.19-21
However, iodinated intravenous contrast materials are still
necessary with this technique.
For a number of technical reasons, MRA of the pulmonary arteries
historically has been extremely limited. As mentioned above,
uncompensated respiratory and cardiac motion significantly degrade
vascular detail. Overlapping venous and arterial vessels produce
complex images which may be very difficult to interpret.6 Also,
depending on the age of a thrombus, its appearance on a standard
spin-echo sequence can range from low to high signal
intensity.22-24 Contrast-enhanced thoracic 3D MRA has the potential
to overcome at least some of these limitations.
The homogeneous intravascular signal intensity provided by the
IV gadolinium allows thrombus to appear as a true filling defect,
regardless of blood flow.6 Advanced high speed MR systems permit
fast 3D acquisitions that are capable of high resolution imaging of
the entire pulmonary vasculature within a single breath-hold (<
20 seconds) (figure 7). In the severely tachypneic patient,
acquisition times can be further reduced by using 1/2 k-space
sampling in the phase-encoding direction, or by reducing the number
of partitions in the 3D volume, though this is at the expense of
spatial resolution.6 Initial reports describe a high diagnostic
accuracy in the evaluation of central pulmonary embolus utilizing
The overlapping arterial and venous vasculature remains a
problem. However, the 3D acquisition permits reformations in
additional planes to help delineate the complex anatomic
architecture and localize filling defects within the arterial
Pulmonary arterial hypertension- MRA has been demonstrated to
effectively and noninvasively diagnose pulmonary artery
hypertension.6,26 With gadolinium-enhanced MRA, we have the ability
to visualize the pulmonary arteries to the sub-segmental level,
allowing clear demonstration of the abrupt caliber change from the
central to the peripheral vasculature. In addition, recent reports
have shown that thoracic MRA may be useful in distinguishing
primary pulmonary hypertension from secondary hypertension due to
chronic thromboembolic disease.26
Dissection and aneurysm-Following the diagnosis of aortic
dissection, evaluation of the extent of disease becomes imperative
in order to determine the direction and urgency of therapeutic
intervention. The efficacy of MR imaging in the assessment and
follow-up of aortic dissection is well established.27,29 However,
conventional MRI suffers technical limitations related to the
variable blood flow present within the false lumen that may
preclude differentiation of clot from sluggish flow;
differentiation of thrombus from turbulent flow in large aortic
aneurysms may be difficult with standard MR techniques for the same
reasons.4,6 In addition, requisite long acquisition times may be
problematic in the hemodynamically unstable patient.
Contrast-enhanced MRA, with or without breath-hold, has been shown
to overcome these limitations.6 Because vascular enhancement is no
longer flow dependent with contrast-enhanced MRA techniques,
differentiation of thrombus from sluggish flow is readily apparent
(figure 5). The 3D acquisition permits detailed reconstructions in
additional planes to accurately characterize the extent of
dissection relative to branch vessels (figure 8). Finally, fast
scanning techniques permit an expedient diagnostic evaluation of
the hemodynamically unstable patient.
Congenital abnormalities-3D MR angiography of the pulmonary
vessels can accurately characterize anatomical anomalies such as a
right-sided or double aortic arch, as well as aberrant branch
vessels. Contrast-enhanced MR angiography can be used in
conjunction with cardiac gating to produce cinegraphic images of
blood flow and provide both physiologic and anatomic information in
disease entities such as aortic coarctation (figure 9) or valvular
stenosis. Intravenous gadolinium enhancement eliminates the
flow-related artifacts caused by turbulent conditions distal to the
stenosis, thus permitting accurate localization and quantification
of the degree of narrowing.6
Coronary arteries-To date, MR imaging of the coronary arteries
has been very limited, mostly due to motion artifacts caused by
cardiac and respiratory activity. These can be at least partially
compensated for with gating techniques such as the navigator echo
sequence described above.3 Because the arteries are close to the
myocardial surface and often embedded in epicardial fat,
magnetization transfer and fat suppression techniques have been
utilized to suppress the signal from the myocardial and fatty
tissues, respectively, and improve conspicuity of the proximal
vessels.11 Nonetheless, accurate characterization of coronary
artery disease with MRA is not currently possible. Recent reports
describe the experimental use of ultrafast MR imaging with a
first-pass bolus gadolinium-enhancement technique to measure and
quantify myocardial perfusion in a single heart beat, thus
providing indirect evidence of coronary artery disease.30 However,
at the time of this writing, gadolinium-enhanced MR coronary
angiography remains largely experimental.
High speed imaging techniques in combination with the routine
use of intravenous gadolinium is currently capable of producing
diagnostic quality, three-dimensional magnetic resonance
angiographic images of the thoracic vasculature in a single,
comfortable breath-hold. Cardiac gating can be utilized to segment
image acquisitions and produce high quality cinegraphic images of
cardiac activity. In short, 3D TOF-MRA with intravenous
administration of gadolinium-DTPA contrast has proven efficacy in
the diagnosis of thoracic vasculature pathology. With further
development and improvement of intravenous contrast agents and the
routine use of sub-second imaging techniques, thoracic MRA will
likely see an expanded role in the diagnosis of thoracic vascular
1. Finn JP, Goldman A, Edelman RR: Magnetic resonance
angiography in the body. Magn Reson Q 8(1):1-22, 1992.
2. Grave MJ: Magnetic resonance angiography. Br J Radiol
3. Wang Y, Rossman PJ, Grimm RC, et al: Navigator-echo-based
real-time respiratory gating and triggering for reduction of
respiratory effects in three-dimensional coronary MR angiography.
Radiology 198:55-60, 1996.
4. Krinsky G, Weinreb J: Gadolinium-enhanced three-dimensional
MR angiography of the thoracoabdominal aorta. Semin Ultrasound CT
MR 17(4):280-303, 1996.
5. Hofman MBM, Paschal CB, Li D, et al: MRI of coronary
arteries: 2D breath-hold vs 3D respiratory-gated acquisition. J
Comput Assist Tomogr 19(1):56-62, 1995.
6. Leung DA, Debatin JF: Three-dimensional contrast-enhanced
magnetic resonance angiography of the thoracic vasculature. Eur
Radiol 7:981-987, 1997.
7. Tu R, Kennel T, Turski P, et al: Preliminary assessment of
gadodiamide-enhanced, complex-difference phase-contrast magnetic
resonance angiography. Acad Radiol 1:S47-S55, 1994.
8. Prince MR, Yucel EK, Kayfman JA, et al: Dynamic
gadolinium-enhanced three-dimensional abdominal MR arteriography. J
Magn Reson Imaging 3:877-881, 1993.
9. Prince MR: Gadolinium-enhanced MR aortography. Radiology
10. Edelman RK, Manning WJ, Burstein D, Paulin S: Coronary
arteries: Breath-hold MR angiography. Radiology 181:641-643,
11. Li D, Paschal CB, Haacke EM, Adler LP: Coronary arteries:
Three-dimensional MR imaging with fat saturation and magnetization
contrast. Radiology 187:401-406, 1993.
12. Liu Y, Riederer SJ, Ehman RL: Magnetization-prepped cardiac
imaging using gradient
echo acquisition. Magn Reson Med 30:271-275, 1993.
13. Liu Y, Riederer SJ, Rossman PJ, et al: A monitoring,
feedback, and triggering system for reproducible breath-hold MR
imaging. Magn Reson Med 30:507-511, 1993.
14. Wang Y, Christy PS, Korosec FR, et al: Coronary MRI with a
respiratory feedback monitor: The 2D imaging case. Magn Reson Med
15. Wang Y, Grimm RC, Rossman PJ, et al: 3D coronary MR
angiography in multiple breath-holds using a respiratory feedback
monitor. Magn Reson Med 34:11-16, 1995.
16. Williams AH, Riederer SJ, Grimm RC, et al: Multiple
breath-hold 3D time-of-flight MR angiography of the renal arteries
using real-time respiratory feedback. Proceedings from the 3rd
Annual Meeting of the Society of Magnetic Resonance, p 1568. Nice,
17. Wang Y, Riederer SJ, Ehman RL: Respiratory motion of the
heart: Kinematics and the implications for the spatial resolution
of the coronary MR imaging. Magn Reson Med 33:713-719 1995.
18. Dalen JE, Alpert JS: Natural history of pulmonary embolism.
Prog Cardiovasc Dis 17:259-270, 1975.
19. van Rossum AB, Pattynama PM, Ton ER, et al: Pulmonary
embolism: Validation of spiral CT angiography in 149 patients.
20. van Erkel AR, van Rossum AB, Bloem JL, et al: Spiral CT
angiography for suspected pulmonary embolism: A cost-effectiveness
analysis. Radiology 201(1):29-36, 1996.
21. Cauvain O, Remy-Jardin M, Remy J, et al: Spiral CT
angiography in the diagnosis of central pulmonary embolism:
Comparison with pulmonary angiography and scintigraphy. Rev Mal
Respir 13(2):141-153, 1996.
22. Thickmann D, Kressel HY, Axel L: Demonstration of pulmonary
embolus by magnetic resonance imaging. AJR 142:921-922, 1984.
23. Pope CF, Sostman D, Carbo P, et al: The detection of
pulmonary emboli by magnetic
resonance imaging. Invest Radiol 22:937-947, 1986.
24. Gamsu G, Hirji M, Moore EH, et al: Experimental pulmonary
emboli detected by magnetic resonance imaging. Radiology
25. Debatin JF, Leung DA, Steiner P, et al: Contrast-enhanced 3D
MR angiography of the pulmonary arteries in under 20 seconds. p
161. In: Book of abstracts, p 161. New York, Society for Magnetic
Resonance in Medicine, 1996.
26. Bergin CJ, Hauschildt J, Rios G, et al: Accuracy of MR
angiography compared with radionuclide scanning in identifying the
cause of pulmonary arterial hypertension. AJR 168(6): 1549-1555,
27. Kersting-Sommerhoff BA, Higgins CB, White RD, et al: Aortic
dissection: Sensitivity and specificity of MR imaging. Radiology
28. Nienaber CA, von Kodolitsch Y, Nicolas V, et al: The
diagnosis of thoracic aortic dissection by noninvasive imaging
procedures. N Engl J Med 328:1-9, 1993.
29. Gaubert J, Moulin G, Mesana T, et al: Type A dissection of
the thoracic aorta: Use of MR imaging for long term follow up.
30. Crnac J, Schmidt MC, Theissen P, Sechtem Y: Assessment of
myocardial perfusion by magnetic resonance imaging. Herz
Dr. Klioze and Dr. Mergo are with the Department of Radiology at
the University of Florida College of Medicine in Gainesville,