is Chief Resident in Diagnostic Radiology at New Jersey Medical
School, Newark, NJ. He completed medical education from Seth G.S.
Medical College/King Edward Memorial Hospital, Mumbai, India and
preliminary medicine internship at the University of Tennessee at
Memphis in June 2000. He will start a Fellowship in
Neuroradiology at University of Pennsylvania, Philadelphia, PA,
in July 2004.
Traditionally, time of flight (TOF) and phase contrast (PC)
magnetic resonance angiography (MRA) have been used to study
arteries and veins. In TOF MRA, a gradient pulse is repeatedly
applied to a slice/slab (two-dimensional [2D]/three-dimensional
[3D]) of tissue. The signal from the stationary tissue decreases
and becomes more saturated with each pulse. The moving blood brings
unsaturated protons into the tissue slice, which generates a high
signal intensity. The contrast in TOF imaging is thus provided by
suppressing the stationary tissue signal intensity in conjunction
with the high signal intensity from moving blood.
In PC MRA, the applied gradient imparts a phase shift to the
inflowing protons proportional to their velocity. Two data sets are
acquired with opposite sensitization resulting in opposite phase
for moving protons and indentical phase for the stationary protons.
On subtraction of the data sets, signal from stationary protons is
eliminated and just the moving protons are seen. Information about
the quantity and direction of flow is obtained.
In TOF MRA, if the vessel is oriented horizontally within the
selected slice, the gradients that suppress the stationary tissue
also suppress the signal from moving blood. This leads to a
decrease in signal due to saturation. In PC MRA, turbulent flow
leads to intravoxel dephasing of protons resulting in a decrease in
signal that is also responsible for overestimation of a stenosis.
Appropriate velocity encoding is vital for correct interpretation.
These techniques have largely been replaced by contrast-enhanced
MRA (CE-MRA), which uses a 3D spoiled gradient-echo sequence with
intravenously administered gadolinium (Gd). Gadolinium reduces the
T1 relaxation of blood significantly, so as to produce a very high
signal from the blood compared with that from surrounding tissue.
The timing of image acquisition is of paramount importance to image
the artery without significant venous signal contamination. Various
commercial methods like SmartPrep (GE Medical systems, Waukesha,
WI), CareBolus (Siemens Medical Systems, Iselin, NJ), and BolusTrak
(Phillips Medical Systems, Bothell, WA) are available to
automatically determine the optimal imaging time. Administering a
test bolus followed by MR fluoroscopic imaging to observe the time
of maximum arterial concentration is often very helpful.
K-space contains the spatial frequencies of the signal obtained
from a tissue slice. These spatial frequencies are converted to an
image by inverse Fourier transformation. The center of the k-space
contains the contrast information. Hence, the center of the k-space
should be acquired at the peak arterial concentration of Gd. Two
such alternate k-space schemes are centric phase ordering and
elliptical centric-phase ordering.
For each, data can be processed to view maximum intensity
projection (MIP) images, multiplanar reformations (MPR), or 3D
surface projection images.
In single-detector computed tomography (SDCT), a fan-shaped beam
traverses the patient and is detected by a single row of detectors.
The information at the edges of the beam is lost, resulting in poor
resolution along the z-axis. The pitch is equal to table feed per
rotation divided by slice thickness. Increasing the scan speed or
the area of coverage necessitates increase in pitch. This results
in degradation of the z-axis resolution because noncontiguous
images are obtained.
Multiple rows of detectors were added in the Z direction to solve
this problem. The detectors at the periphery are now able to detect
data at the edge of the fan beam, thereby improving spatial
resolution along the z-axis. In this method of imaging, a contrast
bolus is administered through a peripheral vein, and scanning is
commenced during the arterial phase. As with CE-MRA, commercial
products that optimize the timing of the scan are available. MIP
images, MPR images, and volume-rendered images, are used for
analysis. With 16-channel multidetector CT (MDCT), volumetric data
are obtained with near-isotropic resolution
(resolution equal in all planes), which is very important to
visualize small vessels, and produce excellent reformations. The
capability for near-isotropic imaging is responsible for the
superior spatial resolution of CT over MR imaging. MDCT has faster
acquisitions, covers a larger area, and significantly reduces the
amount of injected iodinated contrast necessary as compared with
The single breath- hold acquisition eliminates artifacts due to
respiratory excursion and body movement, giving a superior image
Clinical applications: Computed tomography angiography
and magnetic resonance angiography
Intracranial and extracranial carotid arteries
The prime indications for imaging the intracranial vasculature
are for the detection and treatment planning of aneurysms and the
evaluation of specific clinical situations related to
steno-occlusive disease. Autopsy studies indicate that aneurysms
<5 mm in diameter are unlikely to rupture.
This dimension is usually the cutoff for treating asymptomatic
CT angiography (CTA) and MRA combined have the potential to replace
catheter angiography for diagnosis (Figures 1 and 2). A
thin-section (1.25 mm/2.5 mm), contrast-enhanced MDCTA is performed
with 3D volu-metric and MIP reformations. Traditionally, 2D/3D TOF
had been the MR imaging technique. Currently, 3D TOF CE-MRA and
time-resolved CE-MRA are preferred. 3D TOF CE-MRA has improved
spatial resolution, less saturation effects and intravoxel
dephasing, and better evaluation of aneurysm lumen. With this
technique, imaging is not significantly affected by the presence of
hemorrhage. The high signal from the stationary tissue due to T1
contamination artifact of blood is eliminated by subtraction of the
pre-enhancement volume from the contrast-enhanced time frame.
MDCTA and MRA reliably detect aneurysms >=3 mm in diameter.
Several reports confirm their sensitivity and specificity as being
similar to catheter angiography.
MDCTA better defines the anatomic configuration and surroundings of
the aneurysm and has the advantage of showing the relationship of
the aneurysm to bone, which is necessary for surgical planning.
The entire scan can be completed in 30 seconds, minimizing motion
artifacts. Aneurysms in the cavernous sinus and skull base can be
difficult to detect by MDCTA. MRA is better suited for their
The postprocessing of CTA mandates the time and attention of a
skilled operator. In addition to general advantages, CE-MRA on a 3T
magnet with fast gradient techniques provides an especially
accurate depiction of cerebral aneurysms.
Three-dimensional TOF (noncontrast at 1T) MRA has a sensitivity
and specificity of 92% and 91%, respectively, for detecting
intracranial stenosis exceeding 50% of the width of the vessel
CTA has a sensitivity and specificity of 100% and 99%,
respectively, for the same degree of stenosis. MRA has a tendency
to overestimate the extent of stenosis; however, calcification in
the arterial wall can eventuate in a nondiagnostic study by CTA. A
limitation of MRA is that it cannot be performed in claustrophobic
patients or in patients with pacemakers, certain implants, or
aneurysm clips. Sixteen-channel MDCT has the potential to reduce
artifacts due to surgical clips and calcification (this requires
Atherosclerosis is the most frequent indication to image the
extracranial carotid arteries. Endarterectomy has been shown to be
beneficial in patients with a stenosis of greater than 70%.
Investigators from the Asymptomatic Carotid Artery Study
have suggested that asymptomatic patients with a stenosis of 60%
can benefit from endarterectomy by reducing stroke risk. MDCTA and
are accurate and reliable methods of detecting carotid stenosis
(Figure 3). Recently, Linera et al
reported that the sensitivity and specificity of CE-MRA in
detecting carotid stenosis of greater than 70% to be 97.1% and
95.2%, respectively. In the same study the sensitivity and
specificity of CTA were quoted to be 74.3% and 97.6%, respectively.
The results of CTA in this study may be suboptimal, as a
single-detector CT was used to perform the CTA. Sensitivity and
specificity for MDCTA are typically close to 100%.
Besides discerning the degree and length of stenosis, plaque
irregularity, ulceration, and characterization (in terms of lipid,
thrombus and calcification) can be done with both techniques.
MDCTA has been shown to be slightly superior to MRA for detecting
plaque irregularities, whereas plaque ulceration, which is a
frequent precursor of emboli, is better seen with MRA. It has been
suggested that plaque characterization by MRA may be used in
treatment planning of atherosclerotic disease
(Figure 4). One must be aware of pitfalls of each technique. Vessel
wall calcification limits the accuracy of MDCTA. Overestimation of
stenosis can occur with CE-MRA.
In the year 2000, 1.32 million in-patient diagnostic coronary
angiographic examinations were performed in the United States.
The number of diagnostic examinations exceeded those performed as a
part of coronary angioplasty or stent placement.
There is a need and a great potential to image the coronary
arteries noninvasively so as to avoid the risks associated with
catheter angiograms and to minimize exposure to ionizing
MDCTA of the coronary arteries is performed after intravenous
injection of 80 to 120 mL of contrast medium at the rate of 3 to 5
mL/sec followed by a saline bolus chase. Spiral scanning using 8 or
16 slices is performed with a digital electrocardiography (EKG) and
oversampling of scan projections.
This allows retrospective EKG gating, creating images in the same
phase of the cardiac cycle.
Currently, 3D segmented k-space gradient-echo/echoplanar MRA is
performed with the navigator respiratory compensation technique. A
30-mm slab with overlapping 20 slices of 3-mm thickness is
obtained. The process takes only 10 to 12 minutes per 3D
Applications: Coronary vascular anomalies--
Both CE-MRA and MDCTA have been shown to be equivalent to catheter
angiography for the detection of coronary vascular anomalies. Due
to enhanced temporal resolution of CE-MRA, it has the ability to
visualize the anatomic path taken by the contrast (Personal
communication, Martin Prince, MD, PhD, Weill Medical College of
Cornell University, New York, New York, July 2003). CE-MRA could be
used primarily to detect vascular anomalies and to evaluate cases
deemed equivocal by other modalities (Figure 5).
Applications: Detection of bypass graft stenosis--
Engelmann et al
have reported contrast-enhanced 3D gradient- echo technique to have
a sensitivity of 92% and a specificity of 85% for the detection of
graft stenosis. The use of 2D spin-echo and gradient-echo
techniques allows the visualization of flow in 2 successive slices
to correlate with patency, while flow in 1 slice is equivocal, and
no flow indicates occlusion.
The sensitivity and specificity of MDCTA for graft stenosis are 97%
and 89%, respectively, slightly higher than those reported for the
current CE-MRA technique.
Applications: Detection of native vessel stenosis--
In a recent multicenter study, Kim et al
have reported a sensitivity of 88% to 100%, and a specificity of
44% to 88% for the detection of significant (>50%) native
coronary stenosis. Several recent studies have determined the
sensitivity and specificity of MDCTA to range from 85% to 89% and
76% to 99%, respectively.
MDCTA can now reveal near-isotropic resolution of at least 0.6 mm,
allowing visualization of the distal small coronary branches
(Figure 6). A calcium score can be obtained. A low calcium score
has a high negative predictive value for coronary artery disease
Retrospective EKG gating also allows functional information such as
left ventricular ejection fraction and anatomic detail such as left
ventricular wall thickness to be obtained from the same
acquisition; however, this is best evaluated by MR imaging. With
the latest improvement in spatial resolution coupled with better
techniques to suppress motion and the potentially confounding
influence of epicardial fat, MRA can serve as a "one-stop"
examination in the detection of CAD.
Coronary artery disease is a major health problem causing
significant morbidity in the expanding elderly population. A test
to detect CAD should be noninvasive, reliable, safe (does not
involve ionizing radiation), and inexpensive. Coronary MRA has some
of these virtues. A meta-analysis was performed to investigate the
efficacy of MRA compared with catheter angiography in the detection
of significant coronary artery stenosis.
A Medline search was performed with the key words "coronary MR
angiography." English language studies from 1993 to 2003 comparing
current MRA technique (3D gradient or echo- planar imaging) with
catheter angiography were selected. After excluding 2D MR
techniques, 20 studies were identified. Strict inclusion criteria
based on recommendations of Oxman et al
were applied (Table 1).
Nine studies (Table 2), met the criteria, and 11 studies were
excluded (Table 3). For each study, 2 * 2 tables (true positive,
true negative * false positive, false negative) were constructed.
Based on the number of coronary segments evaluated, overall
sensitivity and specificity of coronary MRA for the detection of
significant stenosis were calculated. The meta-analysis was
limited, as only two variables (sensitivity and specificity) were
evaluated. Due to the small number of studies that met the
inclusion criteria, other variables, such as the effect of sample
size on sensitivity and specificity, could not be evaluated.
The study by Kim et al
had the maximum number of coronary segments evaluated, more than
all other studies combined. The total visible segments varied from
62 to 636, the sensitivity ranged from 44% to 93%, and the
specificity ranged from 42% to 95% (Figure 7). There was a high
level of interobserver variability for the determination of
sensitivity of coronary MRA for significant stenosis. The
specificity was near constant. The overall sensitivity and
specificity of coronary MRA in the detection
of significant stenosis were 83% and 82%, respectively.
MRA holds promise to be a safe, noninvasive, and an efficient
test in evaluating the coronary arteries. Technical advancements,
such as faster imaging sequences, and prospective double-blinded
studies are needed to evaluate the full extent of its clinical
The most frequent indication to image the pulmonary arteries is
pulmonary embolism (PE). Conventional angiography is not possible
in up to 20% of patients due to its invasiveness, associated
sporadic mortality, and morbidity.
CTA and MRA are faster, less invasive, and associated with a
smaller number of complications.
MDCTA and MRA have similar efficacy in diagnosis of main, lobar,
and segmental pulmonary artery emboli (Figures 8 and 9). Oudkerk et
report a sensitivity of 84% for segmental, and 100% for main and
lobar arteries when using CE-MRA with a specificity of 98%, when
compared with catheter angiography. Studies using helical CT have
quoted sensitivity and specificity of 90% for diagnosis of
The advantage of MDCTA is fast imaging and wide availability. The
disadvantage is that in a patient population with a high risk of
PE, it is not performed in up to 12% due to renal failure.
In 1% to 10% of patients, CTA is suboptimal for radiologic
The drawback of CE-MRA is long imaging time. For good-quality
images, breath-holding for at least 10 to 15 seconds for each lung
acquisition is necessary,
which may not be possible in patients with dyspnea. Haage et al
report a detection rate of emboli of 97.7% by using real-time MRA
without breath-holding in animal studies. Initial reports of
protocols to image the pulmonary arteries in <4 seconds
have also been described and are very promising, but lack
prospective confirmation. The advantage of CE-MRA is simultaneous
evaluation of lung ventilation and perfusion in the same
In patients with PE and negative compression duplex ultrasonography
(CDUS), MR venography shows thrombus in pelvic veins in up to
one-third of the patients.
This is done during the same examination, and avoids exposure to
radiation and iodinated contrast as compared with methods for CT
venography. Recent experimental studies show CE-MRA to be superior
to CTA in diagnosis of subsegmental pulmonary emboli
; however, the management of patients with isolated subsegmental PE
is controversial. At present, MDCTA is widely employed; however,
CE-MRA has the same efficiency and more advantages than MDCTA.
Imaging of the aorta
Acute aortic dissection is a cardiovascular emergency with a
preadmission mortality of 21%.
The role of imaging is to recognize the dissection, reveal the
entry point, display branch vessel involvement and demonstrate
aortic regurgitation and hemopericardium. In the ascending aorta,
intramural hematoma (wall thickening >7 mm due to hemorrhage in
the vessel wall, in the absence of an intimal flap) is a precursor
to dissection. Its presence is associated with rapid progression to
MDCTA is a rapid imaging tool to detect dissections. The
sensitivity/specificity of CTA has improved from 94%/87%
as CT technology has evolved from single-detector to MDCT (Figure
10). Because of high spatial resolution, CTA is considered superior
to MRA and transesophageal echocardiography (TEE) in identifying
branch vessel involvement.
Both CTA and MRA are more accurate than TEE for determining the
distal extent of a dissection (Figure 11). However, CTA cannot
detect aortic regurgitation, which is often the cause of death.
In a large retrospective study, evaluating the current imaging
modalities for the detection of dissection, Moore et al
have reported a sensitivity and specificity of 100% using MRA; a
sensitivity of 93% for CT; 87% using TEE; and 88% using catheter
angiography. An additional feature of MRA is accurate diagnosis of
aortic regurgitation, which affects prognosis.
Intramural hematoma, branch extension, and hemopericardium are also
reliably imaged by MRA. The choice between MDCTA and MRA is
governed by availability and physician preference and experience
with the modality. A large retrospective review showed that most
patients with aortic dissection get imaged by at least 2
this rationale is questionable with the use of MRA.
Abdominal aortic aneurysm is defined as enlargement of its arterial
diameter by >=50% of its normal caliber (>=3 cm in diameter).
Abdominal aortic aneurysms are increasingly being treated by the
Criteria for endovascular repair includes the following: 1)
infrarenal neck diameter of <3 cm; 2) minimum distance of 1.5 cm
between the renal artery origin and the aneurysm; 3) absence of
extensive mural thrombus; and 4) if the aneurysm is not too large
or does not have a very wide neck.
The use of automated volume-rendering software for CTA and MRA has
made both modalities equally efficient in preoperative evaluation
Because MRA does not use radiation or nephrotoxic contrast agents,
prospective studies have advocated CE-MRA as the sole imaging
modality for preoperative planning.
Engellau et al
have reported that measurement of aneurysm length is more accurate
on CE-MRAMIP images than by conventional angiography, probably due
to inherent magnification in conventional angiographic images.
Imaging is valuable in postprocedure evaluation. It can document
proper stent placement, moniter stability of aneurysm size, and
diagnose complications like stent migration, stenosis, and
endoleaks. The advantage of CTA is its usefulness in the presence
of multiple surgical clips adjacent to the aneurysm and when
stainless steel stents and stent-grafts have been used, as these
would cause significant artifacts on MR imaging. MR imaging with
steel stents is contraindicated,
though nitinol can be safely imaged.
Endoleaks are classified into four types. Type I are small leaks
adjacent to the proximal or distal end of the prosthesis. The type
II endoleak is reopacification of the aneurysm sac by vessels
arising from the lumbar arteries or inferior mesenteric artery.
Type III endoleaks are due to fracture of the struts of the graft.
Type IV leaks are due to porosity of stent material, and are very
rare. In evaluating type I endoleaks, CE-MRA and CTA have equal
efficacy. For type II endoleaks CE-MRA has a sensitivity of 94%, as
opposed to 50% by CTA (Figure 13).
Due to excellent temporal resolution, CE-MRA is suggested to be the
procedure of choice in evaluating all type II endoleaks.
MRA and CTA also help in detecting type III endoleaks.
Renal artery stenosis (RAS) is the cause of hypertension in 1%
to 5% of patients.
Atherosclerosis is the most common cause (60% to 70%) and usually
affects proximal renal arteries. Patients with atherosclerotic
renal artery stenosis often progress to renal failure if untreated.
Fibromuscular dysplasia (FMD) is the second most common cause (30%
to 40%) of renovascular hypertension with a more diffuse
involvement of the vessel.
Imaging is used to detect, grade, and evaluate hemodynamic
significance of the stenosis; provide a road map for interventional
procedures; and can help assess the functional status of the
Typical 16-channel MDCT evaluation has a slice thickness of 1 mm
with reconstruction at 0.5 mm, providing 500 images in 7 seconds
after intravenous administration of 80 mL of iodinated contrast
The sensitivity and specificity of CTA in the detection of
hemodynamically significant RAS in the main renal artery are 100%
and 97%, respectively (Figure 14).
The advantages of MDCTA include fast imaging, better spatial
resolution, depiction of smaller vessels, and better efficacy in
detecting RAS due to FMD. It can be easily standardized, does not
necessitate a higher level of technical ability, and is less
expensive. The disadvantages are in the use of ionizing radiation
and iodinated contrast, which may be a limiting factor for its use
in patients with compromised renal function. CTA provides limited
functional information as compared with MRA, and temporal
evaluation using CTA is not possible without increasing the
MRA can serve as a comprehensive examination in assessing RAS as
it allows simultaneous assessment of anatomic and functional
information. Generally, CE-MRA tends to overestimate the stenosis,
though to a much lesser extent as compared with 3D TOF. Additional
sequences can be used to avoid this pitfall. Three-dimensional PC
MRA detects a signal void to show a significant stenosis, provided
an appropriate velocity encoding value is chosen depending on the
patient's age and cardiac status. Wasser et al
reported equal efficacy of 3D PC MRA and catheter angiography for
the detection of hemodynamically significant stenosis, when
compared with pressure measurements.
Cine PC technique looks at the temporal pattern of blood flow in
a vessel. Velocity measurements are plotted in relation to time.
The acquisition is prospectively or retrospectively ECG-gated.
Multiple 2D images are obtained at the same location perpendicular
to the long axis of the artery. As stenosis develops, changes in
vascular resistance alter the shape of the curve correlating with
the degree of stenosis. Cardiac-gated PC flow measurements have a
sensitivity of 100% and a specificity of 93% for significant
stenosis when compared with catheter angiography
(Figure 15). A perfusion deficit must be shown to assess the
hemodynamic significance of a stenosis. A technique known as
extra-slice spin-tagging perfusion-weighted imaging
shows asymmetric perfusion of the kidneys in cases of unilateral
renal artery stenosis.
Examination time and spatial resolution are crucial factors in
comparing CTA and MRA. Weiger et al
have reported considerable improvement in temporal and/or spatial
resolution of MRA when using sensitivity encoding. Gadolinium has
been shown to be safe in patients with compromised renal function.
Lower extremity arteries
Peripheral vascular occlusive disease (PVOD) is a major health
problem, with an annual incidence of 4.5% to 8.0%
in men 55 years and older. The role of imaging in PVOD is to
establish the diagnosis, provide a preoperative roadmap for the
vascular surgeon or the interventionalist, and evaluate the grafts
to determine the presence and extent of collateral circulation. An
MDCTA is performed with an injection of 120 mL of iodinated
contrast at the rate of 4 mL/sec. Anatomic coverage from the celiac
artery to the toes is obtained in a single acquisition with a total
examination time of 10 minutes. A 3-station (abdominal aorta to
toes in 3 acquisitions) bolus-chase CE-MRA is performed with
intravenous administration of 40 mL of Gd. Upper and middle
stations are acquired coronally, and the last station in either
coronal or sagittal planes.
Phase encoding is reverse linear, linear, and elliptically centered
for each station, respectively.
The sensitivity of MDCTA for the occlusion of femoropopliteal,
popliteal, and tibial arteries is 100%, 100%, and 94%,
respectively. The specificities are 100%, 99%, and 98%,
For stenosis >75%, the overall sensitivity is reported to be
92.2% with a specificity of 96.8%.
Thus MDCTA is a fast, noninvasive, and accurate method for
evaluating PVOD. The whole- body radiation dose is 3.9 times
reduced as compared with that of digital subtraction angiography,
and the contrast requirements are less. The drawbacks of MDCTA
include limited function of the MIP software in distal anterior
tibial and peroneal vessels, and inadequate imaging of the vessels
in the presence of dense mural calcification. MDCTA is thus limited
when evaluating for a distal bypass if proximal narrowing is
A distal bypass done secondary to trauma can be evaluated (Figure
16). The postprocessing of CTA data is time consuming. Up to 45% of
patients who have PVOD also have renal artery stenosis
and compromised renal function, and a contrast load may precipitate
The accuracy of MRA in proximal lower extremities has improved
as the technique advanced from 2D TOF to CE-MRA. The specific
challenge for CE-MRA is to obtain good spatial resolution at
proximal stations, and yet scan fast enough to image the entire
lower extremity without significant venous contamination.
Cellulitis and osteomyelitis reduce the arterial-venous transit
time, leading to venous contamination.
CE-MRA can visualize stents adequately; however, it can miss stent
stenosis in the absence of secondary signs such as collateral
vessels. Time-resolved CE-MRA and parallel imaging have given
reduced scan times with subcentimeter spatial resolution. The
sensitivity for femoral, popliteal, and anterior tibial vessel
occlusion (>70% stenosis) is reported as 96.2%, 96.9%, and
95.5%, respectively. The specificities are 98%, 96.4%, and 91.8%,
respectively (Figure 17).
Improving spatial resolution, obtaining faster imaging, and
avoiding ionizing radiation and nephrotoxic iodinated contrast
makes CE-MRA the examination of choice for imaging the distal
A prototype four-dimensional (4D) CT with 256 detectors is under
development. A 4D image is a 3D image with the additional dimension
of time. This CT could allow 0.5 mm true isotropic voxel data to be
acquired much faster than the present 16-channel MDCT. MR imaging
is rapidly advancing. The future holds clinical imaging at magnet
strengths >3T, MR microscopy, complete functional assessment of
organ systems, 3D whole-body high resolution imaging with
16-channel phase array coils
(multidetector MR), advanced MR-guided interventional procedures,
new MRA contrast agents, and use of higher concentration