The recent development of the use of magnetic resonance imaging (MRI) to guide and monitor medical procedures holds much promise.
received his MD and PhD from the University of Pennsylvania,
Philadelphia, PA. He completed his residency in Diagnostic
Radiology at Beth Israel Deaconess Medical Center in Boston. He is
currently completing his fellowship in Interventional Radiology at
the University of Pennsylvania Medical Center and affiliated
hospitals. He will be taking a position in Cardiovascualr Imaging
and Interventional Radiology at St. Luke's Medical Center,
Milwaukee, WI, where he plans to perform Interventional Procedures
on their 0.7T open MR scanner.
The recent development of the use of magnetic resonance
imaging (MRI) to guide and monitor medical procedures holds much
promise. At present, MR-guided biopsies, wire localizations, and
chemical ablations can be performed on both open and closed MRI
systems with little or no specialized hardware. Thermal ablations
require specialized hardware, much of which already exists,
although this procedure needs to be further studied and developed
before it is accepted for widespread use. Finally, intravascular
procedures utilizing MR guidance are still in preliminary
research and development, but offer the potential for vascular
imaging and treatments not available with any other
As the revolution in minimally invasive surgery progresses, the
need for more advanced and accurate methods for guiding and
evaluating procedures is growing. Most intravascular radiological
procedures are performed under fluoroscopy and X-ray guidance.
Cross-sectional imaging, including ultrasound, computed tomography
(CT), and magnetic resonance imaging (MRI), is used for most
percutaneous procedures, such as biopsies and ablations. As these
cross-sectional imaging modalities have become faster and easier to
use, they have become more available to aid interventional
radiologists in these procedures.
Each modality has its own advantages and disadvantages. Although
the advantages of ultrasound (particularly real-time rapid imaging)
are frequently exploited, the disadvantages of poor tissue
resolution and poor penetration can limit the ability to perform
certain procedures. CT has better penetration and good contrast
between air, fat, and tissue. However, CT often has difficulty in
distinguishing lesions without the use of exogenous contrast
agents. Such agents, however, only aid conspicuity during a
transient imaging interval. In addition, CT provides only axial (or
limited off-axial) imaging guidance and can not be performed in
MRI, a newer modality for procedure guidance, has many potential
advantages that may make it more attractive than other currently
available modalities. One such advantage is the lack of radiation
exposure. Radiologists are subject to high levels of radiation
during their lifetime career; use of nonX-ray modalities such as
MRI will lower this risk. In addition, the multiplanar imaging
capabilities of MRI permit non-axial needle tracking, such as is
desired in reaching liver lesions from a subcostal approach. These
areas and lesions are often not seen well with ultrasound, and
tracks are difficult to follow with axial CT images. MRI is often
the only imaging modality that affords the inherent tissue contrast
needed to visualize lesions in steady state (ie, not during
transient contrast phases). MRI, with its high inherent tissue
contrast, can visualize different structures within an organ,
including vessels and ducts, without exogenous contrast agents.
Finally, and probably most importantly, MRI permits monitoring of
treatment, such as thermal ablation, directly and in real time
while the procedure is under way.
The recent advances in scanner design, MR-compatible
instruments, and rapid imaging have propelled MR into the vanguard
of imaging. This article will review the scanner designs, current
applications, and recent developments that will enable MR guidance
to be a reality in the next few years. This article will focus on
applications of MRI in abdominal and vascular procedures, while
briefly mentioning some of the neurointerventional, breast, and
The ideal MR system for interventional and surgical procedures
provides adequate access to the patient from all directions, yet
maintains a good signal to noise ratio, rapid imaging, and low
artifacts. As yet, there is no one perfect scanner system, but
there are many system designs, and many commercially available and
prototype systems in current use.
The first system design used for biopsy and neurosurgical
procedures in the United States is still used today at the Brigham
and Women's Hospital in Boston, MA. Their design, by GE Medical
Systems (Waukesha, WI), is called the "double donut" and consists
of a cylindrical scanner with 54 inches of the center portion
removed. This center area provides access to the patient at the
scanner's isocenter for the surgeon or interventionalist.
Disadvantages of this design include lower magnetic field strength
at the center of the bore (0.5 T) and lower field homogeniety. This
costly system may not suitable for most hospitals, but similar
systems have recently been installed in other large research
Several commercially available systems also permit access to the
patient in the scanner. Such open systems, which can also be used
for imaging larger or claustrophobic patients, generally use
biplanar horizontal magnets. These systems come in varying field
strengths. Permanent or resistive magnets, the least expensive,
have low field strength (<0.1 T to 0.3 T) and therefore have
limited tissue contrast, temporal, and spatial resolution. The more
expensive "mid-field" magnets are usually superconducting types and
have main magnet strengths of 0.5 T to 0.7 T. These are the highest
field strengths designed with open configuration. The C-arm designs
(produced by Seimens Medical Solutions, Erlangen, Germany) look
like conventional fluoroscopic C-arms and permit the widest access
to the patient but have the lowest field strength. Scanners by GE
Medical Systems, Hitachi Medical Systems America (Twinsburg, OH)
and Toshiba America Medical Systems (Tustin, CA) have wider magnets
with higher field strength, but require 2 to 4 supporting posts,
decreasing access to the patient (Figure 1A). A new prototype
design by Fonar (Melville, NY) incorporates the magnet hardware
into the ceiling and floor, thereby providing the widest space
between the magnets with 360š access (Figure 1B). This scanner is
still in the development stages.
Alternative ideas for MR-guided procedures include use of a
combination fluoroscopy/MRI suite. These rooms, such as that
developed by Phillips Medical Systems (Bothell, WA) and installed
at the University of Minnesota, Minneapolis, MN, center around a
custom table that slides a patient from a C-arm fluoroscope into a
short bore, high-field magnet.
This flexible design permits high-resolution rapid imaging, and is
suitable for monitoring procedures after instrument placement using
fluoroscopy or ultrasound. It can also be used for intravascular
procedures, whereby both MR and X-ray guidance can be used for
catheter and guidewire positioning, imaging, and treatment
Hardware developments continue to occur rapidly, and new designs
and advances will continue to appear as the technology matures.
Recent developments include X-ray systems incorporated into
mid-field MR systems.
Biopsies and localization procedures
Early MR pioneers first reported on the use of MR-compatible
devices for percutaneous biopsies and fluid drainages in the late
Following these early proofs of principle papers, there has been a
very slow translation of these methods into clinical practice.
Limitations include limited access to the patient and the
time-consuming nature of these procedures due to closed bore
systems. In closed-bore magnets, as in CT-guided procedures, the
following steps are performed: 1) imaging the lesion (with or
without intravenous contrast); 2) marking the location on the skin
surface; 3) advancing a needle toward the lesion; 4) securing the
needle; and 5) re-imaging to correct the position. Steps 3, 4, and
5 are repeated until the needle is visualized at the periphery of
the lesion. Each step requires the operator to move the patient in
and out of the magnet bore and view the images on the monitor
outside of the scanner room. In this manner, lesions in the liver,
kidney, adrenal glands, prostate, and breast have been biopsied
successfully when the lesions were not detected on any other
The most extensive experience in MR-guided procedures has come
from non-abdominal applications. In 2001, Salomonowitz
published his experience with 361 MR-guided procedures on 250
patients over 3 years. Procedures included spinal nerve root
injections, intrathoracic and vertebral body biopsies,
intervertebral joint blockades, and intra-abdominal (mostly liver,
pancreas, and lymph node) biopsies. He reported no
MR guidance has also been used for procedures on other organ
systems. Specialized neurosurgical procedures are performed in MR
suites like the one found at Brigham and Women's Hospital in
At the Hospital of the University of Pennsylvania (Philadelphia,
PA), breast biopsies are the most commonly performed procedures.
Most of these lesions are only detected on contrast-enhanced MRI.
These procedures require custom designed grids and compression
devices, which will not be reviewed here.
Reports have described MR-guided prostate biopsies for lesions
detected with endorectal prostate MRI.
Recent experiments with MR-guided radioactive seed placement and
cryotherapy for prostate cancer have also been reported.
In general, at the Hospital of the University of Pennsylvania,
MR is still reserved for biopsy and treatment procedures for
lesions not visualized with other modalities. We do not have an
open configuration scanner, and therefore use the algorithm
previously outlined: Imaging followed by needle manipulation with
the patient outside the magnet bore, then re-imaging. Procedures
can take hours, but often could not be done with any other
modality, other than open surgical biopsy.
Imaging is usually a combination of breath-hold (usually in
expiration) T1-weighted gradient-echo and/or single-shot fast
spin-echo T2-weighted images of the region of interest. If the
lesion is not well visualized, a contrast-enhanced
three-dimensional (3D) gradient-echo image with fat saturation is
obtained. After localizing the lesion, axial, sagittal, or coronal
thin-section images are obtained. An entry site is chosen from the
images, and that site on the skin is marked, using the axial
position LASER and the right/left distance from isocenter.
Following appropriate preparation, draping, and use of local
anesthesia, the needle is advanced toward the lesion from the
predetermined orientation, as calculated from the images. As the
needle is advanced, serial in-plane imaging can be acquired to
verify orientation and position. Needle artifacts are related to
pulse sequence and orientation of the needle with respect to the
main magnetic field. To minimize artifacts, imaging with highly
refocused pulse sequences such as fast- (turbo-) spin echo can be
After advancing to the correct position, perpendicular images along
the track of the needle can be obtained to verify position. Figure
2 shows two images obtained during needle placement in a liver
lesion near the inferior vena cava (IVC).
When the needle is at the appropriate position, fine needle
aspirates, core biopsy samples, or instillation of chemical
ablation agents can be performed. Following ablation or biopsy,
post-procedure imaging (T1-weighted or contrast-enhanced images)
are obtained to confirm treatment and/or evaluate for
The risk of complications and the length of time required for
such procedures could be lessened significantly with open
configuration systems; in-room monitors; and rapid, user-friendly
image prescription software. Over the last few years, such systems
have been demonstrated at research conferences. They continue to be
improved and are slowly being installed at many hospitals and
One of the largest areas of growth in interventional radiology
over the last few years is in the percutaneous treatment of cancer.
Cancersincluding hepatocellular carcinoma, lung cancer, and
metastatic colon cancerhave been treated in the liver, lung, and
bones. The indications for regional or local ablation continue to
grow, but currently include nonresectable hepatoma, colon cancer
metastases to the liver or bone, metastatic disease to the lung,
and nonresectable primary lung cancers. MRI is best suited for
liver lesions, as other modalities can be used to visualize lung
and bone tumors.
The incidence of hepatoma continues to increase as a result of
increases in hepatitis C in the United States. Although the only
definitive therapy is surgical resection, many patients are not
candidates due to liver dysfunction or high surgical risk.
Percutaneous procedures aimed at destroying the tumor and a small
margin of normal tissue can be performed in a number of ways.
Thermal therapy, with heat (radiofrequency [RF] ablation, lasers,
or focused ultrasound) or cold (cryotherapy), has been used for a
number of years. Chemical ablation with alcohol or acetic acid has
also been shown to be effective in the treatment of hepatomas.
However, the methods and doses for delivery of these agents have
not been defined completely, therefore more accurate monitoring of
the ablation procedure should be done to maximize efficacy and
RF ablation is the delivery of alternating electric current,
modulating in the low radiofrequency range of <1 MHz. This
electric current results in rapid ion movements in the adjacent
tissue that cause heating. The heat induces cellular damage
resulting in tumor necrosis. Necrosis can occur instantaneously at
temperatures >60šC, but can also occur using lower temperatures
for prolonged periods of time. MR is the only imaging modality with
the potential to directly measure tissue temperature to determine
when a lesion is fully treated.
There are multiple approaches to measuring temperature. The
three most practical methods are described below. The most
straightforward approach is the use of T1 maps to follow
temperature changes. T1 measures the longitudinal relaxation time
and is commonly evaluated qualitatively with T1-weighted images. T1
temperature images are easy to implement, provide relatively high
temporal and spatial resolution, but suffer from poor linearity
with temperature change and a somewhat poor signal-to-noise
Diffusion-weighted images are also temperature sensitive with
good sensitivity and similar temporal and spatial resolution to T1
images. However, factors other than temperature affect diffusion
weighting, including ischemia and changes in tissue characteristics
that occur during the coagulation process. In addition, diffusion
images are direction dependent, requiring additional images to
measure the diffusion tensor. Finally, fat can cause artifacts in
diffusion images acquired near fat/water interfaces.
The third and most promising method of measuring temperature is
based on the chemical shift of the water/proton frequency with
changes in temperature (proton resonance frequency method). Using
fast spoiled gradient-echo sequences that correct for most of the
local magnetic field inhomogenieties, the proton frequency shift is
related linearly to the temperature change. These PRF sequences
have high temporal resolution, good sensitivity and spatial
resolution, and have been shown to be more precise than T1- or
diffusion-weighted image sequences. Several published studies
describe the use of MR thermometric maps during tumor ablation.
Recent experience on low-field (0.2 T) systems show that both
conventional T2-weighted imaging and turbo spin-echo imaging
sequences provide conspicuity of the treated region similar to that
of contrast-enhanced imaging, permitting continuous monitoring
Further research is necessary to validate these findings before
this method can be put into widespread clinical use.
Other delivery systems can also be monitored with MR
thermometry. Cryoablation probes are MR compatible and can be
positioned with MR guidance, and monitored using similar pulse
sequences. Other heat delivery systems include focused ultrasound
technology in which arrays of high-frequency ultrasound transducers
placed outside the body deliver energy resulting in heating below
the skin surface. Integrated 3D imaging with computer tracking and
MR thermometric monitoring allows for the automated determination
of appropriate heat delivery.
This treatment has been applied in the brain, breast, and uterus
Early experience with totally noninvasive MR-guided focused
ultrasound therapy of benign fibroadenomas has been shown to be
although preliminary results suggest more work is needed in
monitoring and targeting lesions.
While thermal therapies require specialized and expensive
equipment that must be MR compatible, chemical ablation, on the
other hand, requires only a thin needle (or possibly one with
multiple side holes). Research in Asia, and more recently in Europe
and the United States, has shown efficacy and increased survival
with both alcohol and acetic acid instillations as treatment for
hepatocellular carcinoma and liver metastases.
Alcohol causes decreased T1-, T2-, and diffusion-weighted signal
on rapid imaging sequences. Following ablation with alcohol, the
low signal intensity on the T2-weighted images corresponds with the
area of nonenhancement on the post-gadolinium images and represents
the area of cell death. Immediately following instillation,
contrast-enhanced images can be obtained to verify the size of the
treatment area. Figure 3 shows some edema surrounding a lesion
recently treated with ethanol injection. The treated area appears
low in signal intensity, corresponding to the area of cell
necrosis. A small amount of subcapsular fluid is also noted.
An advantage of using acetic acid is the difference in the
resonant frequency of acetic acid versus water. Selective chemical
shift imaging can be performed to suppress the water during acetic
acid instillation. Thus the treated area will appear bright on an
acetic acid selective image. Figure 4 shows an image acquired
during treatment of a nonresectable hepatoma in a patient with poor
liver function and high surgical risk. The series of images shows
the increasing size of the treated region using chemical selective
The rationale for the use of MRI for interventional procedures
is based on recent studies describing atherosclerotic plaques with
luminal appearances that do not necessarily correspond with the
potential for harm. These vulnerable plaques--in which a thin
fibrous layer covers a large lipid core--have the potential to
rupture, leading to vessel occlusion even though the lesion may not
cause significant luminal narrowing. This phenomenon has been seen
in carotid and coronary vessels. MRI that permits visualization of
the vessel wall with a catheter in place will reveal the
characteristics of the plaque and vessel more clearly.
Additionally, MR-guided procedures can be performed without the
use of ionizing radiation or nephrotoxic contrast agents. Current
advancements in this field include development of intravascular
catheters and wires that can be tracked (ie, visualized) with the
MR scanner and that act as antennae for high-resolution vascular
imaging. Because synthetic catheter material lacks any MR signal,
various strategies have been developed for making them visible.
These strategies can be divided into passive or active catheter
Passive tracking uses material or contrast agents added to the
catheter to make them visible on MR sequences. Some commercially
made catheters incorporate dysprosium rings near the end of the
catheter, which appear as focal signal voids in bright blood
Such catheters were used by Bucker et al
to deploy IVC filters in pigs using real-time imaging in a combined
MR-fluoroscopy system. Omary et al filled catheters with dilute
GD-DTPA and imaged using two-dimensional (2D) and 3D spoiled
gradient-echo imaging sequences.
They successfully tracked catheter manipulation during renal artery
angioplasty in pigs (Figure 5A and 5B).
Passive devices can be tracked with conventional or fast imaging,
and/or background suppressed images, or can be superimposed upon
Passive catheters, however, can have poor conspicuity within
certain images, especially those with large section thickness or
low in-plane resolution.
Alternatively, active catheters produce inherent signal,
eliminating the problem of poor conspicuity. Active catheter
tracking uses catheters or guidewires embedded with an RF coil
directly attached to the scanner. The coil causes a peak in the
Fourier-transformed images corresponding to the position of the tip
of the coil.
After appropriate three-plane imaging, the position of the tip of
the coil is known and can be superimposed on any image. To
visualize more than just the catheter tip, modifications can be
made to the catheters. These modifications include using multiple
coils along the catheter length or using short linear segment
coils. The limitations of active catheter tracking include catheter
expense, local heating, and poor tracking if the patient moves.
However, these same catheters can be used for imaging the vessel
wall. This dual function is exemplified in Figure 6, which shows an
active catheter being tracked (Figure 6A) in a rabbit aorta and the
corresponding high-resolution vessel wall image (Figure 6B).
Semi-active catheters provide MR circuitry that enhances imaging
without direct electrical connections with the scanner.
This type of system reduces the risks of coil heating and artifacts
from standing waves. In addition, wireless inductive coupling can
result in signal enhancement when imaging the tissue immediately
adjacent to the coil. The coil can be mounted on an inflatable
balloon at the end of a catheter. Alternative strategies include
the use of a laser-controlled inductive coil that can be tuned or
detuned with photo-optics, for use in tracking or imaging without
risks of additional heating.
The preliminary research on this method is a very exciting
beginning to the nascent field of intravascular MRI. However,
before it becomes standard in clinical use, further developments in
active and passive catheter and guidewires are required.
Interventional MRI is a rapidly developing area. Cancer
treatment represents the fastest growing area among many
interventional radiology practices. MR currently represents the
most accurate modality for biopsies and cancer treatments.
Improvements in open MR systems, device technology, and image
prescription software will make these procedures easier and faster
in the next few years. Finally, although intravascular MRI is still
in its infancy, the possibilities inherent in this technology offer
a new paradigm for intravascular therapy based on direct
vessel-wall pathology, rather than luminal irregularity.
The author would like to thank Dr. David Roberts (University of
Pennsylvania Medical Center, Philadelphia, PA) and Dr. Reed Omary
(Northwestern Memorial Hospital, Chicago, IL) for their helpful
discussions and images and Dr. Albert Lardo (Johns Hopkins Medical
Center, Baltimore, MD) for his images.