Magnetic resonance-guided interventions offer the promise of seamless integration of anatomy and physiology without the use of ionizing radiation. Standard procedures, such as biopsy, tumor ablation, and endovascular stenting, can already be achieved with magnetic resonance imaging (MRI) safely and efficiently. Interventional MRI stands ready to be a gateway technology, allowing novel forms of noninvasive and minimally invasive therapies to be applied to a variety of diseases. Several applications that may spur interventional MR to be adopted widely are now visible on the horizon.
Dr. Hetts
is currently a fourth-year Resident in Diagnostic Radiology at
the University of California, San Francisco (UCSF), CA. He
received his AB from Harvard College and MD from Harvard Medical
School, Boston, MA. He completed an Internship in Internal
Medicine at Stanford University Hospital, Stanford, CA. He plans
to remain at UCSF for a Fellowship in Diagnostic and
Interventional Neuroradiology following completion of his
residency.
Interventional magnetic resonance imaging (iMRI) encompasses
diagnostic and therapeutic procedures that use magnetic resonance
(MR), in whole or part, for the guidance of, assistance in, or
immediate assessment of therapy. The advent of fast imaging
techniques, allowing real-time MR fluoroscopy, and innovations in
both MR and X-ray equipment design, have permitted the application
of MR guidance to a variety of clinical and research situations.
Widespread adoption of MR guidance, however, will depend in large
part on overcoming both technical and financial challenges.
1
This article will provide an overview of the burgeoning field of
iMRI and offer a glimpse of its future potential.
Advantages and disadvantages
When compared with interventions performed without imaging
guidance and procedures performed under X-ray fluoroscopic or
computed tomographic guidance alone, iMRI has several distinct
advantages and disadvantages (Table 1). If an intervention is
accomplished entirely under MR guidance, then neither patient nor
practitioner is exposed to ionizing radiation, an aspect of
particular importance when interventions are being performed on
children or on any patient likely to undergo frequent repeat
intervention. Most iMRI today is conducted under a combination of
X-ray and MR guidance due to the limited vascular spatial
resolution of conventional MR angiography at 0.5T and 1.5T, as
compared with X-ray angiography, thereby permitting reduction but
not elimination of the radiation dose.
2
Current research on MR angiography at 3T, in which, for example,
lenticulostriate arteries are visible (whereas they are usually
below the resolution of conventional digital subtraction
angiography), promises to eliminate the spatial resolution
disadvantage of MR in the intermediate term.
3
MR imaging (MRI) is the gold standard in the evaluation of most
soft-tissue structures but can also provide more and a greater
variety of physiologic information than can X-ray-based imaging
systems. Diffusion of water within tissue (a commonly used
surrogate marker for tissue ischemia), tissue perfusion, cardiac
wall motion, collateral blood flow around vascular stenoses, and
tissue temperature can all be measured noninvasively by MRI.
4,5
In addition, the strong main (B
0
) field of the scanner (0.5T to 3T for current clinical units)
allows the creation of a torque when a current is run through a
microcoil at the tip of a catheter, offering the possibility of
remote catheter-tip control.
6
Several disadvantages of iMRI must be noted. Exposure to loud
sound during real-time imaging sequences such as fast
gradient-recalled echo (GRE) and fast spoiled GRE (up to 99
decibels, the equivalent of a New York City subway train)
necessitates use of earplugs for patients and ear-covering
headphones for the operating team.
7,8
In addition, the biologic effects of repeated exposure to strong,
rapidly shifting magnetic fields are not fully understood, are not
yet routinely quantitated, and, hence, remain a challenge for
personnel monitoring. Limited access to the patient can be a
problem, depending on the geometry of the interventional magnet. A
relative dearth of MRI-compatible devices, ranging from patient
vital-sign monitoring equipment to surgical tools, adds to
procedural cost, as does the increased time often needed to perform
procedures under MR guidance.
Types of interventional MR systems available
As a nascent technique, iMRI is currently in a transitional
phase. Few interventions are performed solely under MR guidance;
instead, many also involve steps guided by X-ray fluoroscopy. Thus,
there has been impetus in recent years to develop systems that
combine clinical MR scanners with clinical X-ray fluoros-copy or
C-arm systems. Two approaches have been taken (Figure 1): 1)
placing a short-bore 1.5T MR scanner in line with an X-ray
angiography suite, with the patient transferable in minutes between
the 2 units via a sliding table
9
; and 2) fitting a specially designed X-ray tube in the gap between
magnets of a "double-donut" open 0.5T MR scanner.
10
Both systems have merits: the former produces better quality
(shorter acquisition time, higher resolution) X-ray and MR (1.5T)
images, and the latter allows nearly simultaneous X-ray and MR
acquisitions without cumbersome patient movement during procedures.
Interventions, such as brain tumor resection, in which MR imaging
is required only intermittently and no X-ray data are necessary,
can be achieved simply by removing the patient from a standard
clinical MR scanner stationed immediately adjacent to an operating
theater and returning the patient intermittently when new MR data
are needed.
1,11
For example, intermittent MR can be used to evaluate the margins of
resection not just versus a preacquired 3D stereotactic data set,
but also against a newly acquired data set that delineates
intraprocedural "brain sag" and intraprocedural complications, such
as hemorrhage or infarct.
Common challenges in developing interventional MR
devices
Devices used in the MR suite must be MR-safe, MR-compatible, and
visible to the interventionalist. To be safe, a device must use
little or no ferromagnetic material, as the missile effect of the
strong magnetic field of clinical magnets is well known and
hazardous. Electrical cables can also pose a danger through induced
heating by the magnetic gradients and radiofrequency (RF) pulses
used in imaging
12
; thus, the development of wireless forms of interventional devices
is currently under investigation.
13
Devices also need to be MR-compatible, not creating undue imaging
artifacts, yet still visible during the intervention. Visibility
can be achieved through active or passive means. Passive methods
rely on the MRI artifact created by paramagnetic materials, such as
gadolinium or dysprosium, which can be used to coat all or part of
the device. Placing one or more small RF antennas on the
interventional tool (eg, catheter, guidewire, needle) creates very
high MR signal around the device, greatly enhancing conspicuity.
Active devices, however, may be somewhat limited by local heating,
especially when used in small blood vessels or solid tissues, where
flowing blood does not dissipate heat as efficiently as it does in
large vessels.
An exciting development related to placing multiple coils on
catheter tips is target navigation, essentially allowing the tip of
the catheter and its intended target to define the plane of imaging
in 3D space. In this type of active visualization, the invasive
device and target are always included in a single image,
facilitating catheter tracking toward the desired target.
14
Further development of user-friendly software interfaces and rapid
imaging techniques continue to increase the speed and ease of MR
interventions, such as the use of real-time undersampled projection
reconstruction combined with the temporal filtering technique of
reduced field-of-view.
15
Current and future applications
Interventional MR is being applied both to procedures that are
already clinically routine-biopsies, tumor ablations, angioplasty,
and stent-ing-and to novel procedures made possible by this new
guidance modality.
Breast biopsy
MRI has the ability to detect benign and malignant breast
lesions not visible by X-ray- or ultrasound-based imaging. As such,
MR-guided biopsy or hook-wire localization of these lesions is the
only practical option and has quickly become the first iMRI
technique to be adopted in clinical practice at large academic
institutions.
16
Stereotactic
17
and freehand
18
techniques have been developed for large-core (14-gauge) needle
biopsy using dedicated breast coils in clinical short-bore 1.5T and
open 0.5T scanners, respectively. Stereotactic core biopsy results
are promising: 99% of biopsies in one study resulted in histologic
diagnosis, with 98% diagnostic accuracy, and a change in treatment
in 70% of patients.
19
The freehand method offers more flexibility in needle and patient
positioning, greater access to lesions near the chest wall and
axilla, less tissue distortion because compression is not used,
and, because it is performed in an open magnet, the ability to
sample one or more lesions without removing the patient from the
magnet (Figure 2).
20
Tumor ablation
Both primary and metastatic tumors in various organs have long
been ablated by RF thermocoagulation, often under ultrasound
guidance. With its superior soft-tissue contrast and the ability to
measure temperature nonivasively,
5,21
MRI stands to become the modality of choice for monitoring
minimally invasive and noninvasive ablation of a variety of benign
and malignant lesions. Focused ultrasound (FUS) is a gateway
technology that permits completely noninvasive thermocoagulation of
lesions in breast, uterus, liver, kidney, prostate, and brain.
22
Focused ultrasound beams generated by a piezoelectric phased-array
transducer mounted in the MR table (ExAblate 2000;
InSightec-TxSonics, Haifa, Israel and Dallas, TX) penetrate soft
tissue and can be focused deep within the patient's body on a few
cubic millimeters of tissue (Figure 3A). Energy deposition at the
focus causes rapid temperature elevations (50°C to 90°C in a few
seconds) with sharply demarcated boundaries, sparing surrounding
tissue.
23,24
Recent feasibility studies of MR-guided focused ultrasound surgery
of benign breast fibroadenomas and invasive breast carcinomas have
shown that the technique is well tolerated (with minor skin burns
being the only complication noted in less than 20% of patients) and
effective at producing necrosis of between half and 95% of tumor
volumes (Figure 3B).
24,25
Future applications, particularly to malignant lesions, will need
to overframe (ie, treat slightly beyond the borders of) the
gadolinium-enhancing portion of tumors with more sonication foci,
as most of the residual tumor in these studies was present at the
periphery of lesions. Focused ultrasound thermocoagulation and
necrosis of uterine leiomyomas has also been shown to be visible by
MR in >98% of sonications in a small feasibility study. A phase
III multicenter clinical trial of FUS for uterine fibroid treatment
is ongoing and will yield risk/benefit profiles for this
noninvasive therapy of this extremely prevalent condition.
22,26
Another novel approach to MR-guided tumor ablation is the use of
magnetic targeted chemotherapy administered via transcatheter
embolization. In a phase I/II clinical trial involving 4 patients
with inoperable hepatocellular carcinoma, doxorubicin linked to
metallic iron-activated carbon (MTC-DOX, FeRx, San Diego, CA) was
administered endovascularly to the hepatic artery branch(es)
supplying the tumor under X-ray fluoroscopic guidance.
27
A 5-kilogauss portable external magnet (Magnet Sales, Culver City,
CA) was positioned over the liver in the approximate location of
the tumor during MTC-DOX injection to attract the chemotherapeutic
agent. MRI performed before and after MTC-DOX administration showed
preferential deposition of chemotherapy within the tumor as opposed
to normal adjacent liver tissue (Figure 4). Increased accuracy of
chemotherapy delivery may allow not only dose reduction and
concomitant reduction in toxicity, but also a reduced number of
chemotherapy sessions.
Neurosurgery
Death rates from malignant primary brain neoplasms have remained
virtually static for more than 30 years despite numerous advances
in neurosurgical technique and diagnostic imaging. Resection of a
brain tumor presents a difficult clinical choice for the surgeon:
maximal tumor resection versus creation of new neurologic deficits.
Initial experience with intraoperative MRI in an open-bore 0.5T
magnet was favorable in guiding craniotomies, biopsies,
intracranial cyst evaluations, subdural drainages, and
trans-sphenoidal pituitary resections, with complication and
infection rates similar to those seen in standard surgery and with
the added benefit of detecting hyperacute hemorrhage in about 1% of
patients.
28
Other centers have advocated use of intermittent intraprocedural
imaging, rotating the patient into the magnet for anatomic
evaluation and rotating the patient out to the 5-gauss line
(generally considered the safe line at which MR incompatible
instruments may be used) where a standard operating microscope is
mounted. A clear advantage of intraoperative MRI over preoperative
stereotaxy is its ability to depict brain shift during tumor
resection, as there is no correlation between the volume of
resection and the direction of brain shift-only a correlation in
the amount of shift.
11
Concomitantly, intraoperative diffusion tensor tractography holds
promise in remapping shifted white matter tracts after partial
resections, allowing the surgeon to plan further resection based on
updated anatomic information. Research is also under way to
incorporate intraprocedural functional MRI for motor mapping and MR
spectroscopy for tumor metabolite mapping.
11
Animal studies are also ongoing to develop focused ultrasound as a
minimally invasive tool for brain tumor thermocoagulation, either
through small craniotomies or through the intact skull, thus coming
one step closer to the holy grail of "bloodless neurosurgery,"
which has been long espoused but is, as yet, unrealized.
29
Drug, gene, and cell delivery
Pulsed sonication with focused ultrasound at lower power than is
used for thermocoagulation (as low as 0.2W versus 30W to 450W) can
be used to open the blood-brain barrier (BBB) transiently without
damaging adjacent tissue, thus allowing delivery of drugs or cells
to sites in the central nervous system (CNS) that are normally
sequestered from the vasculature.
30,31
Transient opening of the BBB is shown on T1-weight-ed MRI by focal
enhancement after gadolinium administration, an effect that fades
over hours to days and has been correlated histologically with the
presence of normally impermeable test molecules in brain
parenchyma. The basis for BBB opening is thought to be mechanical
stress imparted on vascular walls when FUS is used to cavitate an
intravascular preformed gas microbubble ultrasonographic contrast
agent (Optison, Mallinkrodt Inc., St. Louis, MO) injected shortly
before sonication. Recent work in rabbits has shown transient BBB
opening though an intact skull with induced pressures of 0.5 MPa.
32
The normally impermeable large molecule horseradish peroxidase was
injected intravascularly and was shown by electron microscopy to
pass into the brain parenchyma interstitium adjacent to
capillaries, arterioles, and venules by both transendothelial and
paraendothelial routes. Given that many novel therapies (drugs,
gene constructs, stem cells) for CNS disorders ranging from
ischemia to Parkinson's disease are impermeable to the BBB, the use
of noninvasive FUS to allow focal delivery of these agents is very
exciting.
Delivery of stem cells and extracellular markers under MR
guidance has also been shown by using endovascular catheters.
Implantation of mesenchymal stem cells (MSCs) to myocardial infarct
borders may allow therapeutic regeneration.
33
To that end, a group of investigators at the National Heart, Lung,
and Blood Institute used a clinical 1.5T MR scanner (CV/i, GE
Medical Systems, Milwaukee, WI) customized with rapid imaging (8
frames per second), independent color highlighting of catheter
channels, mulislice 3D rendering, catheter-only viewing mode, and
infarct-enhanced imaging to guide an MR-compatible catheter
(modified Stiletto 2, Boston Scientific, Natick, MA) to the margin
of precreated anterior wall infarcts in swine and to deposit
iron-fluorescent particle-labeled MSCs (Figure 5).
34
Although infarcts were created using platinum coils deployed in the
left anterior descending coronary artery under X-ray guidance, the
therapeutic portion of the experiments (precise delivery of MSCs)
was conducted solely under MR fluoroscopic guidance. Similar work,
using a combination hybrid interventional X-ray/MR fluoroscopy
system (XMR) facility, has also been successful in delivering
extracellular marker material to the myocardium in dogs.
35
Endovascular interventions
Catheter-based iMRI (CBiMRI) was reviewed in detail in last
year's
Applied Radiology
Residents' Forum.
36
Two areas in which MR shows pathophysiology, and thus can be used
for immediate evaluation of treatment efficacy, are aortic
coarctation repair and carotid stenting. In a canine study,
37
acute aortic coarctation was created with plastic ligatures placed
around the descending thoracic aorta and collateral blood flow
around the coarctation (comparing blood flow in the aorta above the
coarctation and at the diaphragm) was assessed by retrospectively
gated balanced T1 fast field-echo MR. Subsequently, a
self-expanding nitinol stent was deployed across the coarctation
under MR fluoroscopic guidance (using a balanced fast field-echo
sequence), with poststenting flow measurements showing no residual
collateral flow, implying complete physiologic reversal of the
coarctation. Similarly, no residual high-velocity jet was apparent
on velocity-encoded cine MRI. However, evaluation of the aortic
lumen in the area of the stent was possible only using X-ray
angiography due to MR signal dropout within the nitinol stent, thus
emphasizing the utility of XMR facilities until, and unless, MR
transparent stents become available. Similar MR angiography signal
dropout is evident in human carotid stenting,
38
where evaluation of cerebral perfusion by arterial spin labeling
before and after intervention is under way. For more on the
evaluation of tissue perfusion using MR, please refer to the review
in this issue by Dr. Zaharchuk.
Remote-controlled interventions
The strong ambient (B
0
) magnetic field of the MR scanner allows a special opportunity for
iMRI not available in other forms of interventional guidance.
6
As illustrated in Figure 6, if an electrical current is run through
a coil mounted on the tip of a catheter, a magnetic moment is
generated. This magnetic moment will then interact with the B
0
field, creating a torque, which, in turn, causes deflection of the
catheter tip. If 3 coils are mounted orthogonally on a catheter
tip, then virtually any tip orientation can be achieved through
selective activation of each coil. Initial in vitro experiments
have shown the ability to navigate simulated vascular bifurcations
up to 90°. The combination of technologies like this with advances
in robotics
1
and nanotechnology offers the possibility of truly integrated
remote-controlled interventions under MRI guidance in vessels too
small or tortuous to be reached with current methods.
Conclusion
Interventional MRI is beginning to come into its own through
applications using old procedures and as the gateway technology to
novel, minimally invasive therapies. Although significant hurdles
in terms of expense and procedure time will limit the rate of its
adoption, iMRI is already reaching clinical utility at academic
institutions in the areas of breast biopsy, tumor ablation, and
neurosurgical guidance. As a gateway to new forms of intervention,
however, MRI promises to become a major player in the not too
distant future.
Acknowledgments
The author would like to thank Anthony Bernhardt, PhD, Michael
Guttman, MS, Gloria Hwang, MD, Timothy Roberts, PhD, Mark Wilson,
MD, and Greg Zaharchuk, MD, PhD for helpful discussions and figure
images.