A truly hybrid interventional MR/X-ray system: feasibility demonstration

Article (PDF Available)inJournal of Magnetic Resonance Imaging 13(2):294-300 · March 2001with27 Reads
DOI: 10.1002/1522-2586(200102)13:2<294::AID-JMRI1042>3.3.CO;2-O · Source: PubMed
A system enabling both x-ray fluoroscopy and MRI in a single exam, without requiring patient repositioning, would be a powerful tool for image-guided interventions. We studied the technical issues related to acquisition of x-ray images inside an open MRI system (GE Signa SP). The system includes a flat-panel x-ray detector (GE Medical Systems) placed under the patient bed, a fixed-anode x-ray tube overhead with the anode-cathode axis aligned with the main magnetic field and a high-frequency x-ray generator (Lunar Corp.). New challenges investigated related to: 1) deflection and defocusing of the electron beam of the x-ray tube; 2) proper functioning of the flat panel; 3) effects on B0 field homogeneity; and 4) additional RF noise in the MR images. We have acquired high-quality x-ray and MR images without repositioning the object using our hybrid system, which demonstrates the feasibility of this new configuration. Further work is required to ensure that the highest possible image quality is achieved with both MR and x-ray modalities.



Full-text (PDF)

Available from: David L. Ergun
Original Research
A Truly Hybrid Interventional MR/X-Ray System:
Feasibility Demonstration
Rebecca Fahrig, PhD,
Kim Butts, PhD,
John A. Rowlands, PhD,
Rowland Saunders, BS,
John Stanton, BSE,
Grant M. Stevens, MS,
Bruce L. Daniel, MD,
Zhifei Wen, BS,
David L. Ergun, PhD,
and Norbert J. Pelc, ScD
A system enabling both x-ray fluoroscopy and MRI in a
single exam, without requiring patient repositioning,
would be a powerful tool for image-guided interventions.
We studied the technical issues related to acquisition of
x-ray images inside an open MRI system (GE Signa SP). The
system includes a flat-panel x-ray detector (GE Medical
Systems) placed under the patient bed, a fixed-anode x-ray
tube overhead with the anode-cathode axis aligned with
the main magnetic field and a high-frequency x-ray gener-
ator (Lunar Corp.). New challenges investigated related to:
1) deflection and defocusing of the electron beam of the
x-ray tube; 2) proper functioning of the flat panel; 3) effects
on B
field homogeneity; and 4) additional RF noise in the
MR images. We have acquired high-quality x-ray and MR
images without repositioning the object using our hybrid
system, which demonstrates the feasibility of this new
configuration. Further work is required to ensure that the
highest possible image quality is achieved with both MR
and x-ray modalities. J. Magn. Reson. Imaging 2001;13:
294–300. © 2001 Wiley-Liss, Inc.
tools for guiding interventional procedures. MRI has
exquisite soft tissue contrast, excellent three-dimen-
sional visualization, the ability to image in any scan
plane, and the possibility of providing physiological in-
formation. X-ray fluoroscopy provides high-resolution
( 2 lp/mm, 1024
matrix), real-time (30 frames/sec)
two-dimensional projections with excellent contrast for
guidance and placement of catheters, stents, platinum
coils, and other metallic devices. Although recent ad-
vances in MR fluoroscopic imaging (1–4) and MR-com-
patible devices (5) have been significant, the combina-
tion of temporal and spatial resolution provided by
x-ray fluoroscopy still exceeds any realistic expecta-
tions for MRI. In addition, the two-dimensional projec-
tion format of x-ray fluoroscopy is an advantage in
many applications. While MRI can collect projection
images, these often suffer from serious artifacts.
There are several interventional procedures that
could benefit from using both x-ray and MR for image
guidance. Transjugular intrahepatic portosystemic
shunt (TIPS) is a common clinical procedure to treat
bleeding esophageal varices due to portal venous hy-
pertension. A shunt is placed between the right hepatic
vein and the right portal vein. While generally done with
x-ray guidance, previous work has shown that the use
of MR to guide the portal-venous puncture is more
accurate than x-ray fluoroscopy, while catheter and
stent placement are more easily achieved using x-ray
guidance (6 –7). The chemoembolization of hepatic tu-
mors might also benefit from a hybrid system. In this
procedure, an embolizing agent is subselectively in-
jected through a catheter placed in one of the hepatic
arteries. Steering and guidance of catheter placement is
best achieved using x-ray fluoroscopy; the lesions and
the volume of liver tissue supplied by the branch can be
easily visualized using three-dimensional MR images in
conjunction with an MR contrast agent. Other proce-
dures that might be improved include vascular appli-
cations, biliary drainages, abscess drainages, gallstone
removal, precutaneous nephrostomy, and kidney stone
removal. It is also thought that nonvascular procedures
such as vertebroplasty, placement of pedicle screws,
and other minimally invasive spinal procedures could
benefit from multimodality imaging (8). It is likely that a
hybrid system could increase the accuracy and shorten
the duration of these and other procedures.
Other groups have recognized the potential utility of a
combined system and have developed configurations
that allow acquisition of x-ray and MR images within
the same room. Probably due to concerns about each
system affecting the other, these configurations have
placed the x-ray hardware some distance from the mag-
net and have stopped short of a true hybrid system. The
Department of Radiology, Stanford University, Stanford, California
Department of Applied Physics, Stanford University, Stanford, Califor-
Department of Physics, Stanford University, Stanford, California.
Department of Medical Biophysics, University of Toronto, and Sunny-
brook Health Sciences Center.
GE Medical Systems, Milwaukee, Wisconsin.
Lunar Corporation, Madison, Wisconsin.
Contract grant sponsor: Terry Fox NCIC Program Project; Contract
grant number: NIH P41 RR09784.
Portions of this work were presented at the ISMRM 2000 in Denver,
*Address reprint requests to: R.F., Department of Radiology, Lucas
MRS Center, 1201 Welch Rd., Stanford University, Stanford, CA 94305-
5488. E-mail: fahrig@s-word.stanford.edu
Received May 15, 2000; Accepted August 10, 2000.
© 2001 Wiley-Liss, Inc. 294
patient (and table) must be moved between imaging
modalities (9 –) and United States Patent No. 6.031,888
(2000), which can be cumbersome and time consum-
ing, and registration of the images acquired on the two
systems is not easily maintained. Recently introduced
flat-panel digital x-ray detectors (12–15) are far more
immune to magnetic fields than x-ray image intensifi-
ers, thereby simplifying a true integration. We have
embarked on a project to place an x-ray fluoroscopy
unit completely within the bore of an interventional MRI
system. Both x-ray fluoroscopy and MR images would
be available to the clinician without moving the patient,
although, at least initially, each system would be off, or
inactive, while the other system is active.
In the final implementation, the x-ray components
should have minimal impact on the operation and im-
age quality of the MR system, and the x-ray system
should have performance (dose efficiency, SNR, spatial
resolution, and distortion) comparable to that of a con-
ventional x-ray fluoroscope. As will be shown, the x-ray
performance can be assured. The main impacts on the
MR system are the possibilities of added noise and
reduction in main field homogeneity. It is difficult to set
absolute limits on these, but it is reasonable to require
that the SNR of the MR system decrease by no more
than 10%. As will be discussed below, a reasonable
target for the allowed inhomogeneity is the imposition
of gradients of on the order of 25 Hz/cm or less. The
purpose of this article is to demonstrate that such a
system is feasible by showing that these requirements
are already met or can reasonably be expected to be met
in the near future.
We briefly describe our system and outline the tech-
nical challenges that must be overcome before x-ray
and MR images can be acquired successfully using a
truly hybrid system. Feasibility is demonstrated by ad-
dressing each of the aspects outlined above. We then
show high-quality x-ray and MR images of the same
object that were acquired using our hybrid system, fur-
ther demonstrating the feasibility of this new configu-
A standard x-ray system consists of an x-ray source
(x-ray tube and housing), a high-voltage (HV) generator,
an x-ray detector, a detector power supply, data acqui-
sition electronics, and a display. The two components
that have to be placed in the magnetic field are the x-ray
source and the x-ray detector. Both must therefore be
compatible with the MR system. The locations of the
x-ray source and detector within our iMRI system are
shown in Fig. 1. The x-ray tube and collimator are
placed next to the upper connection between the two
‘donuts’ and the x-ray detector is placed in the bridge
below the x-ray transparent patient cradle. This place-
ment allows imaging at an angle of 18° to the vertical
(i.e., in an approximately vertical AP projection if the
patient is supine). Ultimately, the source could be
placed within the upper enclosure. The magnetic field
at the two locations is the same.
Most x-ray tubes used in diagnostic x-ray imaging
have rotating anodes, allowing high exposure rates
without target vaporization (16). We have, instead, used
a fixed-anode x-ray tube (Brand X-ray Co., Addison, IL)
for two reasons: 1) so that an induction motor is not
required and 2) because the tube and housing can be
compact, thereby maximizing mounting flexibility and
available working volume in the bore. The source was
adapted from that of a bone densitometry system (DPX,
Lunar Corporation, Madison, WI). In order to ensure
maximum MR compatibility, most magnetic compo-
nents within the glass envelope of the x-ray tube (such
as the outer sleeve of the cathode) were replaced by the
vendor with equivalent nonmagnetic stainless steel
components. The remainder of the x-ray source, includ-
ing housing (aluminum), shielding (lead), and HV ca-
bles (copper) were nonmagnetic. The tube and housing
are cooled by passive convection of oil and water, re-
The SNR in x-ray fluoroscopy is strongly determined
by the x-ray flux. Typical x-ray techniques during fluo-
roscopic imaging are 60–100 kVp and 0.5–5.0 mA (17).
Our tube can operate at 80 kVp and 2.75 mA (220W)
continuously. As the duty cycle is decreased, as is typ-
ical in fluoro studies, the instantaneous power can be
increased. It can operate at 400W for more than 10 sec.
Thus, fixed anode tubes can provide sufficient x-ray
flux for long, low-dose fluoroscopic exposures. Indeed,
some commercial C-arm x-ray systems use fixed anode
tubes. The size of the x-ray tube focal spot can limit the
Figure 1. Interventional MR system showing the locations of
the x-ray system components. The x-ray flat panel detector fits
under the patient cradle, and the x-ray source is placed next to
the upper horizontal enclosure, allowing acquisition of (nearly)
AP projections through a supine patient.
Hybrid Interventional MR/X-Ray System
spatial resolution of the system. Typical systems use
tubes with projected focal spots in the range of 1 mm in
size. The specification for our tube is 0.5 mm, well
below the current standard. Even higher x-ray flux can
be available as the focal spot is made larger.
X-rays are generated in an x-ray tube by directing a
beam of high-energy (80 keV) electrons at the x-ray
tube target. When the tube is within the bore of the
magnet, the electron beam will experience the main B
field. The strength of the field at the location of the x-ray
tube is indicated by the arrow in Fig. 2. Interaction
between the large magnetic field and the electron beam
may lead to deflection or defocusing of the beam and, in
the worst case, may cause the electrons to miss the
target entirely. We therefore sought to align the anode-
cathode direction with the direction of the main mag-
netic field. By symmetry, we reasoned that the field is
perpendicular to a plane drawn midway between the
two magnet cryostats and, therefore, aligning the an-
ode-cathode axis perpendicular to this plane should
minimize deflection.
The HV generator (Lunar Corporation, Madison, WI)
provides both the accelerating voltage between cathode
and anode and the AC current required for heating the
cathode filament. We chose a high-frequency (50 kHz
HV, 25 kHz power to the filament) generator that uses
standard (115V AC / 60 Hz) line power. In initial tests,
the generator was placed outside the magnet room,
with connection to the x-ray tube provided by 40-foot
long, shielded HV cables. The present generator has a
peak power rating of 240W; higher-power generators
are available if needed.
The x-ray flat panel detector (prototype of the Revo-
lution™ detector, GE Medical Systems, Milwaukee, WI)
contains a large (20 cm 20 cm) amorphous silicon
panel that is used to image the light generated by a
CsI(Tl) phosphor conversion layer. As in x-ray image
intensifiers (XRII), the CsI is grown in needles to mini-
mize light diffusion and image blurring. The detection
efficiency is comparable to that in an XRII. The panel
contains 1024 1024, 0.2 0.2 mm
pixels, each
having a photodiode and an integrator that collects
electric charge in response to incident light. Thin-film
transistors at each pixel act as switches and are con-
trolled, row by row, using common gate lines. Readout
lines along columns allow the charge switched out by
the transistor to be measured and digitized. The detec-
tor has good noise properties, permitting it to be oper-
ated at the low signal levels of fluoroscopic imaging at
30 frames per second (18). There is no fundamental
reason why a detector of this type will not work in a
magnetic field since the underlying electronics are very
similar to the LCD displays already in use within the
MR suite. It is essentially impossible for such a detector
to exhibit image distortion. Nonetheless, the detector
had not been designed with operation at high magnetic
fields in mind. All of the components in the detector
were evaluated and it was determined that, with the
exception of a magnetically operated relay that was
eliminated, none of the electronic components were
likely to be affected by magnetic fields.
The presence of the x-ray system within the bore of
the magnet can have two major effects on MR image
quality. First, main field homogeneity may be degraded
due to the presence of the x-ray tube and housing on
one side and the detector (with its cooling plate and
housing) on the other. Second, additional RF loading
and RF interference may lead to increased noise in the
MR image.
Investigation I—MR Image Quality with X-ray
Components in Place
field homogeneity was measured for three configu
rations: 1) baseline; 2) x-ray source was placed in the
proposed location; and 3) the flat-panel detector was
placed in the proposed location. Gradient echo images
of a 28 cm diameter spherical phantom were acquired
using a transmit-receive body birdcage coil (GRE, TR/
TE 150/20, 30 30 cm
FOV, 256 256 matrix, 1
cm slice thickness, 30° flip angle, 16 kHz bandwidth).
This was done for the central axial, sagittal, and coronal
planes. From the complex images, the maximum fre-
quency offset and the largest frequency gradient were
calculated for each of the configurations, before and
after reshimming using linear terms.
To directly investigate image distortion, images of the
MR system’s image quality phantom were acquired
(FSE, TR/TE 500/18, 256 128, FOV 24 24
, 1 cm slice thickness, 16 kHz bandwidth) with no
x-ray equipment in the room and with the detector and
x-ray source in their proposed locations. In addition, to
quantify possible distortion, the ratio of minor and ma-
jor axes of the spherical phantom were measured in the
axial, sagittal, and coronal images that were acquired
during the investigation of field homogeneity. Addi-
tional images were collected with gradient echo se-
quences, as described below.
The possible RF noise increase was investigated by
acquiring images of a water phantom using a birdcage
head coil and a standard clinical multislice imaging
sequence (GRE, TR/TE 150/6.7, flip angle 60°, 24
24 FOV, 16 kHz bandwidth, 1 cm slice thickness).
Figure 2. Field strength as a function of distance from the
isocenter of the magnet. The field strength at the location of the
x-ray tube is indicated by the arrow.
296 Fahrig et al.
The acquisition conditions are summarized in Table 1.
A baseline image was first acquired with no x-ray com-
ponents in the MR suite. The x-ray components were
then installed and connected as follows: the flat panel
detector was connected to its power supply through
low-pass filters in the penetration panel; the detector
cooling hoses and fiber-optic signal cables were fed
through a waveguide and connected to their respective
systems; and the HV cables were fed through
waveguides and connected to the HV power supply. MR
images were acquired with all of the x-ray components
turned off, and then with the components still off and
the HV cables outside the magnet room wrapped in foil.
Although the expected configuration for the system
would have all of the x-ray components in place but not
on during the acquisition of MR images, we examined
the effect of turning these components on. Thus, MR
images were also acquired with all of the components
turned on, and then with all of the components turned
on and the detector wrapped in aluminum foil. The foil
around the detector and the HV cables represented a
first attempt to provide some RF shielding. The signal
within the phantom and the pixel standard deviation
(SD) in the background of each image were measured in
regions away from zipper artifact, if such artifact was
Investigation II—X-ray Image Quality
in the MR Suite
Investigation of x-ray system performance while in a
magnetic field was carried out using several ap-
proaches. The x-ray source and detector were mounted
on a stand and slowly advanced into the magnet bore,
with the source on the central plane between the cryo-
stats and the anode-cathode axis alignied with the nor-
mal to this plane. Pinhole images of the focal spot loca-
tion (30 m pinhole 9.6 cm from the focal spot,
magnification 9.23) were acquired and compared to a
similar image taken outside the field. Sensitivity to mis-
alignment was investigated by rotating the electron-
beam axis relative to the main magnetic field by 5° and
measuring the resulting motion of the image of the focal
spot. After optimizing the alignment, images of a lead
resolution test pattern, placed 5 mm from the surface of
the detector, were acquired (74 kVp, 1 mA, 30 frames/
s). Finally, x-ray images of a pear and an orange with a
22-gauge needle were acquired (50 kVp, 0.25 mA, 30
frames/s). For comparison, MR images of the same
phantom were acquired using a protocol typically used
during interventional procedures (linearly polarized
flexible RF coil, GRE, TR/TE 150/2.8, 30 30 cm
FOV, 256 192 matrix, 1 cm slice thickness, 16 kHz
bandwidth, 30 sec imaging time) with the x-ray compo-
nents in position and off.
Investigation I—MR Image Quality with X-ray
Components in Place
Magnetic field homogeneity maps showed significant
degradation due to the x-ray components. After shim-
ming with linear terms, the x-ray source had no mea-
surable effect on field homogeneity but the detector had
a significant effect. The worst impact was near the pa-
tient table. The maximum observed frequency shift was
on the order of 350 Hz (16 ppm) and the maximum
frequency gradient was on the order of 120 Hz/cm.
Near the magnet center the effect was much lower. As
would be expected, after shimming with linear terms,
the remaining inhomogeneity appears roughly qua-
dratic. It is assumed that with full reshimming, the
homogeneity would be substantially improved.
The images of the quality assurance phantom (see
Fig. 3) show no visible distortion. In addition, measure-
ments of the distortion ratio (minor-to-major axes in
slices through the sphere) were 1.008, 1.000, and 1.000
Table 1
Summary of Experimental Conditions Under Which Noise
Measurements Were Made
Experimental conditions Noise SNR
Baseline 3.7 216
X-ray equipment in, but turned off 5.7 134
X-ray equipment off, foil around HV cables 4.1 191
X-ray equipment in and turned on 11.5 66
X-ray equipment on, foil around detector 7.7 94
Noise was measured in the background of each image and signal
was measured in the object.
Figure 3. FSE images (TR/TE
2000/25 ms) of the image quality
phantom acquired a) with no x-ray
equipment in the MR suite and b)
with the x-ray source and detector in
their proposed locations within the
bore of the magnet.
Hybrid Interventional MR/X-Ray System
in the axial, sagittal, and coronal slices, respectively,
indicating that global distortion is minimal.
Noise and SNR measured in the MR images under the
various conditions is summarized in Table 1. In the
proposed configuration, with all of the x-ray equipment
in place but off, and with modest shielding around the
HV cables, SNR decreased from baseline by only 11.6%.
When the x-ray components were turned on, the stan-
dard deviation in the background was three times
larger than baseline, and also contained a significant
“zipper” artifact, indicative of strong narrow-band
noise. Not surprisingly, the detector is a significant
source of RF noise.
Investigation II—X-ray Image Quality
in the MR Suite
Three images of the focal spot are shown in Fig. 4. When
alignment was maintained between the B
field and the
anode-cathode axis, no motion of the focal spot was
detected. Focal spot area did, however, increase by
17%; the projected focal spot size at the anode was
0.30 mm
in the absence of magnetic field, and 0.35
with the x-ray tube in the bore of the magnet. The
equivalent diameters of the focal spot were 0.55 and
0.59 mm, respectively. Misalignment by caused
the focal spot to move by 1.75 mm. Visual inspection of
the bar pattern (Fig. 5) indicates that the resolution of
the detector is not affected by the presence of the mag-
netic field. The limiting resolution of 2.2 lp/mm is close
to the Nyguist limit of the detector (2.5 lp/mm).
X-ray fluoroscopy and MR images of the fruit phan-
tom are shown in Fig. 6. On the x-ray image, which is a
single frame from the fluoro run, the size and location of
the 22-gauge needle are clearly delineated. The RF coil
and one of its capacitors can also be seen. The MR
images show impressive soft-tissue contrast. The metal
artifact in the MR images causes the needle to appear
wider than in the x-ray images. Not evident from these
images is the difference in imaging times, 30 ms for the
x-ray image and 30 sec for the MR image, although
clearly MR images could be acquired more quickly with
some compromise in image quality.
These first images acquired using our new configura-
tion are extremely promising. Although MR field homo-
geneity is affected by the presence of the x-ray system,
particularly the detector, MR images were acquired that
exhibited insignificant global distortion. This is not sur-
prising since, at worst, the frequency shift caused by
the increase in field inhomogeneity corrresponds to 2–3
pixels. This degree of spatial shift over an image can be
relatively benign when tracking a device in an interven-
tional procedure, but if present over a small region (eg,
1 cm) can be harmful to the device visualization and
localization. Since chemical shift selective pulses are
not reliable on this system due to its field strength and
limited homogeneity, the more important impacts of
degradation in field homogeneity are local distortions
and signal loss in GRE images. With respect to local
distortions, a reasonable homogeneity specification
Figure 4. Three x-ray pinhole images (the same 699 550
pixel subregions of the original 1024 1024 images) of the
x-ray tube focal spot with the x-ray system as follows: a) the
focal spot 260 cm from the isocenter of the magnet, and
the anode-cathode axis aligned with the direction of B
line image); b) the focal spot 70 cm from the isocenter of the
magnet at its proposed location and the anode-cathode axis
aligned with B
; and c) as in b, but with the anode-cathode axis
rotated by 5° relative to B
298 Fahrig et al.
would be a frequency shift of one pixel in a distance of
1 cm. For a 8 kHz readout with 256 points, this con-
verts to a frequency gradient of 62.5 Hz/cm. Another
potential impact is signal loss in GRE images. For a 1
cm thick slice and TE 40 ms, a frequency gradient in
the slice direction of 25 Hz/cm would introduce a phase
difference spanning a full cycle. Thus, limiting the im-
pact of the x-ray components to no more than 25 Hz/cm
frequency gradient seems reasonable. Our current con-
figuration caused a local gradient of 120 Hz/cm near
the detector. However, we believe this degree of impact
on field homogeneity, though certainly undesirable, is
not a fundamental characteristic of the detector panel
but most likely due to particular components, and that
this problem can be solved. The decrease in SNR in MR
images (acquired with the x-ray components off) was
only 12%. Our experiments indicate that the noise is
being conducted into the MR suite by the HV cables. We
are confident that permanent installation of the HV
power supply and cables within an RF shield will re-
duce the RF noise essentially to baseline.
The performance of the x-ray system was completely
satisfactory. Alignment of the tube with the magnetic
field is important but can be achieved. The size of the
x-ray focal spot was well below 1 mm, and the x-ray
detector functioned normally. A review of these experi-
mental results in parallel with the criteria outlined
above shows that our geometry is indeed feasible if the
x-ray system is off during the acquisition of MR images.
The initial investigations presented above do, how-
ever, highlight several technical issues that should be
addressed in order to achieve the highest possible im-
age quality with both MR and x-ray systems. The de-
crease in field homogeneity— due in large part to the
presence of the detector—should be addressed by iden-
tification and replacement of the detector components
causing the inhomogeneity and, if needed, higher-order
shimming. In addition, detailed measurement of the
eddy currents produced in the x-ray tube housing and
in the flat-panel detector must be made. Ideally, the
x-ray detector should be in thermal equilibrium during
a procedure in order to maintain the stability of dark
current and gain characteristics; turning the system off
and on repeatedly might not allow the system to equil-
ibrate. RF shielding of the detector will therefore be
investigated. Further detailed studies of focal spot size
Figure 5. X-ray image (single fluoro frame) of a resolution bar
pattern placed 5 mm above the flat panel detector (74 kVp,
0.03 mAs). The x-ray tube was at the proposed location, and
the x-ray detector was 1 m directly below the focal spot (at
the same height as the proposed location of the detector, but
displaced from the center of the bore by 40 cm). The limiting
resolution seen here of 2.2 lp/mm is close to the sampling
limit of the detector (2.5 lp/mm).
Figure 6. a) Projection x-ray image (single fluoro frame, 50
kVp, 0.01 mAs) acquired with the x-ray tube at the proposed
location within the bore of the magnet, and the x-ray detector
below the patient bed. b) GRE image (TR/TE 150/2.8 ms)
acquired with the x-ray components in their proposed loca-
Hybrid Interventional MR/X-Ray System
and position as a function of x-ray tube orientation are
necessary to increase the flexibility of the image acqui-
sition geometry (eg, other projections). In addition,
when heated with AC currents, the x-ray tube fila-
ment’s life might be limited due to the mechanical vi-
bration produced by the AC power in conjunction with
the magnetic field. An investigation using isolated fila-
ments is underway. Although detector resolution and
distortion cannot be affected by the magnetic field,
careful measurements of noise levels under different
exposure conditions need to be carried out. Component
changes (larger focal spot, increased housing cooling,
more powerful generator) should further expand the
capabilities of the system, allowing short, high expo-
sure DSA image sequences to be acquired. The mount-
ing of tube and detector could be modified to provide
projections ranging from AP to lateral, although this
would require a significant increase in the complexity of
the system. Finally, while these technical projects will
increase our understanding of the system and its per-
formance, the most important next step is the demon-
stration of potential applications in animals and hu-
The truly hybrid system described here is particularly
suited for use during interventional procedures. Inte-
gration of the x-ray system into the bore of the Signa SP
magnet allows the physician to switch back and forth
rapidly between imaging modalities without patient
transport, maintaining maximum flexibility. This is not
possible with other semihybrid systems that place a
fluoroscopy system in the same room as (but some
distance from) a standard long-bore magnet. A recent
workshop on the technical requirements for image-
guided spine procedures concluded that the ideal sys-
tem would combine multiple imaging modalities with a
high degree of integration (8). However, the semihybrid
configuration does have some advantages. There is no
limit on the size and complexity of the x-ray gantry or on
the size of the x-ray detector, thereby allowing arbitrary
projections and patient positioning and large field of
view. In addition, one can use a high power rotating
anode tube. While this does not impact fluoroscopy, it
does enable very short exposure time, high-exposure
angiographic imaging. With a fixed anode tube, lower
noise images and even digital subtraction angiography
(DSA) runs can be acquired but require a longer expo-
sure time per frame, possibly resulting in motion blur-
ring. Nonetheless, we believe the fully integrated, hy-
brid approach is very attractive for guiding a number of
interventional procedures. The utility of an integrated
approach for TIPS and chemoembolization was dis-
cussed above. There are others. For example, during a
craniotomy, soft tissue deformation could be visualized
using MR, while the motion of small vessels is simulta-
neously tracked using the x-ray fluoroscopy system.
During vascular interventions, catheter and device
placement could be guided using x-ray imaging while
MR is used to monitor physiological consequences and
to provide three-dimensional visualization.
In conclusion, we have shown that it is feasible to
fully integrate an x-ray fluoroscopy system within the
bore of an iMRI system, enabling dual modality imaging
without requiring patient repositioning. While further
work is needed to perfect this technology and clinical
applications have not yet been proven, we are optimistic
that this configuration can have an important impact in
image-guided interventions.
The authors gratefully acknowledge support from GE
Medical Systems and Lunar Corporation, and funding
from NIH grant P41 RR09784, the Terry Fox NCIC Pro-
gram Project grant entitled “Imaging for Cancer”, and
the Lucas Foundation. The first author is supported by
the Medical Research Council of Canada. Valuable as-
sistance was provided by G. Glover, S. Germann, J.
Routledge, and G. DeCrescenzo.
1. Busse RF, Kruger DG, Debbins JP, Fain SB, Riederer SJ. A flexible
view ordering technique for high-quality real-time 2DFT MR fluo-
roscopy. Magn Reson Med 1999;42:6981.
2. Kerr AB, Pauly JM, Hu BS, et al. Real-time interactive MRI on a
conventional scanner. Magn Reson Med 1997;38:355–367.
3. Gmitro AF, Ehsani AR, Berchem TA, Snell RJ. A real-time recon-
struction system for magnetic resonance imaging. Magn Reson Med
4. Wright RC, Riederer SJ, Farzaneh F, Rossman PJ, Liu Y. Real-time
MR fluoroscopic data acquisition and image reconstruction. Magn
Reson Med 1989;12:407– 415.
5. Jolesz FA, Morrison PR, Koran SJ, et al. Compatible instrumentation
for intraoperative MRI: Expanding resources. JMRI 1998;8:8 –11.
6. Kee S, Rhee JS, Butts K, et al. MR-guided transjugular portosys-
temic shunt placement in a swine model. JVIR 1999;10:529 –535.
7. Butts K, Pauly JM, Kerr A, Daniel BL, Kee S. Real-time interactive
MR imaging on an open MR system. RSNA EJ 1999;3.
8. Duerk DL, Gro¨nemeyer D, Bascle B, et al. Intraprocedural Imaging
and Endoscopy. In: Technical Requirements for Image-Guided
Spine Procedures Workshop Report. Washington, DC: Imaging Sci-
ence and Information Systems (ISIS) Center Radiology Department,
Georgetown University Medical Center; 1999.
9. Adam G, Neuerburg J, Bucker A, et al. Initial clinical experience
with a 1.5-tesla magnetic resonance system combined with C-arm
fluoroscopy. Invest Radiol 1997;32:191–197.
10. Neuerburg J-M, Adam G, Buecker A, et al. MRI-guided biopsy of
bone in a hybrid system. JMRI 1998;8:85–90.
11. Hall WA, Martin AJ, Haiying L, et al. Brain biopsy using high-field
strength interventional magnetic resonance imaging. Neurosurg
12. Antonuk LE, el-Mohri Y, Siewerdsen JH, et al. Empirical investiga-
tion of the signal performance of a high-resolution, indirect detec-
tion, active matrix flat-panel imager (AMFPI) for fluoroscopic and
radiographic operation. Med Phys 1997;24:51–70.
13. Siewerdsen JH, Antonuk LE, el-Mohri Y, et al. Empirical and the-
oretical investigation of the noise performance of indirect detection,
active matrix flat-panel imagers (AMFPIs) for diagnostic radiology.
Med Phys 1997;24:71– 89.
14. Matsuura N, Zhao W, Huang Z, Rowlands JA. Digital radiology
using active matrix readout: amplified pixel detector array for flu-
oroscopy. Med Phys 1999;26:672–681.
15. Zhao W, Blevis I, Germann S, Rowlands JA, Waechter D, Huang Z.
Digital radiology using active matrix readout of amorphous sele-
nium: construction and evaluation of a prototype real-time detec-
tor. Med Phys 1997;24:1834–1843.
16. Bushberg JT, Siebert JA, Leidholdt Jr. EM, Boone JM. The Essen-
tial Physics of Medical Imaging. Baltimore: Williams & Wilkins;
17. Christensen EE, Curry TS III, Nunnally J. An Introduction to the
Physics of Diagnostic Radiology. Philadelphia: Lea & Sebiger; 1972.
18. Granfors P. Performance characteristics of an amorphous silicon
flat panel x-ray imaging detector. SPIE 1999;3659:480 490.
300 Fahrig et al.
    • "Although challenging, PET/MR systems have also been developed (Marsden et al 2002, Antoch and Bockisch 2009). Progress has also been made in the development of hybrid MRI/x-ray systems (Fahrig et al 2001, Ganguly et al 2005). Multimodality imaging systems can provide important scientific or diagnostic information not readily attainable using two separate imaging systems, and, where possible, the performance of each imaging system is preserved (Cherry 2006). "
    [Show abstract] [Hide abstract] ABSTRACT: Imaging of the microvasculature is often performed using contrast agents in combination with either ultrasound (US) or magnetic resonance (MR) imaging. Contrast agents are used to enhance medical imaging by highlighting microvascular properties and function. Dynamic signal changes arising from the passage of contrast agents through the microvasculature can be used to characterize different pathologies; however, comparisons across modalities are difficult due to differences in the interactions of contrast agents with the microvasculature. Better knowledge of the relationship of contrast enhancement patterns with both modalities could enable better characterization of tissue microvasculature. We developed a co-registration platform for multi-modal US and MR imaging using clinical imaging systems in order to study the relationship between US and MR contrast enhancement. A preliminary validation study was performed in phantoms to determine the registration accuracy of the platform. In phantoms, the in-plane registration accuracy was measured to be 0.2 ± 0.2 and 0.3 ± 0.2 mm, in the lateral and axial directions, respectively. The out-of-plane registration accuracy was estimated to be 0.5 mm ±0.1. Co-registered US and MR imaging was performed in a rabbit model to evaluate contrast kinetics in different tissue types after bolus injections of US and MR contrast agents. The arrival time of the contrast agent in the plane of imaging was relatively similar for both modalities. We studied three different tissue types: muscle, large vessels and fat. In US, the temporal kinetics of signal enhancement were not strongly dependent on tissue type. In MR, however, due to the different amounts of agent extravasation in each tissue type, tissue-specific contrast kinetics were observed. This study demonstrates the feasibility of performing in vivo co-registered contrast US and MR imaging to study the relationships of the enhancement patterns with each modality.
    Full-text · Article · Feb 2011
    • "These are not optimized with respect to radiation transparency, and when irradiated they will cast a shadow in the radiation field. Rieke et al (2005) showed already the feasibility of building x-ray compatible RF coils for their hybrid interventional MR/x-ray system (Fahrig et al 2001). Together with Philips, megavoltage (MV) transparent RF coils will be developed. "
    [Show abstract] [Hide abstract] ABSTRACT: At the UMC Utrecht, The Netherlands, we have constructed a prototype MRI accelerator. The prototype is a modified 6 MV Elekta (Crawley, UK) accelerator next to a modified 1.5 T Philips Achieva (Best, The Netherlands) MRI system. From the initial design onwards, modifications to both systems were aimed to yield simultaneous and unhampered operation of the MRI and the accelerator. Indeed, the simultaneous operation is shown by performing diagnostic quality 1.5 T MRI with the radiation beam on. No degradation of the performance of either system was found. The integrated 1.5 T MRI system and radiotherapy accelerator allow simultaneous irradiation and MR imaging. The full diagnostic imaging capacities of the MRI can be used; dedicated sequences for MRI-guided radiotherapy treatments will be developed. This proof of concept opens the door towards a clinical prototype to start testing MRI-guided radiation therapy (MRIgRT) in the clinic.
    Full-text · Article · Jun 2009
    • "Fahrig and colleagues at Stanford University developed a hybrid X-ray and MR system based on a double-donut MR scanner, with an X-ray source and detector mounted inside the magnet bore [158,159]. Images can be acquired alternately from X-ray and MR without moving the patient, providing images that are automatically co-registered . "
    [Show abstract] [Hide abstract] ABSTRACT: The often touted advantages of MR guidance remain largely unrealized for cardiovascular interventional procedures in patients. Many procedures have been simulated in animal models. We argue these opportunities for clinical interventional MR will be met in the near future. This paper reviews technical and clinical considerations and offers advice on how to implement a clinical-grade interventional cardiovascular MR (iCMR) laboratory. We caution that this reflects our personal view of the "state of the art."
    Full-text · Article · Jan 2009
Show more