IMAGING REPORTS

COMBINED X-RAY AND MAGNETIC RESONANCE

IMAGING FACILITY: APPLICATION TO IMAGE-GUIDED
STEREOTACTIC AND FUNCTIONAL NEUROSURGERY
Stefan Hunsche, Ph.D. OBJECTIVE: To assess the feasibility of a hybrid imaging setup combining x-ray and
Department of Stereotactic and magnetic resonance imaging (MRI) in the setting of both stereotactic and functional
Functional Neurosurgery,
neurosurgery.
University of Cologne,
Cologne, Germany METHODS: A combined x-ray and MRI scanning facility with a trolley system for a fast
patient transfer between both modalities was installed in a neurosurgical setting. A reg-
Dieter Sauner, M.D. istration algorithm for fusion of MRI scans and x-ray images was derived for augmen-
Municipal Clinics of Duisburg, tation of fluoroscopic x-ray projection images with MRI scan data, such as anatomic
Departments of Radiology and
Neuroradiology,
structures and planned probe trajectories. Phantom measurements were obtained
Duisburg, Germany between both modalities for estimation of registration accuracy. Practical application
of our system in stereotactic and functional neurosurgery was tested in brachytherapy,
Mohammad Maarouf, M.D. deep brain stimulation, and motor cortex stimulation.
Department of Stereotactic and RESULTS: Phantom measurements yielded a mean spatial deviation of 0.7 ⫾ 0.3 mm
Functional Neurosurgery,
University of Cologne, with a maximum deviation of 1.1 mm for MRI scans versus x-rays. Augmentation of
Cologne, Germany x-ray images with MRI scan data allowed intraoperative verification of the planned tra-
jectory and target in three types of neurosurgical procedures: positioning iodine seeds
Klaus Lackner, M.D. in brachytherapy in one case with cerebellar metastasis, placement of electrodes for
Department of Radiology, deep brain stimulation in two cases of advanced Parkinson’s disease, and placement
University of Cologne,
Cologne, Germany
of an epidural grid for motor cortex stimulation in two cases of intractable pain.
CONCLUSION: Combined x-ray and MRI-guided stereotactic and functional neuro-
Volker Sturm, M.D. surgery is feasible. Augmentation of x-ray projection images with MRI scan data, such
Department of Stereotactic and as planned probe trajectories and MRI scan segmented anatomic structures may be
Functional Neurosurgery, beneficial for probe guidance in stereotactic and functional neurosurgery.
University of Cologne,
Cologne, Germany KEY WORDS: Image fusion, Magnetic resonance imaging, Stereotactic neurosurgery, X-ray

Harald Treuer, Ph.D. Neurosurgery 60[ONS Suppl 2]:ONS-352–ONS-361, 2007 DOI: 10.1227/01.NEU.0000255423.24173.42

Department of Stereotactic and

Functional Neurosurgery,
University of Cologne,

C
Cologne, Germany ombined x-ray and magnetic resonance Usually, x-ray imaging is used for the guid-
imaging (MRI) facilities (XMR) are of ance of the surgical instruments and monitor-
Reprint requests: great interest in the field of intraopera- ing of the vascular therapy, whereas MRI scan-
Stefan Hunsche, Ph.D., tive imaging (1, 4, 25). The basic premise is to ning is used for the imaging of anatomic
Department of Stereotactic and combine x-ray fluoroscopy functionality with structures and monitoring therapeutic conse-
Functional Neurosurgery,
Kerpener Straße 62,
soft tissue contrast MRI scanning. Both imag- quences for the tissue. Although MRI scanning
50937 Cologne, Germany. ing modalities can be combined in a truly can be used for instrumental guidance, safety
Email: stefan.hunsche@uk-koeln.de hybrid system, whereby the x-ray system is issues must be taken into account.
integrated into the MRI scanner or operate as A technical challenge for an XMR facility is
Received, July 28, 2005. separate but related MRI scanning and C-arm the ability of the system to integrate informa-
Accepted, November 1, 2006. devices with a mobile table for a fast patient tion from either imaging modality into the
transfer. In either case, switching from the monitoring display of the other device. For
imaging modality x-ray to MRI scans and vice example, the “anatomic background” of MRI
versa is easily accomplished. Feasibility has scanning could be added to fluoroscopic
already been shown in interventional radiol- images, thus setting surgical instruments in a
ogy and neurosurgery (3, 5, 6, 13, 18, 26, 37). tissue context. For such a system, an accurate

ONS-352 | VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 www.neurosurgery-online.com

 COMBINED X-RAY AND MAGNETIC RESONANCE IMAGING-GUIDED NEUROSURGERY

registration between both modalities would be of utmost Before estimating the transformation between x-ray projec-
importance. tion and stereotactic coordinates, x-ray images were corrected
We present a hybrid imaging facility for neurosurgery, com- for geometric distortion induced by the image intensifier (28).
bining high-field 1.5-T MRI scanning and fluoroscopic x-ray For this purpose, we imaged a regular calibration grid of 400
projection imaging in a separate device installation with a cus- steel bullets with a diameter of 3 mm equally spaced within
tom-made trolley system for patient transfer. A registration 1 cm. The bullets were segmented in the projection images
algorithm with fast and high precision fusion ability for x-rays automatically by cluster analysis (2), and an “unwarping” poly-
and MRI scans was developed for stereotactic neurosurgery, in nomial was estimated.
which probes are introduced in the brain through a small burr For estimation of the transformation between stereotactic and
hole (32). Registration accuracy was estimated with phantom x-ray projection coordinates, a point-based approach was used.
measurements. The neurosurgeon can make use of MRI scans An x-ray localizer consisting of steel wire squares of 0.3-mm
showing anatomic structures and MRI scan-planned trajecto- diameter (Pastyr, Heidelberg, Germany) and known stereotac-
ries in the x-ray display with high spatial reliability and at any tic coordinates (60-mm square length) were attached to the
time during the surgery. Practical feasibility of such MRI scan- right, left, front, and back of the frame. By manual segmentation
augmented x-ray-based probe guidance is demonstrated for of the localizer in two-dimensional-projection x-ray images, we
stereotactic and functional neurosurgery. estimated the transformation by multiresolution optimization in
a least-square approach. We assumed a perspective projection
with a projection vector perpendicular to the image plane. First,
MATERIALS AND METHODS a transformation MSTX⫺Carm between the stereotactic and the
C-arm device coordinate system was derived. The origin of the
XMR Suite C-arm system was defined by the focus of the system. The
Our XMR-facility consists of two neighboring rooms image plane was assumed to be perpendicular to the x axis.
divided by a radiofrequency-shielded sliding door. For steril- The y and z axes were then defined as being parallel to the bor-
ity, both rooms are ventilated under positive air pressure. One ders of the image plane. Thus, a rigid body transformation
radiofrequency-shielded room contains the short-bore whole- between both coordinate systems was assumed:
body MRI scanner operating at 1.5 T with an in-room opera-

冤 冥冤 冥冤 冥
cos α sin α 0 0 cos β 0 – sin β 0 1 0 0 0
tor’s console (Gyroscan Intera; Philips Medical Systems, Best,
– sin α cos α 0 0 0 1 0 0 0 cos γ sin γ 0
The Netherlands). The adjacent room is a stereotactic operat- MSTX–Carm = . . .
0 0 1 0 sin β 0 cos β 0 0 – sin γ cos γ 0
ing theater with a mobile x-ray C-arm unit (Exposcop CB7-D;
0 0 0 1 0 0 0 1 0 0 0 1
Ziehm, Nuremberg, Germany). Either room can be used fully

冤 冥
independently as an MRI scanning system with its own control 1 0 0 – xF
room or as a stereotactic operation room. For combined use, a 0 1 0 – yF
custom-made trolley system with a MRI scan-compatible and 0 0 1 – zF
x-ray-transparent floating tabletop serves for the docking and 0 0 0 1
delivery of patients.
with α, β, γ, xF, yF, and zF as unknown angulations and trans-
Calibration and Registration of XMR lation parameters. For estimation of transformation from
C-arm device to x-ray image coordinates, we assumed a
The reference system for registration of both modalities and
perspective projection Px with a projection vector perpendicu-
the patient’s head is a stereotactic MRI scan-compatible ceramic
lar to the image plane. Thus, the transformation from stereotac-
frame, which is fixed to the patient’s head (MRC, Heidelberg,
tic coordinates to x-ray image coordinates was MSTX–Xray =
Germany). The frame defines the stereotactic coordinate system
PX . MSTX–Carm with:

冤 冥
within the head, a Cartesian right-handed coordinate system 1 0 0 0
with its origin in the center of the frame. 0 1 0 0
Px =
For each modality, a separate localizer is mounted on the 0 0 1 0
frame for estimation of image-to-stereotactic coordinate trans- 1
— 0 0 0
–λ
formation. The combination of both transformations (MRI scan
to stereotactic coordinates and stereotactic to x-ray-image coor- The image plane-to-focus distance was assumed as known. A
dinates) results in an intermodality registration. frame grabber card (FlashBus V; Integral Technologies, Inc.,
For estimation of transformation between MRI scan and Indianapolis, IN) was used to digitize the x-ray projection
stereotactic coordinates, a V-shaped localizer device (Leibinger, images in real-time.
Freiburg, Germany), consisting of tubes filled with a mixture of
water and the paramagnetic contrast medium, gadolinium Estimation of Registration Accuracy
diethylenetriamine penta-acetic acid (Magnevist; Schering, For estimation of registration accuracy between both modali-
Berlin, Germany), was used during MRI scanning. The trans- ties, a custom-made XMR phantom was build out of a Perspex
formation was estimated with stereotactic neurosurgery guid- cube filled with water and 0.2-mm thick steel wires. The wires
ance software (STP3; Stryker Leibinger, Freiburg, Germany). were visible in both modalities by the attenuation effect in x-ray

NEUROSURGERY VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 | ONS-353

HUNSCHE ET AL.

and by signal void in MRI scanning. The junctures of the wire images, the MRI scan-planned probe trajectory was projected
were defined as fiducial markers (38). The wires were attached into the x-ray images with the transformation estimated in the
in a diagonal fashion that allowed estimating the stereotactic registration procedure, allowing on-line control for instanta-
coordinates with two x-ray projection images by calculating the neous detection of arising differences between the planned and
intersection of x-rays (29). For estimating MRI scanning stereo- real trajectory of the probe.
tactic coordinates, 70 transverse turbo spin-echo T2-weighted
slices were imaged. The imaging parameters were: repetition
time, 2000 milliseconds; echo time, 120 milliseconds; turbo factor,
A B
19; slice thickness, 2 mm; matrix size, 512 ⫻ 512; field of view,
290 mm; and bandwidth, 120 Hz per pixel. The stereotactic coor-
dinates of the 12 fiducial markers were defined manually in both
modalities; the mean and maximum deviations in each direction,
dx, dy, and dz, as well as the spatial deviation were estimated.
Patients
Five patients underwent different stereotactic neurosurgical
operations with XMR guidance. One patient underwent a
brachytherapy, two patients underwent deep brain stimulation
(DBS), and two patients underwent epidural motor cortex stim- FIGURE 1. Phantom images. A, lateral x-ray projection image. The white
ulation (MCS). All patients gave written informed consent arrow indicates a fiducial (crossing wires). B, resliced T2-weighted turbo
before the imaging was completed. The institutional review spin-echo MRI scan parallel to one wire. The white arrow indicates a fidu-
cial (crossing wires).
board of the university approved the study.
MRI Scanning for Treatment Planning
Imaging for treatment planning for different types of neurosur- TABLE 1. Phantom measurementsa
gical procedures is performed according to a collective imaging Deviation Mean (mm) Maximum (mm)
protocol. First, we fixed the patient’s head to the stereotactic MRI dx –0.4 ⫾ 0.3 –0.9
scan-compatible ceramic frame (32). This frame is fixed to the dy –0.1 ⫾ 0.3 –0.5
floating tabletop, which serves as the operating table. After fixa- dz 0.3 ⫾ 0.2 0.7
tion, MRI scanning was performed for treatment planning. We dr 0.7 ⫾ 0.3 1.1
used T2-weighted spin-echo imaging mainly for delineation of
target structures (repetition time, 2000 ms; echo time, 120 ms; a
Mean ⫾ standard deviation and maximum deviation of x-ray versus magnetic
turbo factor, 19; slice thickness, 2 mm; matrix size, 512 ⫻ 512; field resonance imaging coordinates (n ⫽ 12); Dr, spatial distance.

of view, 290 mm; and bandwidth, 120 Hz/pixel). Paramagnetic

contrast medium (Magnevist) was injected via an antecubital vein A B
before T1-weighted gradient-echo imaging, which was used
mainly for vessel detection and imaging of contrast uptake (echo
time, 15 ms; repetition time, 30 ms; flip angle, 40 degrees; slice
thickness, 2 mm; matrix size, 512 ⫻ 512; field of view, 290 mm;
and bandwidth, 120 Hz/pixel). For somatotopic mapping of the
motor cortex for MCS, blood oxygenation level-dependent
(BOLD) imaging was performed as previously described (12).

Treatment Planning C D
Probe positioning was planned on MRI scans with STP3
stereotactic guidance software. The transformation between
MRI and stereotactic coordinates allows planning of the probe
position in stereotactic coordinates (11).

Image-guided Stereotactic Neurosurgery

After simulation with a phantom, the operator inserted the
probe into the brain under calibrated stereotactic-based x-ray
control. Except for MCS, the insertion of the probe was sup- FIGURE 2. XMR facility. A, trolley system and the entrance to the MRI
ported by the MRI scan-safe Riechert-Mundinger aiming bow scanner. B, x-ray unit with the calibration grid. C, MRI scanner setup. D,
(32), which is attached to the frame and fits within the MRI stereotactic operating theater with a mobile x-ray C-arm unit. The x-ray
scanner bore. The aiming bow uses input coordinates from the images are projected onto the right monitor and the augmented x-ray images
stereotactic guidance software. For augmentation of the x-ray are projected onto the left monitor.

ONS-354 | VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 www.neurosurgery-online.com

 COMBINED X-RAY AND MAGNETIC RESONANCE IMAGING-GUIDED NEUROSURGERY

and before implantation of the

A B second electrode for DBS to
check for brain deformation.
We closed the burr hole to stop
cerebrospinal fluid loss and
possible brain deformation
during the transition from
x-ray imaging to MRI scan-
ning. For visualization of brain
deformation, we applied a gra-
dient and maximum filter (31)
to the preoperative images and
overlaid the resulting contour
images onto the intraoperative
MRI scans consecutively.

RESULTS

C D Estimation of
Registration Accuracy
Figure 1 demonstrates the
phantom measurements for
the two different modalities.
For x-rays versus MRI scans,
the mean spatial deviation was
0.7 ⫾ 0.3 mm with a maximum
spatial deviation of 1.1 mm
(Table 1). With the exception of
the mean dy, there was a sig-
nificant difference of mean dx
and dz from zero (P ⱕ 0.05).

General Experience
Figure 2 shows the XMR
facility. The trolley transfer of
FIGURE 3. Planning on contrast-enhanced fast field echo T1-weighted sagittal (A) and transverse images (B) for the patient from one modality
brachytherapy with two seeds in one catheter (small black bars). C and D, x-ray guidance with lateral projection. The to the other required approx-
gray bars are the seeds. The white bars (overlaid on the gray bars) show the planned positions of the seeds. The real
imately 5 to 10 minutes, in-
and planned seed positions match well.
cluding the mounting of the
different localizer device. The
In MCS, additional augmentation of x-ray images with MRI segmentation of the localizer and the estimation of the trans-
scan information was performed by projecting the MRI scan formation between both modalities required approximately 1
segmented and rendered cortex (Visualization Toolkit 4.2; to 2 minutes. The whole operation lasted approximately 4 to
Kitware Inc., Clifton Park, NY) onto the stereotactic x-ray 6 hours. None of the patients experienced early postoperative
images. This enables orientation during the freehand posi- complications.
tioning of the MCS electrode (DBS Model 3587A; Medtronic,
Minneapolis, MN) through a small burr hole. Because of the Brachytherapy
lack of a genuine guidance system for the MCS electrode, Planning and XMR-guidance for iodine brachytherapy were
only the trepanation was supported by the mechanical guid- performed for the treatment of cerebellar metastasis in one
ance system. patient (Fig. 3). Two seeds (IMC 6711; GE Healthcare, Arlington
For augmentation of the x-ray images during DBS (DBS Heights, IL) with a size of 4.5 mm and a diameter of 0.8 mm
Model 3389; Medtronic), we fused the MRI scan segmented were stereotactically implanted in one catheter. Over the
and rendered target volume (30) and the subthalamic nucleus course of 42 days, a cumulative irradiation dose of 50 Gy (sur-
(STN) with the x-ray images. Additionally, intraoperative MRI face dose) was applied to the tumor. The stereotactic x-ray
scanning was performed after implantation of one electrode guidance was augmented with MRI scan-planned seed posi-

NEUROSURGERY VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 | ONS-355

HUNSCHE ET AL.

Epidural MCS
A B
Epidural MCS was per-
formed with XMR guidance in
two patients. One patient who
lost his left arm 20 years ago
developed an intractable phan-
tom pain with pain projection
mainly to the upper arm. MCS
was planned for the right hemi-
sphere, where functional activa-
tion indicated his imagination
of finger movements on the left
body side (Fig. 6). Figure 7
shows the stereotactic-based
x-ray guidance of the electrode
through a small burr hole in the
cranium. For augmentation of
C D x-ray images, the MRI scan-
planned trajectory of the elec-
trode and the MRI scan vol-
ume-rendered cortex were
fused with the x-ray images.
The cortical projection depicted
the motor cortex and allowed
an optimized electrode position
online on the basis of the clini-
cal stimulation results. The
patient reported a subjective
pain reduction of 75% intra-
and postoperatively.
The second patient devel-
oped intractable facial pain
after dissection of the internal
FIGURE 4. STN stimulation planning on turbo spin-echo transverse slices (A) and x-ray probe guidance (B–D). The carotid artery. Functional acti-
MRI scan-segmented STN is projected onto the x-ray images (lateral projection). The approach of the probe to the pro- vation during mouth move-
jected target is shown in B and C. D, the planned trajectory and the real position of the probe match properly. ment was used for localizing
the motor cortex. Two days
tions by fusing them with the x-ray images, allowing immedi- after XMR-guided MCS, the patient reported a subjective pain
ate detection of arising differences between real and planned reduction of approximately 90%.
positions of the seeds.
DISCUSSION
DBS
In two patients with advanced Parkinson’s disease, STN We presented a separate XMR imaging facility with a regis-
stimulation was performed with XMR guidance. The target tration algorithm for stereotactic neurosurgery, allowing probe
point of the STN stimulation electrode was planned on guidance by augmentation of stereotactic-based x-ray projec-
T2-weighted images, which sufficiently show the borders of tion images with MRI scan information.
the STN (Fig. 4). The intended trajectory and the T2-weighted
segmented STN were fused with the calibrated stereotactic- Estimation of Registration Accuracy
based x-ray images, allowing probe guidance and correlation The phantom registration measurements showed an accuracy
of clinical stimulation results at distinct electrode positions of 0.7 mm, which is approximately the limitation of accuracy
relative to the STN. The overlay of gradient and maximum fil- produced by the discrete nature of the MRI scan (36), thus indi-
tered preoperative images on intraoperative images after cating negligible systematic errors. Significant differences in the
implantation of one electrode indicated no brain deformation mean δx and δz may be caused by the sensitivity of frequency
on the contralateral side and in the target region of the first and slice encoding directions to geometric distortions (16). Our
electrode (Fig. 5). measurements of registration accuracy did not take object-

ONS-356 | VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 www.neurosurgery-online.com

 COMBINED X-RAY AND MAGNETIC RESONANCE IMAGING-GUIDED NEUROSURGERY

tion algorithms or offered registration accuracies that were not

A B suitable for stereotactic neurosurgical purposes (25). The main
difference in our algorithm is in its use of optical tracking for reg-
istration of both modalities with the inherent errors and obstacles
of any additional measurement procedure. Because we only had
to localize the stereotactic frame in both modalities, our registra-
tion procedure yielded a higher precision than optical tracking.

General Experiences
For selected patients, we chose three different surgical sce-
narios, brachytherapy and two stimulation settings, DBS and
MCS, to demonstrate the practical feasibility and clinical poten-
tial of XMR-guided stereotactic neurosurgery.
Our XMR-system comprises two adjacent rooms and allows
C D the independent use of both rooms as either an MRI scanning
site or an x-ray-equipped operating room. Patient transfer from
one to the other modality required only 5 to 10 minutes, which
is fast enough for stereotactic neurosurgery. In general, a
modality switch was intended only from initial MRI scanning
for operation planning to x-ray probe guidance during the
actual neurosurgery. However, switching back to MRI scan-
ning without losing registration is possible if there is a need to
check for brain deformation or intracranial bleeding. For DBS,
we used this option after implantation of the first electrode
and before implantation of the second electrode to check for
possible brain deformation. We used this possibility for DBS be-
cause the clinical testing and electrophysiological measure-
FIGURE 5. A and B, preoperative contrast-enhanced, fast field echo, T1- ments necessary for electrode implantation is time-consuming
weighted transverse MRI scans. C and D, intraoperative, fast field echo, T1- and carries a substantial risk for brain deformation.
weighted transverse MRI scans after implantation of an electrode. Gradient Five to 10 minutes for patient transfer is a long time compared
and maximum filter-corrected preoperative images (white contours) are over- with other systems (3, 18, 26), but our trolley system offers the
laid onto the intraoperative images. No brain deformation was detectable. possibility to switch easily to further modalities, such as com-
Note the signal void from the electrode (white arrows).
puted or positron emission tomography. An easy switch to other
operation areas, such as stereotactic radiosurgery, is also possible.
A B The actual surgery was not performed in the magnet for dif-
ferent reasons. A narrow burr hole and the noise lead to low
surgical comfort. Furthermore, many safety issues must be
taken into account regarding heating and “missile” effects (7, 9,
14, 23, 24). To reduce heating risk for DBS electrodes, a very low
head-specific absorption rate level (⬍0.1 W/kg) is recom-
mended by the manufacturer, thus resulting in poor image
quality. In addition, probe insertion under MRI scan guidance
is not straightforward because of “blooming” artifacts of the
probe (15, 19). High localization accuracy is desirable in stereo-
tactic neurosurgery. Estimating the real position of the probe
FIGURE 6. Planning of MCS for Patient 1 with intractable left arm phantom with MRI scanning is difficult compared with x-ray imaging
pain on transverse (A) and lateral (parallel to probe) (B) T1-weighted fast field with its high contrast for metallic devices. Depending on the
echo images. Activation from BOLD imaging is overlaid onto the MRI scans. measurement protocol and geometry, the MRI scan artifact of
the probe can be different. Furthermore, the “blooming” arti-
dependent geometric MRI scan distortions into account, but it is facts of the probe can mask the target region in DBS (17). For
conceivable that reduced accuracy has to be expected at the cor- these reasons, augmentation of x-ray guidance with MRI scan
tex or near the base of the brain because of susceptibility arti- data such as probe trajectories and MRI scan-segmented struc-
facts (16). However, for MCS, such distortion should be negligi- tures may be an alternative to MRI scan guidance alone for
ble compared with the extensive effect of the electrode. burr hole neurosurgery (8, 17, 35).
To our knowledge, previously presented XMR facilities with a One problem of our approach is brain deformation (10),
separate device design had either no implementation of registra- which can occur during the interval between MRI scanning and

NEUROSURGERY VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 | ONS-357

HUNSCHE ET AL.

dose distribution. The main

A B task of the surgeon is the posi-
tioning of several radioactive
iodine seeds in such a way that
the above criteria are fulfilled.
The seeds are positioned with
the help of catheters, which are
fixed at the cranium.
In conventional surgery
without XMR after position-
ing, the true positions of the
seeds are determined with
film x-ray imaging (33). The
x-ray coordinates are then
imported into the planning
program to obtain the real
C D dose distribution. If the dis-
tribution is satisfactory, the
operation ends; otherwise re-
positioning takes place, with
repeated position control,
until the dosage distribution is
acceptable. Treuer et al. (34)
determined that target-point
deviations larger than 1.5 mm
have a considerable influence
on surface dose and conform-
ity and, thus, seed positioning
should have an accuracy of
less than 1.5 mm.
FIGURE 7. Guidance of an MCS electrode. A, the electrode immediately after insertion through the burr hole with respect For our brachytherapy pa-
to its planned position (white bars). B and D, the planned electrode position (white bars) and the MRI scan- tient, x-ray guidance for the
segmented cortex are fused into the lateral x-ray images. C and D, the final position of the electrode at the postcentral gyrus. seed positioning was useful
for two reasons. First, the
x-ray guidance and may invalidate the registration between the online control of catheter trajectories reduced the risk for
x-ray and MRI scan. To overcome this problem, especially in harming vessels with consecutive bleeding. Secondly, it was
DBS, an update of the MRI scan data is possible at any time dur- possible to assess the current dose distribution online by
ing the operation without losing registration. Brain deformation importing the x-ray data into the planning program, thus,
is not expected for epidural MCS without incision of the dura enabling the neurosurgeon to stop the positioning of the
mater or for a brachytherapy case with implantation of only catheters immediately when dose coverage was satisfying.
one catheter in a short time. The burr hole and especially the Thus, the dose burden of the surrounding healthy brain tissue
incision of the dura mater must be as small as possible to min- and operation time were minimized.
imize the loss of cerebrospinal fluid and, thus, the risk of brain
deformation (10). A current study with our XMR installation DBS
and a significant number of patients is underway to investigate For DBS, the STN was well demarcated as a target structure
the need to check for brain deformation and the necessity to in T2-weighted spin-echo images (30), and delineation of elo-
update the MRI scan data in stereotactic neurosurgery. We can- quent brain regions and vessels was easily performed in
not recommend our XMR installation for conventional brain T1-weighted gradient echo images. By projecting the trajectory
tumor surgery. Because of brain deformation, intraoperative of the electrode into the x-ray view, probe guidance was effec-
MRI scan alone seem to be better for that purpose (20). tively possible. The MRI scan plus x-ray fusion was a useful
feature because of the limited rigidity of the electrode, again
Brachytherapy reducing the risk for harming vessels with consecutive bleed-
In brachytherapy, contrast-enhanced T1-weighted sequences ing. Projecting the MRI scan-segmented STN into the x-ray
are usually used for delineation of the tumor. Depending on view enabled an online correlation between the spatial rela-
tumor pathology, the neurosurgeon aims to deliver the tumor tionship of the STN and the electrode and the clinical stimula-
surface dose in a certain time window with an optimized local tion results during the operation. In the future, this online

ONS-358 | VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 www.neurosurgery-online.com

 COMBINED X-RAY AND MAGNETIC RESONANCE IMAGING-GUIDED NEUROSURGERY

feedback may help to elucidate the electrophysiological fine- made trolley system and an implementation of algorithms for
tuning in relation to the electrode position for further opti- registration optimization allowing augmentation of x-ray
mized stereotactic position planning. images with MRI scan information. Usually, data input for
After implantation of one electrode, changing the modality to stereotactic frame-based neurosurgery consists of plain coordi-
MRI scanning allowed us to check for brain deformation in the nate information regarding the start and the target region of a
target region, one way of validating the target point estimation neurosurgical intervention without any possibility for visuali-
by MRI scanning. Checking of brain deformation is also essen- zation of trajectories or anatomic information during the actual
tial in cases of implantation of a second electrode on the con- neurosurgery. Our system differs from existent systems in that
tralateral side. In DBS, brain deformation could pose a problem it delivers such image information by overlaying MRI scans
because of the CSF loss during the time-consuming electro- and information regarding the surgical planning onto the x-ray
physiological measurements. images. Such x-ray-based probe guidance enables the neurosur-
geon to place electrodes freehand, without a guidance system,
Epidural MCS as we demonstrated for MCS.
A third promising application of XMR for stereotactic neuro- Exemplary patient cases with either brachytherapy or stim-
surgery is MCS. The position at which the epidural stimulation ulative neurosurgical interventions demonstrate the feasibility
was most effective in pain reduction corresponded to the indi- of registered XMR for stereotactic and functional neurosurgery.
vidual somatotopic representation of the somatomotor cortical In our opinion, stereotactic neurosurgery may benefit from
projection (21) of the body area in question, which was identi- XMR guidance.
fied by BOLD imaging. Although BOLD imaging is not
unproblematic for somatotopic mapping because of its depend- REFERENCES
ency on the vascular coupling, which may change in patholog-
1. Adam G, Neuerburg J, Bücker A, Glowinski A, Vorwerk D, Stargardt A, Van
ical altered tissue (27), our clinical results seem to confirm our Vaals JJ, Günther RW: Interventional magnetic resonance. Initial clinical expe-
approach for targeting eloquent brain regions. rience with a 1.5-tesla magnetic resonance system combined with c-arm flu-
By projecting the MRI scan-planned position of the electrode oroscopy. Invest Radiol 32:191–197, 1997.
into the x-ray view, x-ray imaging allowed us an exact electrode 2. Cho PS, Johnson RH: Automated detection of BB pixel clusters in digital flu-
oroscopic images. Phys Med Biol 43:2677–2683, 1998.
positioning. Interestingly, our current installation with only
3. Dick AJ, Raman VK, Raval AN, Guttman MA, Thompson RB, Ozturk C,
one projection for x-ray view posed no problem at all because Peters DC, Stine AM, Wright VJ, Schenke WH, Ledermann RJ: Invasive
of the epidural position of the electrode on a surface with only human magnetic resonance imaging: Feasibility during revascularisation in a
two degrees of freedom. Furthermore, the projection of the MRI combined XMR suite. Catheter Cardiovasc Interv 64:265–274, 2005.
scan-rendered cortex into the x-ray view allowed an online cor- 4. Fahrig R, Butts K, Wen Z, Saunders R, Kee ST, Sze DY, Daniel BL, Laerum F,
Pelc NJ: Truly hybrid interventional MR/X-ray system: Investigation of in
rection of the electrode position through a small burr hole ori- vivo applications. Acad Radiol 8:1200–1207, 2001.
entated on anatomic position and clinical stimulation results. 5. Fahrig R, Heit G, Wen Z, Daniel BL, Butts K, Pelc NJ: First use of a truly
As in the other applications, a craniotomy risking infection and hybrid X-ray/MR imaging system for guidance of brain biopsy. Acta
bleeding could be circumvented. Neurochir (Wien) 145:995–997, 2003.
6. Ganguly A, Wen Z, Daniel BL, Butts K, Kee ST, Rieke V, Do HM, Pelc NJ,
Fahrig R: Truly hybrid X-ray/MR imaging: Toward a streamlined clinical
Current Technical Limitations system. Acad Radiol 12:1167–1177, 2005.
C-arm guidance is limited to one x-ray view at a time. For 7. Hall WA, Liu H, Martin AJ, Pozza CH, Maxwell RE, Truwit CL: Safety, effi-
full coordinate determination, the C-arm has to be rotated and cacy, and functionality of high-field strength interventional magnetic reso-
nance imaging for neurosurgery. Neurosurgery 46:632–642, 2000.
calibrated, which is time-consuming. For the next generation of 8. Hall WA, Liu H, Truwit CL: Navigus trajectory guide. Neurosurgery
our XMR unit, we plan to replace the C-arm with two flat panel 46:502–504, 2000.
detectors (22) and to replace the two x-ray tubes with one tube 9. Henderson JM, Tkach J, Phillips M, Baker K, Shellock FG, Rezai AR:
mounted at the ceiling and the other at the wall. The detectors Permanent neurological deficit related to magnetic resonance imaging in a
patient with implanted deep brain stimulation electrodes for parkinson’s dis-
will be mounted in a fixed geometric relation to the stereotac-
ease: Case report. Neurosurgery 57:E1063, 2005.
tic frame, allowing a left-to-right and anterior-to-posterior x-ray 10. Hill DL, Maurer CR Jr, Maciunas RJ, Barwise JA, Fitzpatrick JM, Wang MY:
view, thus defining a probe position in all three coordinates. Measurement of intraoperative brain surface deformation under a cran-
Probe guidance should then be possible in all three directions, iotomy. Neurosurgery 43:514–528, 1998.
a feature that is already implemented in the current software 11. Hunsche S, Sauner D, Maarouf M, Hoevels M, Luyken K, Schulte O, Lackner
K, Sturm V, Treuer H: MR-guided stereotactic neurosurgery—Comparison of
release. Calibration has to be performed only one time before fiducial-based and anatomical landmark transformation approaches. Phys
neurosurgery. One further advantage of this system is the “flat- Med Biol 49:2705–2716, 2004.
ness” of the detectors, which make “unwarping” for geometric 12. Hunsche S, Sauner D, Treuer H, Hoevels M, Hesselmann V, Schulte O,
distortion obsolete. Lackner K, Volker S: Optimized distortion correction of epi-based statistical
parametrical maps for stereotactic neurosurgery. Magn Reson Imaging
22:163–170, 2004.
CONCLUSION 13. Kee ST, Ganguly A, Daniel BL, Wen Z, Butts K, Shimikawa A, Pelc NJ, Fahrig
R, Dake MD: MR-guided transjugular intrahepatic portosystemic shunt cre-
We present a complete XMR installation with technical inno- ation with use of a hybrid radiography/MR system. J Vasc Interv Radiol
vations on the hardware and the software sides; i.e., a custom- 16:227–234, 2005.

NEUROSURGERY VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 | ONS-359

HUNSCHE ET AL.

14. Kettenbach J, Kacher DF, Kanan AR, Rostenberg B, Fairhurst J, Stadler A, 35. Truwit CL, Hall WA: Intraoperative magnetic resonance imaging-guided neu-
Kienreich K, Jolesz FA: Intraoperative and interventional MRI: rosurgery at 3-T. Neurosurgery 58 [Suppl 2]:ONS338–ONS346, 2006.
Recommendations for a safe environment. Minim Invasive Ther Allied 36. Wang MY, Maurer CR Jr, Fitzpatrick JM, Maciunas RJ: An automatic tech-
Technol 15:53–64, 2006. nique for finding and localizing externally attached markers in CT and MR
15. Liu H, Hall WA, Martin AJ, Truwit CL: Biopsy needle tip artifact in MR- volume images of the head. IEEE Trans Biomed Eng 43:627–637, 1996.
guided neurosurgery. J Magn Reson Imaging 13:16–22, 2001. 37. Wilson MW, Fidelman N, Weber OM, Martin AJ, Gordon RL, LaBerge JM,
16. Lüdeke KM, Roschmann P, Tischler R: Susceptibility artifacts in NMR imag- Kerlan RK, Wolanske KA, Saeed M: Experimental renal artery embolization
ing. Magn Reson Imaging 3:329–343, 1985. in a combined MR imaging/angiography unit. J Vasc Interv Radiol
17. Martin AJ, Larson PS, Ostrem JL, Sootsman WK, Talke P, Weber OM, 14:1169–1175, 2003.
Levesque N, Myers J, Starr PA: Placement of deep brain stimulator electrodes 38. Yu H, Fahrig R, Pelc NJ: Co-registration of x-ray and MR fields of view in a
using real-time high-field interventional magnetic resonance imaging. Magn hybrid XMR system. J Magn Reson Imaging 22:291–301, 2005.
Reson Med 54:1107–1114, 2005.
18. Martin AJ, Saloner DA, Roberts TP, Roberts H, Weber OM, Dillon W, Cullen Acknowledgment
S, Halbach V, Dowd CF, Higashida RT: Carotid stent delivery in an XMR
We thank the staff of the Department of Radiology for performing the MRI
suite: Immediate assessment of the physiologic impact of extracranial revas-
scans. We received no financial support in conjunction with the generation of
cularization. AJNR Am J Neuroradiol 26:531–537, 2005.
this article.
19. Müller-Bierl B, Graf H, Lauer U, Steidle G, Schick F: Numerical modeling of
needle tip artifacts in MR gradient echo imaging. Med Phys 31:579–587, 2004.
20. Nimsky C, Ganslandt O, Von Keller B, Romstock J, Fahlbusch R: COMMENTS
Intraoperative high-field strength MR imaging: Implementation and experi-
ence in 200 patients. Radiology 233:67–78, 2004.
21. Nguyen JP, Lefaucheur JP, Decq P, Uchiyama T, Carpentier A, Fontaine D,
H unsche et al. report their contribution to the implementation of
hybrid x-ray and magnetic resonance imaging (MRI) technology in
stereotactic and functional neurosurgery. Five patients were treated:
Brugieres P, Pollin B, Feve A, Rostaing S, Cesaro P, Keravel Y: Chronic motor
cortex stimulation in the treatment of central and neuropathic pain. one patient received iodine brachytherapy for cerebellar metastasis,
Correlations between clinical, electrophysiological and anatomical data. Pain two patients received subthalamic nucleus (STN) deep brain stimula-
82:245–251, 1999. tion (DBS) for treatment of Parkinson’s disease, and two patients
22. Pang G, Rowlands JA: Development of high quantum efficiency flat panel underwent motor cortex stimulation for intractable pain.
detectors for portal imaging: Intrinsic spatial resolution. Med Phys The neurosurgical application of this concept was initially reported
29:2274–2285, 2002. in 2003 (1). The current report is unique because the authors developed
23. Rezai AR, Finelli D, Nyenhuis JA, Hrdlicka G, Tkach J, Sharan A, Rugieri P, a registration algorithm for fusion of magnetic resonance and x-ray
Stypulkowski PH, Shellock FG: Neurostimulation systems for deep brain images, applying the technology in a variety of neurosurgical opera-
stimulation: In vitro evaluation of magnetic resonance imaging-related heat- tions. We do not know whether or not this technology improves out-
ing at 1.5 tesla. J Magn Reson Imaging 15:241–250, 2002.
come for frame based stereotaxy for subcortical targeting or targeting
24. Rezai AR, Phillips M, Baker KB, Sharan AD, Nyenhuis J, Tkach J, Henderson
of superficial structures. The number of patients treated was small,
J, Shellock FG: Neurostimulation systems used for deep brain stimulation
(DBS): MR safety issues and implications of failing to follow safety recom-
there were no controls, and the outcomes were incomplete. This appli-
mendations. Invest Radiol 39:300–303, 2004. cation may prove to be beneficial; however, further experience is
25. Rhode KS, Hill DL, Edwards PJ, Hipwell J, Rueckert D, Sanchez-Ortiz G, needed to understand the value of this advance. Though the purpose
Hedge S, Rahunathan V, Razavi R: Registration and tracking to integrate of the report is technique assessment, and the number of patients
X-ray and MR images in an XMR facility. IEEE Trans Med Imag treated was small, it is important for the authors to state whether or not
22:1369–1378, 2003. they experienced any early complications. It might also be instructive
26. Rhode KS, Sermesant M, Brogan D, Hegde S, Hipwell J, Lambiase P, to know whether or not postoperative imaging was performed for the
Rosenthal E, Bucknall C, Qureshi SA, Gill JS, Razavi R, Hill DL: A system for subthalamic nucleus STN deep brain stimulation DBS patients. We are
real-time XMR guided cardiovascular intervention. IEEE Trans Med Imaging told that the application of this technology was clearly advantageous
24:1428–1440, 2005. for brachytherapy, but it is not clear whether or not there was an
27. Rother J, Knab R, Hamzei F, Fiehler J, Reichenbach JR, Buchel C, Weiller C:
advantage for the functional procedures.
Negative dip in BOLD fMRI is caused by blood flow—Oxygen consumption
uncoupling in humans. Neuroimage 15:98–102, 2002. Andres M. Lozano
28. Rudin S, Bednarek DR, Wong R: Accurate characterization of image intensi- Toronto, Canada
fier distortion. Med Phys 18:1145–1151, 1991.
29. Siddon RL, Barth NH: Stereotaxic localization of intracranial targets. Int J 1. Fahrig R, Heit G, Wen Z, Daniel BL, Butts K, Pelc, NJ: First use of a truly
Radiat Oncol Biol Phys 13:1241–1246, 1987. hybrid X-ray/MR imaging system for guidance of brain biopsy. Acta
30. Starr PA, Vitek JL, DeLong M, Bakay RA: Magnetic resonance imaging-based Neurochir (Wien) 145:995–997, 2003.
stereotactic localization of the globus pallidus and subthalamic nucleus.
Neurosurgery 44:303–314, 1999.
31. Smith SM, De Stefano N, Jenkinson M, Matthews PM: Normalized accurate
measurement of longitudinal brain change. J Comput Assist Tomogr
25:466–475, 2001.
T he authors describe a hybrid imaging system that combines x-ray
and MRI for brachytherapy, the placement of deep brain stimula-
tors, and epidural grid placement for chronic pain. For the placement
32. Sturm V, Pastyr O, Schlegel W, Scharfenberg H, Zabel HJ, Netzband G, of stimulators in the subthalamic nuclei or a grid in the epidural space
Schabbert S, Berberich W: Stereotactic computer tomography with a modified where cerebrospinal fluid (CSF) will not be lost, this technique main-
Riechert-Mundinger device as the basis for integrated stereotactic neuroradi-
tains sufficient accuracy to assure the success of the procedure. Keeping
ological investigations. Acta Neurochir (Wien) 68:11–17, 1983.
the burr hole small during stimulator placement will help to minimize
33. Treuer H, Hunsche S, Hoevels M, Luyken K, Maarouf M, Voges J, Sturm V:
The influence of head frame distortions on stereotactic localization and target- any potential shift of the target because of CSF egress. However, I
ing. Phys Med Biol 49:3877–3887, 2004. would be cautious in the posterior fossa where a significant amount of
34. Treuer H, Klein D, Maarouf M, Lehrke R, Voges J, Sturm V: Accuracy and con- CSF can be lost and any preoperative coordinates would then be
formity of stereotactically guided interstitial brain tumour therapy using I-125 deemed inaccurate. Fortunately, the ability to obtain an intraoperative
seeds. Radiother Oncol 77:202–209, 2005. MRI can identify any aberrant positioning and allow for dynamic

ONS-360 | VOLUME 60 | OPERATIVE NEUROSURGERY 2 | APRIL 2007 www.neurosurgery-online.com

 COMBINED X-RAY AND MAGNETIC RESONANCE IMAGING-GUIDED NEUROSURGERY

adjustment. The accuracy recorded in this work (0.7 + 0.3 mm) sup- 2. Hall WA, Liu H, Martin AJ, Pozza CH, Maxwell RE, Truwit CL: Safety, efficacy,
ports the use of x-ray where no distortion as can be seen with magnetic and functionality of high-field strength interventional magnetic resonance
fields as they move away from isocenter. imaging for neurosurgery. Neurosurgery 46:632–642, 2000.
Most of the reported intraoperative MR-guided surgery has been for 3. Truwit CL, Hall WA: Intraoperative magnetic resonance imaging-guided neu-
rosurgery at 3-T. Neurosurgery 58[ONS Suppl 4]:ONS338–ONS346, 2006.
brain biopsy, epilepsy surgery, or for the resection of tumors (1, 2).
4. Hall WA, Galicich W, Bergman T, Truwit CL: 3-Tesla Intraoperative MR
This work expands the use of intraoperative MR-guided surgery to Imaging for Neurosurgery: J Neurooncol 77:297–303, 2006.
treat other disease processes, such as movement disorders and chronic 5. Hall WA, Liu H, Truwit CL. Functional magnetic resonance imaging guided
pain. As some intraoperative MR-guided neurosurgery centers are resection of low grade gliomas. Surg Neurol 64:20–27, 2005.
exploring the use of very high-field strength magnets (3-Tesla) for neu- 6. Hall WA, Truwit CL: 3-tesla functional magnetic resonance imaging-guided
rosurgery (3, 4), this group has decided to combine standard x-ray tumor resection. Int J Cars 1:223–230, 2006.
with 1.5-Tesla MRI. I applaud the use of brain activation studies or
functional MRI to localize eloquent areas of the brain (i.e., motor cor-
tex) because there are still some neurosurgeons that are reluctant to
accept this imaging technique despite a growing volume of literature
T his article presents a method for computational fusion of x-ray flu-
oroscopic images with MRI. These are obtained in a two-room MRI-
fluoroscopy suite, in which the patient can be readily moved between
substantiating its accuracy (5, 6). Only through continued investigation the fluoroscopy room and the MR bore, on a single gantry, if the MRI
and application of functional MRI to a broad spectrum of neurosurgi- needs to be updated. For the stereotactic procedures, patients were fit-
cal patients will this radiologic imaging modality finally gain wide- ted with a cranial mounted stereotactic frame, with an arc that appar-
spread acceptance by skeptics who are reluctant to accept its validity. ently fit within the MR bore. Therefore, the arc did not have to be
removed to update the MRI. Few details are given on sterile draping of
Walter A. Hall
the MR bore, which can be challenging when using a closed configura-
Syracuse, New York
tion MRI in the middle of an open surgery. The fusion algorithm for rep-
resenting MRI projected onto an x-ray fluoroscopic image is innovative.
1. Hall WA, Martin AJ, Liu H, Nussbaum ES, Maxwell RE, Truwit CL: Brain
biopsy using high-field strength interventional magnetic resonance imaging. Philip A. Starr
Neurosurgery 44:807–814, 1999. San Francisco, California