Keywords endonasal transsphenoidal surgery - navigation - endoscopic surgery - patient positioning
- pituitary adenoma - skull base tumor
Introduction
Intraoperative navigation for intracranial surgery has become increasingly routine
and popular within the last two decades. This technology has significantly contributed
to the progression and safety of endonasal pituitary and skull base surgery and clearly
serves as a valuable tool in these procedures. In 2002, the American Academy of Otolaryngology/Head
and Neck Surgery published recommendations for intraoperative use of image-guided
surgery (IGS) for specific pathologies involving the frontal, posterior ethmoidal,
and sphenoid sinuses; disease abutting the skull base, orbit, optic nerves, and ICA;
CSF leak; and benign and malignant sinonasal neoplasm.[1 ]
[2 ] Beyond confirmation of midline position, navigation is especially useful for correlation
of specific anatomical landmarks in patient space in the axial, coronal, and sagittal
planes.
Most frameless navigation systems require rigid head fixation to facilitate accurate
registration with scalp fiducials (SFs) or bone fiducials (BFs) or laser surface matching.[2 ]
[3 ]
[4 ]
[5 ] However, in endonasal pituitary surgery, many surgeons prefer to keep the head only
semi-immobilized in a horseshoe head holder that allows slight head movements to facilitate
surgeon access and obviates potential postoperative pain and bleeding at pin sites.
Until recently, there did not appear to be an IGS system that allowed accurate navigation
without rigid head fixation. Since 2005 we have used a noninvasive surface autoregistration
mask that adheres to the patient's face and forehead without the need for scalp marking
or shaving (Stryker Navigation, Freiburg, Germany). The mask holds 31 light-emitting
diodes (LEDs) distributed on one central (nasal) and two lateral arms. In our initial
use of the mask over 4 years, patients' heads were fixed in a Mayfield head holder,
the data was transferred to a tracker, and the mask was removed after registration.
More recently, we have returned to only semirigid immobilization in a horseshoe head
holder for many endonasal cases and have fine-tuned use of the autoregistration mask
with the communication box attached to the mask. We have now used this technique of
facial autoregistration mask with only semirigid immobilization in 50 endonasal cases,
and it has become our preferred method. This study assessed the accuracy and the practicality
of this navigation technique in endonasal endoscopic surgery without rigid head fixation.
Methods
Patient Cohort
All consecutive patients undergoing surgery at Saint John's Health Center for sellar
and/or parasellar pathologies from October to December 2010 were included in this
study. The surgical team comprised a neurosurgeon and an ear, nose, and throat (ENT)
surgeon. The endonasal route was used in all cases using either an endoscope-assisted
or a purely endoscopic technique. Patient consent for the use of the navigation system
was obtained at the time of surgical consent. The Institutional Review Boards of each
institution approved this retrospective study of patient data.
Technique Description
Once the patient is under general anesthesia, the head is placed upon a padded horseshoe
head holder with ~15 degrees of rightward rotation and a slight leftward tilt to facilitate
surgeon access from the patient's right side. Greater degrees of head extension are
used for lesions extending anterior to the sella in the frontal fossa.
All patients undergo a pre-operative sellar magnetic resonance imaging (MRI) and an
additional T1 post-gadolinium thin-slice (2-mm cuts) navigational sequence. The navigational
MRI is uploaded into the Stryker Navigation System (Stryker Instruments, Kalamazoo,
Michigan), with axial, coronal, and sagittal views, and the optical detection camera
is appropriately positioned, typically three to five feet above and behind the patient's
head. The optical camera has a bird's- eye view of the registration mask and communication
box, ensuring constant tracking. Both surgeons stand to the right of the patient's
head and torso. Only occasionally does the person driving the endoscope need to slightly
change the angle with which the scope is being held to avoid line- of- sight obstruction
to the mask. This problem has not been found to be a limitation but is a well-appreciated
issue common to all optical tracking navigation systems.
As shown in [Fig. 1 ], the surface auto-registration mask is positioned over the forehead, nose, and upper
face (Stryker Navigation, Freiburg, Germany). The communication box is attached to
the mask and then supported with 4 × 4 pads and taped to the horseshoe. This securing
of the box prevents it from pulling and distorting the mask on the forehead, which
will otherwise distort the registration process with loss of accuracy. It also prevents
movement of the communication box during the case, should the patient be rotated laterally.
Figure 1 Overview of the use of the autoregistration mask without head fixation.
The communication box is then activated. The optical infrared camera must recognize
at least 21 of the 31 LEDs of the mask. For optimal results, all except one LED should
be seen on each arch. Before proceeding with registration, care must be taken to make
sure that the endotracheal tube tape is not displacing or distorting the cheeks or
nose skin to which the mask is attached. Also, there should be no tape covering the
eyes. Once the system has established a correlation between the image data and patient
data, the accuracy is initially qualitatively assessed by touching five anatomical
landmarks. When assessing the correspondence between image space and patient space,
care must be taken to gently apply the tip on the patient's skin without any pressure
because this can lead to erroneous interpretation of the accuracy. The anatomical
landmarks used in this study include the left outer canthus (point 1), left inner
canthus (point 2), the nasion (point 3), the right inner canthus (point 4), and the
right outer canthus (point 5) ([Figs. 2 ], [3 ]). The distance between the actual and virtual target is assessed on the axial, coronal,
and sagittal planes. If the distance was consistently less than 1 mm through the verification
process, the registration was accepted. The system provides a surface registration
error (SRE) in millimeters. The target registration error (TRE) (or operational accuracy)
was assessed at the beginning and end of the case by measuring the distance between
the tip of the navigation wand and the skin as measured in millimeters on the axial,
coronal, and sagittal views for each landmark. The Stryker Navigation System allows
measuring submillimetric distances on the processed axial, coronal, and sagittal sequences.
Figure 2 Example of navigation accuracy at the beginning of the procedure (case 12). Views
in coronal, sagittal and axial planes at the right outer canthus (A); right medial
canthus (B); nasion (C); left medial canthus (D); left outer canthus (E).
Figure 3 Example of navigation accuracy at the end of the procedure (case 12). Views in coronal,
sagittal and axial planes at the right outer canthus (A); right medial canthus (B);
nasion (C); left medial canthus (D); left outer canthus (E).
The autoregistration mask is left in place throughout the case. The eyes can then
be protected, taking care not to cover the top arches with tape. After registration,
only the LEDs on the top arches of both lateral arms emit a signal for continuous
navigation. Any pulling or pushing on the mask transmitted to the superior arches
can contribute to loss of operational accuracy. Therefore, the branch of LEDs that
extends along the nose is cut after registration is completed. This eliminates potential
navigational error from nasal distortion by the endoscope, endonasal speculum, or
other instruments that can distort the nasal skin contour.
Since the mask and the communication box must remain visible to the optical camera
throughout the case, a clear sterile drape is required. The only opaque drape that
covers the facial skin is below the nares to cover the mouth. The rest of the drapes
are stuck to the clear drape, preventing any distortion by the weight of the drapes.
The rest of the draping proceeds as usual.
At the beginning of each case, the system's accuracy is also validated using the midline
sphenoid keel or an intrasphenoidal bony septation visualized directly with the endoscope.
The system is evaluated as accurate by the surgical team if the operational accuracy
was <1 mm. Intraoperative navigation is used regularly throughout endonasal procedures.
Practicality was also evaluated by measuring the time taken from activation of the
communication box to acceptance of registration. This represents the time in the operating
room (OR) uniquely dedicated to the use of the navigation system. Failure of the system
is also a measure of practicality. Intraoperative complications were also noted.
Results
Over a 7-week period from October to December 2010, 12 consecutive patients underwent
surgery for a sphenoid, sellar, parasellar pathology including pituitary macroadenomas
(n = 9), chordoma (n = 1), craniopharyngioma (n = 1), and sinonasal melanoma (n = 1). Four of these surgeries were for recurrent pathologies ([Table 1 ]).
Table 1
Summary of Data
Case
Pathology
Time (sec)
SRE[a ] (mm)
TRE (mm)[b ] pre-op
TRE (mm)[b ] post-op
1
Recurrent chordoma
92
0.6
0.6
1.3
2
Pituitary macroadenoma
92
0.9
1.5
2.08
3
Recurrent pituitary macroadenoma
129
0.8
0.83
1.75
4
Recurrent sinonasal melanoma
71
0.8
0.2
0.4
5
Pituitary macroadenoma with apoplexy
58
0.6
1.1
1.07
6
Pituitary macroadenoma
59
0.8
0.92
1.04
7
Craniopharyngioma
88
0.8
0.47
0.52
8
Pituitary macroadenoma
78
0.6
0.6
N/A[c ]
9
Pituitary macroadenoma
77
0.8
0.84
0.90
10
Pituitary macroadenoma
64
0.8
0.52
1.08
11
Pituitary macroadenoma
90
0.8
0.8
0.52
12
Pituitary macroadenoma
133
0.9
0.88
0.56
a SRE: Surface registration error, given by the navigation system after registration.
b Target registration error (TRE): Measures the distance between the tip of the navigation
wand and the target. The mean value for each patient is presented.
c The postoperative TRE values were not available (N/A) for this patient; however,
excellent correlation was noted between the image point and the patient point throughout
the case. The calculations presented in the result take this into consideration.
Using the five anatomical landmarks detailed in the methods, the median time required
for registration and accuracy verification took 84 seconds varying between 64 to 129 seconds
([Table 1 ]). No system failure occurred in these 12 cases and no surgical complications occurred.
The mean SRE recorded by the neuronavigational software was 0.8 mm (interval 0.6 to
0.9 mm). At the beginning of the case, the mean TRE was 0.9 ± 0.7 mm and the median
was 1 mm (interval of 0.2 to 3 mm). The mean TRE at the end of the procedure was 1.0 ± 0.8 mm
and the median was 0.5 mm (interval of 0.2 to 3 mm). In 44% of readings, there was
no change of the TRE between the initial values and those measured at the end of the
procedure. The average absolute difference between initial and final TRE was 0.7 ± 0.7 mm
and the median was 0.8 mm (interval of 0 to 1.8 mm). For all the cases in this study
population, both surgeons of the surgical team were satisfied with the intraoperative
navigation operational accuracy verified at the beginning of surgery using a deep
bone landmark (TRE <1 mm). However, in two cases the navigation system appeared to
be intermittently inaccurate. Intraoperative troubleshooting found that this occurred
when some of the LEDs from the mask were hidden either by the surgeon's hand, the
endoscope cables, or the drapes. Once this was addressed and all the LEDs could be
seen by the optical camera, navigation was once again accurate.
Discussion
Navigation and Endonasal Surgery
Throughout the years, numerous navigation systems have been used in image-guided surgeries.
Traditional intraoperative fluoroscopy provides sagittal plane bony orientation and
navigation in transsphenoidal surgery. Although it allows one to work without head
fixation, its monoplanar bony information is clearly limited. Current computed tomography
(CT) or MRI-based navigation systems provide a much greater degree of anatomical detail
without radiation. In the field of sinonasal and endonasal endoscopic skull base surgery,
these systems appear to have increased the safety of surgery and facilitated the learning
curve for many surgeons.[6 ] In transcranial surgery, these systems have allowed tailoring of approaches by planning
smaller yet strategically positioned craniotomies, helping advance the concept of
keyhole surgery. Although navigation does not replace a thorough knowledge of the
sinonasal and endonasal skull base anatomy, provided the system has intraoperative
accuracy, it allows instantaneous correlation of patient space and image space in
three planes. It also enables surgeons to readjust their three-dimensional (3D) perception,
which is especially useful in cases with anatomical variations or prior surgery and
thereby helps prevent catastrophic complications.[6 ]
The BFs represent the gold standard of pair-point registration, yielding the highest
navigational accuracy.[2 ]
[3 ] However, this technique is also the most invasive, potentially causing discomfort
to the patient as well as an additional CT scan and its associated radiation.[2 ]
[5 ] Some less invasive registration methods and navigation systems may result in satisfactory
levels of accuracy (in the 1.0- to 2.0-mm range) for specific regions of interest
such as the midface and skull base.[2 ]
[7 ] We have also tried using a headband technique (BrainLab, Feldkirchen, Germany) without
head fixation, which did not prove sufficiently reliable or accurate.
Using the Autoregistration Mask without Head Fixation
The autoregistration mask is a simple, noninvasive surface registration technique.
It requires no special preparation at the patient's bedside because it is installed
after the patient is intubated and requires no additional imaging after the mask is
positioned. More recently, as shown in this study population, we have stopped fixing
the head in pins and proceed with free head navigation, leaving the mask in place
throughout the entire procedure. In a technical note on endonasal combined microscopic
endoscopic surgery for removal of pituitary adenomas, Al-Mefty and colleagues mention
using the mask for registration without fixation of the patient's head.[8 ] They state that the mask is an excellent adjunct for navigating during pituitary
surgery because of its easy and intuitive use. More recently, Makieses compared in
a cadaver model the operational accuracy of the autoregistration mask with that obtained
with SFs or BFs.[5 ] The accuracy for external anterior targets was 1.96, 3.12, and 3.20 mm for BF, SF,
and autoregistration mask, respectively; however, when reaching internal targets,
the autoregistration mask was associated with an operational accuracy (2.41 mm) statistically
superior to that associated with BF (2.91 mm) or SF (3.83 mm).[5 ] This is the first anatomical study that assessed the accuracy of the autoregistration
mask.[5 ] The authors mention that, given its incorporated tracking system that can be seen
by the optical camera, rigid fixation of the patient's head may be obviated. Additionally,
they state that rigid fixation may cause displacement or distortion of the skin, alter
the integration process of image-patient coordinates, and result in decreased accuracy.[4 ]
[9 ] To our knowledge, this study is the first clinical report showing feasibility and
accuracy of using the autoregistration mask throughout endonasal surgery without rigid
head fixation.
Accuracy and Practicality
In the clinical setting, navigation systems reach operational accuracies between 0.5
and 2.77 mm,[2 ]
[10 ]
[11 ] and several reports have stated that an operational accuracy (TRE) of 1 to 2 mm
is required for safe use of a navigation system.[2 ]
[7 ] For each navigation system, numerous factors can influence the operational accuracy,
including preoperative image quality, the registration procedure, and the navigation
system.[9 ]
[12 ]
[13 ] Intraoperative events can also modify the accuracy, such as movement of the tracking
device. Accuracy values also vary depending on if the assessment was performed on
cadavers,[5 ] skull models,[2 ] or patients.[14 ]
[15 ] This study demonstrates that satisfactory accuracy can be obtained by using the
autoregistration mask without head fixation (free-head) while maintaining excellent
accuracy: a SRE of 0.8 mm (interval 0.6 to 0.9 mm) and a mean TRE of 0.9 mm and 1.0 mm
measured respectively at the beginning and end of the case. In almost half of the
readings (44%), there were no shifts between the initial and final TRE. Numerous adjustments
were brought since the beginning of the use of the autoregistration mask. The set-up
and technique as described in the methods is simple and rapid. The small technical
details—including cutting the nasal branch of the LED display, making sure the eye
dressing does not cover the superior lateral arches of the mask, using a clear drape,
and preventing the LEDs from being hidden from the scope cables—together ensure the
best operational accuracy throughout surgery. Neuromonitoring can be used in the presence
of the autoregistration mask and the communication box but requires that the scalp
sensors be positioned further posterior because of the possible noise in the somatosensory
evoked potential (SSEP) signal. Postoperatively, we have observed fewer patients reporting
scalp and head pain, although we did not objectively document this in the current
study.
Conclusion
Use of the surface autoregistration mask in endonasal pituitary and skull base tumor
surgery without head fixation enables satisfactory operational accuracy throughout
the procedure, and is practical, reliable, and noninvasive. Further assessment will
help determine in which surgical procedures this system is most beneficial and how
to improve this registration technique and navigation system.
