Visualization of anterior skull base defects with intraoperative by mikeholy



Gideon Bachar, MD,1 * Emma Barker, MD,1 * Harley Chan, PhD,2 Michael J. Daly, MSc,2
Sajendra Nithiananthan, BSc,2,3 Al Vescan, MD,4 Jonathan C. Irish, MD,1
Jeffrey H. Siewerdsen, PhD2,3,5
  Department of Otolaryngology–Head and Neck Surgery, Department of Surgical Oncology,
Princess Margaret Hospital, Toronto, Ontario, Canada. E-mail:
  Ontario Cancer Institute, Princess Margaret Hospital, University Health Network,
The Institute of Medical Science at the University of Toronto, Toronto, Ontario, Canada
  Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada
  Department of Otolaryngology–Head and Neck Surgery, Mount Sinai Hospital, Toronto, Canada
  Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD

Accepted 17 June 2009
Published online 19 August 2009 in Wiley InterScience ( DOI: 10.1002/hed.21219

                                                                     Conclusions. The ability to identify ASBD depended on the
Abstract: Background. The role of cone-beam CT (CBCT)
                                                                  size and location of defect. Oblique viewing planes were opti-
in demonstrating anterior skull base defects (ASBDs), differing
                                                                  mal for ASBD visualization. V 2009 Wiley Periodicals, Inc.
in size and location, was investigated. The study was designed
                                                                  Head Neck 32: 504–512, 2010
to describe the potential advantage of CBCT in the setting of
an intraoperative cerebrospinal fluid (CSF) leak.                  Keywords: cone-beam CT; sinus; skull base;            surgery;
   Methods. In all, 120 ASBD were evaluated in 5 cadaver          intraoperative; imaging; cerebrospinal fluid (CSF)
heads. Orthogonal and oblique slices were reconstructed. Ob-
server studies assessed the visibility of ASBD in each location
as a function of defect size.
   Results. For 1-, 2-, and 4-mm defects, the percentage that     Cerebrospinal fluid (CSF) rhinorrhea results
were undetectable ranged from 20% to 33%, 0% to 14%, and          from a breach in the anterior skull base, leading
0% to 5%, respectively. Confident breach detection increased       to a direct communication between the subarach-
with defect size and was most challenging in the lateral
                                                                  noid space and either the sinus or nasal cavity.
lamella and cribriform. CBCT permitted confident detection of
ASBD as small as about 2 mm in the fovea ethmoidalis and          Classified according to etiology,1 approxi-
planum. Oblique views were found to be superior to orthogo-       mately 90% of cases are a result of trauma,
nal planes.                                                       either an accident or a postsurgical (iatrogenic)
                                                                      CSF rhinorrhea resulting from an iatrogenic
                                                                  surgical complication is an increasing concern
Correspondence to: G. Bachar                                      as a result of the increasing popularity of endo-
*The first 2 authors contributed equally to this work.             scopic sinus surgery and closed neurosurgical
V 2009 Wiley Periodicals, Inc.
C                                                                 procedures. The size and anatomic location of

504     Intraoperative Cone-Beam CT                                             HEAD & NECK—DOI 10.1002/hed         April 2010
FIGURE 1. (A) Experimental setup showing the C-arm and cadaveric head. (B) Endoscopic photograph illustrating skull base defects
after total ethmoidectomy, medial maxillectomy, sphenoidotomy, and excision of the pituitary gland. Eight locations were systematically
breached: (1 and 2) left and right cribriform plate; (3 and 4) left and right lateral lamella; (5 and 6) left and right fovea ethmoidalis; and
(7 and 8) left and right planum, respectively. The figure shows a specimen with 2-mm-diameter defects introduced at each location.
[Color figure can be viewed in the online issue, which is available at]

an anterior skull base defect are multifactoral,                          procedure has a significant advantage. It enables
depending on the type of procedure, extent and                            the surgeon to immediately seal the skull base
location of disease, surgical instruments used,                           defect and to minimize the risk of further trauma
and surgical expertise.                                                   and postoperative complications.
    Skull base anatomy presents a complicated                                 To overcome the limitations associated with
architecture predisposed to iatrogenic CSF leaks                          guidance based on preoperative image data, we
that can be difficult to detect intraoperatively.3                         described a technology for 3-dimensional (3D)
CSF rhinorrhea may be associated with both sig-                           cone-beam CT (CBCT) imaging using a mobile C-
nificant morbidity and mortality. Meningitis is                            arm modified to include a high-performance flat-
the most common complication, with a reported                             panel x-ray detector.10–12 The method operates
incidence of 19% in patients with persistent CSF                          similar to CT, except CBCT captures an entire
rhinorrhea.4 Other complications include ortho-                           volume image in a single half-rotation ($180 )
static (upright) headache, intracranial infections,                       about the patient at lower radiation dose. Its
pneumocephalus, and bleeding.3,4                                          application in head and neck surgery has been
    CSF leaks that are diagnosed intraoperatively                         reported in a variety of studies,10–12 and a system
are treated immediately. In cases in which the                            for CBCT-guided head and neck surgery has
leak is diagnosed a few days after surgery, a con-                        entered research trials at our institution. In the
servative approach is generally adopted.5 Patients                        present study, we investigated the role of CBCT
with CSF leaks present a diagnostic challenge to                          in demonstrating anterior skull base defects
the surgeon. Various preoperative imaging modal-                          (ASBDs) varying in size and anatomic location.
ities are available for the diagnosis and identifica-                      The study was designed to evaluate the perform-
tion of a CSF leak. High-resolution CT scans, T2-                         ance of CBCT for intraoperative detection of an-
weighted MRI, MR cisternography, and intrathe-                            terior skull base CSF leak.
cal fluorescein have all been described.5–8 How-
ever, both radionuclide cisternography and
contrast-enhanced CT cisternography have been                             MATERIALS AND METHODS
the mainstay of evaluation of a skull base defect.
The clear disadvantage of the cystenography                               C-Arm for Cone-Beam CT Imaging. A prototype
studies is the need for intrathecal injection. These                      mobile isocentric C-arm (Siemens PowerMobil)
are invasive, time consuming, and include poten-                          was developed in collaboration with Siemens
tial side effects of headache, nausea, vomiting,                          Healthcare (Special Products Division, Erlangen,
seizures, and intracerebral hemorrhage.9 The                              Germany) for intraoperative CBCT. As illustrated
ability to identify a CSF leak during the surgical                        in Figure 1A, the C-arm was modified to include

Intraoperative Cone-Beam CT                                                       HEAD & NECK—DOI 10.1002/hed             April 2010      505
a flat-panel detector (FPD; PaxScan 4030CB,                 The 3D coordinates of each defect were iden-
Varian, Palo Alto, CA), a motorized orbital drive,     tified by an independent observer. For smaller
geometric calibration, and a computer control sys-     defects that were difficult to identify (eg, 1 mm),
tem for 3D reconstruction. CBCT acquisition            the coordinates of the large defects (4 mm) were
involved collection of projections across a rota-      translated to the earlier scans, using anatomic
tional arc of about 178 , nominally 200 projections   landmarks to account for possible displacement
acquired in about 60 s ($3.3 fps). The 3D field of      of the specimen between scans. In all, 120
view (FOV, $20 cm  20 cm  15 cm) is sufficient        defects (5 heads  8 locations  3 diameters)
to encompass the skull base.13–15 The dose associ-     were thus localized.
ated with CBCT imaging on the C-arm has been               For the observer studies described in the fol-
previously described10: approximately 2.9 mGy          lowing text, image slices through each defect
(0.10 mSv) and 9.6 mGy (0.35 mSv) were shown           were extracted from CBCT volume reconstruc-
sufficient for visualization of bone and soft tis-      tions using the ‘‘3D Slicer’’ software. These
sues, respectively. Such dose levels represent a       included ‘‘orthogonal’’ slices that correspond to
fraction ($1/10–1/3) that of diagnostic CT.            conventional triplanar views (coronal, sagittal,
                                                       and axial) as well as ‘‘oblique’’ slices. The
Specimen Preparation and CBCT Imaging. Five            oblique slices included both a coronal and a lon-
cadaver heads were acquired from the Division of       gitudinal slice (ie, a quasi-sagittal slice in the
Anatomy, University of Toronto, in accord with         plane of the drill) and a tangential slice (ie, a
the Anatomy Act of Ontario. All heads were ini-        quasi-axial slice in the plane of the defect).
tially scanned on a 16-slice diagnostic CT scanner         For each defect, an area of interest was
(Discovery PET-CT; General Electric, Milwaukee,        cropped from the image such that the defect
WI) to review any anatomic variations and              was presented in the center of the image. Slice
excluded any anatomic abnormality or disease.          triples (either orthogonal or oblique) were dis-
During the head preparation, each head under-          played in comparison with the corresponding
went a complete surgical clearance of the eth-         ‘‘baseline’’ area of interest (ie, the identical area
moid, maxillary, and sphenoidal sinuses. Air           of skull base before introduction of a defect).
space along the skull base (attributed to postfixa-     Example images are shown in Figures 2 and 3
tion brain shrinkage) was filled with water-equiv-      for each anatomic location.
alent gel via craniotomy. Subsequently, the                Observer studies were conducted to assess
cranial bone flap was reopposed and secured.            visibility of skull base defects in each location as
    After preparation, a baseline CBCT scan was        a function of defect size. Five expert observers
taken. Each head was immobilized in a carbon-          participated, including 1 radiologist, 2 skull
fiber frame using 6 to 8 cranial pins, and was          base surgeons, and 2 head and neck surgeons.
supported at the C-arm isocenter for each scan.        The study was conducted under controlled view-
Eight locations on the anterior skull base were        ing conditions: a darkened reading room on a
selected for introduction of skull base breach as      diagnostic workstation (Dell Precision 380, 3-
illustrated in Figure 1B: the (left and right) cri-    GHz Pentium 4 with dual-head Barco displays,
briform plate, fovea ethmoidalis, lateral lamella,     1536 Â 2048 resolution, 8-bit grayscale; Dell
and planum. After the baseline scan, an experi-        Inc., Round Rock, TX). A fixed viewing distance
enced skull base surgeon used a 1-mm-diameter          of 50 cm was recommended but not enforced.
drill to introduce skull base defects into each        Observers received identical instructions and
site, and a second CBCT scan was acquired.             training. Reading order across 120 cases was in-
Then each of the 1-mm defects was widened              dependently randomized for each observer. For
using a 2-mm drill (2-mm defects are shown in          each case, observers were shown (1) the full
Figure 2), followed by a third CBCT scan.              FOV (nominal resolution) coronal and sagittal
Finally, all sites were drilled to a 4-mm diame-       images for purposes of orientation; (2) the base-
ter, and a final CBCT scan was acquired.                line images (orthogonal or oblique slices with no
                                                       defect); and (3) the defect images (orthogonal or
Visualization of Skull Base Defects: Observer Study    oblique slices with a defect). Observers were
and Data Analysis. The images were viewed in 3D        then asked to rate their satisfaction in detect-
visualization software (3D Slicer v3.2; Brigham &      ing, localizing, and characterizing the defect on
Women’s Hospital, Boston, MA, and Massachu-            a scale of 1 to 5 (1 ¼ Unable to identify any
setts Institute of Technology, Cambridge, MA).         breach, 2 ¼ The breach could be overlooked, 3 ¼

506   Intraoperative Cone-Beam CT                                  HEAD & NECK—DOI 10.1002/hed    April 2010
FIGURE 2. Illustration of variable defect size (baseline, 1-, 2-, and 4-mm-diameter breach). Examples correspond to defects in the pla-
num demonstrated in orthogonal (coronal, sagittal, and axial) views. Cone-beam computed tomographic (CBCT) images. Throughout
the following figures, the full field-of-view (FOV) images (left) at ‘‘nominal’’ resolution (0.8-mm voxels) illustrate the surrounding ana-
tomic context. The zoomed images (right) at ‘‘high’’ resolution (0.4-mm voxels) illustrate the region of the breach. In each case, the
breach is near the center of the image, and its location is demarked by gray triangles at the image boundaries.

The breach is visible, but detection and charac-                       Â 3] ¼ 30 in the planum, 30 in the fovea eth-
terization of subtle features are a bit challeng-                      moidalis, 29 in the lateral lamella, and 13 in the
ing, 4 ¼ The Breach is visible, 5 ¼ The breach                         cribriform). The shortfall of cases in the cribri-
is perfectly obvious).                                                 form was attributed to difficulty in drilling large
    The results were analyzed using both a                             (4-mm) defects, which was technically challeng-
mixed-regression model and trend testing. Mean                         ing because of the thickness of the skull base at
values (and SD) in observer scores were ana-                           this point, resulting in a much bigger defect
lyzed as a function of anatomic location (4 sites)                     being created than desired and thus leading to
and defect size (3 diameters). Results were also                       rejection of nearly 17 cases from the analysis.
analyzed for orthogonal views compared with                                The percentage of defects that were judged
oblique views to see whether the latter more                           undetectable (a score of 1) depended on the loca-
clearly demonstrated defects. A paired t test                          tion and size of the defect. For the 1-mm defects,
(Wilcoxon) was used to test statistical signifi-                        the percentage judged undetectable ranged
cance in observer scores between orthogonal and                        from 20% to 33% (specifically, 30% [cribriform],
oblique views.                                                         33% [lateral lamella], 20% [fovea ethmoidalis],
                                                                       and 21% [planum]). For the 2-mm defects, the
RESULTS                                                                percentage judged undetectable was consider-
Of the 120 defects (5 heads  8 locations  3                          ably lower, 0% to 14% (12% [cribriform], 14% [lat-
diameters), 102 were deemed successful ([5 Â 2                         eral lamella], 0% [fovea ethmoidalis], and 1%

Intraoperative Cone-Beam CT                                                    HEAD & NECK—DOI 10.1002/hed            April 2010     507
FIGURE 3. Illustration of skull base breach in regions of the cribriform plate (A), fovea ethmoidalis (B), lateral lamella (C), and planum
(D). Each example shows baseline images in comparison to images of a 2-mm defect in oblique views.

[planum]). The 4-mm defects were almost always                          moidalis], and 0% [planum]). Besides demon-
detectable, with 0% to 5% rated undetectable (5%                        strating more confident detection for larger
[cribriform], 2% [lateral lamella], 1% [fovea eth-                      defects, these results indicate that breach

508    Intraoperative Cone-Beam CT                                                     HEAD & NECK—DOI 10.1002/hed            April 2010
FIGURE 4. Plots summarizing the visibility of skull base breach (observer score 1–5) as a function of the defect size (1–4 mm). Mixed
linear regression is consistent with the finding that visibility of breach in the cribriform was significantly more challenging than that in
the fovea ethmoidalis, lateral lamella, and planum.

detection was most challenging in the lateral                           pendence in scores from 1 mm to 4 mm, suggest-
lamella and cribriform and most conspicuous in                          ing a trend that is significantly greater than 0.
the fovea ethmoidalis and planum.                                       Specifically, observer score increased by 0.60 for
    Similarly, the results indicate the degree to                       every 1-mm increase in defect size (p < .0001).
which CBCT enables confident detection of skull                          After considering various anatomic sites, the
base breach (score of !4 in observer rating). For                       slopes of regression were found to be 0.68 (pla-
the 1-mm defects, confident detection was                                num), 0.64 (lateral lamella), 0.65 (fovea ethmoi-
achieved in 3% to 14% of cases (specifically, 3%                         dalis), and 0.30 (cribriform). The last is
[cribriform], 7% [lateral lamella], 7% [fovea eth-                      consistent with the previously noted point that
moidalis], and 14% [planum]). For the 2-mm                              breaches were more difficult to detect in the cri-
defects, confident detection was achieved in 44%                         briform. Interestingly, the intercept points for
to 69% of cases (specifically, 45% [cribriform],                         the 4 sites plotted in Figure 4 were not signifi-
44% [lateral lamella], 69% [fovea ethmoidalis],                         cantly different (p ¼ .14).
and 67% [planum]). Finally, for the 4-mm                                    There was a significant difference in the
defects, confident detection was achieved in 47%                         scores associated with orthogonal versus oblique
to 91% of cases (specifically, 47% [cribriform],                         views, as shown in Figure 5. The intercept
71% [lateral lamella], 87% [fovea ethmoidalis],                         implied by the mixed regression analysis was
and 91% [planum]).                                                      significantly different (p ¼ .02) for the 2 types of
    A mixed-regression model allowed analysis of                        views, with a lower intercept for orthogonal
statistical significance in the differences arising                      views (1.75) than that for oblique views (2.06).
from defect size, in a manner that adjusts for                          Interestingly, the slopes in the observer scores
the dependence on anatomic site, showing the                            (ie, the change in score per unit size of the
observer scores for the 3 defect sizes (1, 2, and                       breach) of the 2 types of views were not signifi-
4 mm) to be significantly different (p < .0001).                         cantly different (p ¼ .26). A paired t test
As shown in Figure 4, the average score                                 between scores in orthogonal and oblique views
increased with defect size: average score ¼ 2.51                        suggests a significant difference in the planum,
for 1 mm, 3.82 for 2 mm, and 4.45 for 4 mm.                             combining defect sizes (p < .001) and also
The regression model also tested the linear de-                         when all 4 sites were combined (p < .001) (see

Intraoperative Cone-Beam CT                                                     HEAD & NECK—DOI 10.1002/hed            April 2010     509
FIGURE 5. Plots summarizing the visibility of skull base breach (observer score 1–5) for orthogonal views (light gray bars) and oblique
views (black bars). For individual anatomic sites, only the planum (D) illustrated a statistically significant difference between orthogonal
and oblique views, whereas pooling over all sites and defect sizes demonstrated an overall improvement (p < .001) for oblique views.

Figure 5). However, the cribriform plate, lateral                       base defects. Various authors suggest different
lamella, and fovea ethmoidalis, individually, did                       imaging modalities to evaluate iatrogenic ante-
not show a significant difference between or-                            rior skull base defects,6,8,16 but the reported
thogonal and oblique views. In comparing differ-                        sensitivity and specificity of such tests are
ent anatomic sites, there was a significant                              inconsistent, probably the result of different
difference for 1-mm defects (p < .001) and 2-mm                         imaging modalities and small numbers of cases.
defects (p < .002), but not 4-mm defects (p ¼ .4).                      In addition, the visibility of a defect depends on
                                                                        its location and size, an uncontrolled variable in
                                                                        previous studies. The results in this report con-
                                                                        firm this point: a 2-mm breach in the cribriform
DISCUSSION                                                              plate was significantly more challenging to iden-
This study is the first to investigate the role of                       tify than was a breach of identical size in the
CBCT imaging for visualization of ASBD. The                             planum. Shetty et al7 found that high-resolution
cadaver model provided a somewhat idealized,                            CT was accurate in 93% of patients, whereas
but reasonably robust, tool to evaluate CBCT in                         MR cisternography was accurate in 89% of
the detection of known breaches of variable size                        patients with clinically suspected CSF rhinor-
and within distinct anatomic locations of the                           rhea. The combination of high-resolution CT
anterior skull base. The advantage of the cur-                          and MR cisternography was accurate in 96% of
rent study is that cadaveric specimens allowed a                        patients. In contrast, Eljamel et al17 demon-
strict protocol, enabling the creation of bony                          strated the use of high-resolution CT with an
defects of specific size in predetermined loca-                          identification rate of only 47%.
tions. We were therefore able to evaluate the                               For small (1-mm) skull base defects, 20% to
identification rate/sensitivity of each site sepa-                       33% of cases did not demonstrate visualization
rately and assess how easily individual defects                         of the defect in CBCT, despite observers know-
could be detected.                                                      ing of its existence in the image. This is in con-
    To date, there is no universally accepted mo-                       trast to just 2% of cases for which a 4-mm
dality or algorithm to evaluate anterior skull                          defect could not be identified. Of the 4 anatomic

510    Intraoperative Cone-Beam CT                                                      HEAD & NECK—DOI 10.1002/hed            April 2010
sites, the planum and fovea ethmoidalis demon-       and treated during the primary procedure using
strated clearer visibility of defects of any size.   CBCT. This would minimize the need for a sec-
These and the lateral lamella exhibited similar      ond procedure, reduce patient stress and dis-
‘‘slopes’’ in detectability versus size of breach    comfort, and decrease both operating room costs
(from 1 mm to 4 mm), suggesting an improve-          and time. In addition, practice that aims to
ment at larger defect identification (2 mm then       repair a CSF leak as a second procedure con-
4 mm). The detection rate within the cribriform      tends with a previously operated field with asso-
plate was significantly lower, suggesting that        ciated scar tissue and fibrosis. The ability to
even larger defects (eg, 2 and 4 mm) in the cri-     identify and repair a CSF leak during the pri-
briform were not as easy to identify as in other     mary surgery can eliminate such problems. In
anatomic sites. Interestingly, the ‘‘y-intercept’’   any case, we are not suggesting that CBCT will
on plots of detectability versus size of breach      replace the need for good preoperative imaging.
were not significantly different among the 4              This study has a variety of limitations. First,
anatomic sites, implying that a small breach         the study was performed on cadaver heads and
(<1 mm) would become unidentifiable irrespec-         not on living patients with an actual CSF leak.
tive of anatomic site.                               Second, the observers were asked to identify the
     Overall, CBCT provided confident detection       skull base breach while reviewing a narrow part
(score of !4) in 44% to 69% of 2-mm defects and      of the image, unable to scroll through adjacent
47% to 91% of 4-mm defects, across all anatomic      slices, and the observers were informed that the
sites tested. We acknowledge that the identifica-     breach did, in fact, exist in the image. This is
tion rate may be falsely elevated by the prede-      in contrast to the clinical setting in which the
fect skull preparation (including ethmoidectomy      surgeon or radiologist is required to search an
and mucosa stripping). However, to create a          entire volume to locate the bony defect from
comparable model, between individual skulls,         multiple images collected during the CT scan. In
this preparation was necessary.                      addition, the test (satisfaction rating scale) does
     In a comparison of conventional orthogonal      not allow quantitation of sensitivity or specific-
triplanar views (coronal, axial, and sagittal        ity of breach detection.
planes) with oblique views (coronal, longitudi-          In summary, a cadaver model was used to
nal, and tangential planes through the defect),      test intraoperative CBCT detection of ASBD,
the results confirm that breaches were better         investigating detectability thresholds as a func-
demonstrated in the latter. Therefore, if 3D vis-    tion of defect size and location, each found to
ualization software permits convenient selection     affect detectability. In addition, optimal viewing
of oblique planes to more accurately contain the     planes (oblique) were identified. The utility of
true plane of the breach, the identification rate     this experimental model was demonstrated, and
is improved.                                         implementation of the system in a clinical intra-
     CSF leaks are typically intermittent, and       operative environment is under way.
some imaging (cisternography) requires active
leakage during the examination to improve sen-       Acknowledgments. The authors extend their
sitivity. Moreover, cisternography also requires     thanks to the Guided Therapeutics Program
an intrathecal injection and a transfer to the CT    (GTx) at the University Health Network (Toronto,
scanner, before the operating room. The use of       ON) and to Dr. Eugene Yu (University Health
CBCT overcomes these obstacles and would be          Network, Toronto, ON) for expert assistance with
ideal in sinus/skull base surgery if there were a    the observer studies. The work was supported by
risk of skull base breach. It has the potential to   the Princess Margaret Hospital Foundation and a
demonstrate skull base defects intraoperatively,     grant from the National Institutes of Health
and does not depend on the presence of CSF           (NIH R01-CA-127444).
flow during the examination. It can be com-
pleted in the operating theater without the need
of injection or patient transfer. Moreover, the      REFERENCES
CBCT provides a noninvasive, near real-time          1. Ommaya AK, Di Chiro G, Baldwin M, Pennybacker JB.
images at 1/10 of the time and radiation dose           Non-traumatic cerebrospinal fluid rhinorrhoea. J Neurol
delivered by conventional CT. Therefore, in             Neurosurg Psychiatry 1968;31:214–225.
                                                     2. Loew F, Pertuiset B, Chaumier EE, Jaksche H. Trau-
cases in which suspicion of a skull base breach         matic, spontaneous and postoperative CSF rhinorrhea.
exists, the defect can potentially be localized         Adv Tech Stand Neurosurg 1984;11:169–207.

Intraoperative Cone-Beam CT                                HEAD & NECK—DOI 10.1002/hed       April 2010   511
 3. Heinz Stammberger MH. Essentials of endoscopic sinus       11. Rafferty MA, Siewerdsen JH, Chan Y, et al. Investiga-
    surgery. 2nd ed. St. Louis, MO: Mosby Publication; 1993.       tion of C-arm cone-beam CT-guided surgery of the fron-
 4. Daudia A, Biswas D, Jones NS. Risk of meningitis with          tal recess. Laryngoscope 2005;115:2138–2143.
    cerebrospinal fluid rhinorrhea. Ann Otol Rhinol Laryn-      12. Chan Y, Siewerdsen JH, Rafferty MA, Moseley DJ, Jaf-
    gol 2007;116:902–905.                                          fray DA, Irish JC. Cone-beam computed tomography on
 5. Kerr JT, Chu FW, Bayles SW. Cerebrospinal fluid rhinor-         a mobile C-arm: novel intraoperative imaging technology
    rhea: diagnosis and management. Otolaryngol Clin               for guidance of head and neck surgery. J Otolaryngol
    North Am 2005;38:597–611.                                      Head Neck Surg 2008;37:81–90.
 6. Stone JA, Castillo M, Neelon B, Mukherji SK. Evalua-       13. Roos PG, Colbeth RE, Mollov I, et al. Multiple gain rang-
    tion of CSF leaks: high-resolution CT compared with            ing readout method to extend the dynamic range of
    contrast-enhanced CT and radionuclide cisternography.          amorphous silicon flat panel imagers. Proc SPIE 2004;
    AJNR Am J Neuroradiol 1999;20:706–712.                         5368:139–149.
 7. Shetty PG, Shroff MM, Sahani DV, Kirtane MV. Evalua-       14. Colbeth RE, Cooper VN, Gilblom DL, et al. Characteriza-
    tion of high-resolution CT and MR cisternography in the        tion of a third-generation multimode sensor panel. Proc
    diagnosis of cerebrospinal fluid fistula. AJNR Am J Neu-         SPIE 1999;3659:491–500.
    roradiol 1998;19:633–639.                                  15. Cho Y, Moseley DJ, Siewerdsen JH, Jaffray DA. Accu-
 8. Ibarra R, Jinkins JR, Korvick D, Xiong L, Gao JH. Eval-        rate technique for complete geometric calibration of
    uation of intrathecal gadolinium-enhanced MR cisternog-        cone-beam computed tomography systems. Med Phys
    raphy in a rabbit model of traumatic nasoethmoidal CSF         2005;32:968–983.
    fistula. J Magn Reson Imaging 2000;11:20–24.                16. Meco C, Oberascher G. Comprehensive algorithm for
 9. Sand T, Myhr G, Stovner LJ, Dale LG. Side effects af-          skull base dural lesion and cerebrospinal fluid fistula di-
    ter lumbar iohexol myelography. Relation to radiologi-         agnosis. Laryngoscope 2004;114:991–999.
    cal diagnosis, sex and age. Neuroradiology 1990;31:        17. Eljamel MS, Pidgeon CN. Localization of inactive cere-
    523–528.                                                       brospinal fluid fistulas. J Neurosurg 1995;83:795–798.
10. Rafferty MA, Siewerdsen JH, Chan Y, et al. Intraopera-     18. Manelfe C, Cellerier P, Sobel D, Prevost C, Bonafe A.
    tive cone-beam CT for guidance of temporal bone sur-           Cerebrospinal fluid rhinorrhea: evaluation with metrizamide
    gery. Otolaryngol Head Neck Surg 2006;134:801–808.             cisternography. AJR Am J Roentgenol 1982;138:471–476.

512   Intraoperative Cone-Beam CT                                            HEAD & NECK—DOI 10.1002/hed         April 2010

To top