Eur Arch Otorhinolaryngol (2000) 257 : 578–583
MISCELLANEOUS
Ping Han · Wolfgang Pirsig · Frank Ilgen ·
Johannes Görich · Roman Sokiranski
Virtual endoscopy of the nasal cavity
in comparison with fiberoptic endoscopy
Received: 9 December 1999 / Accepted: 7 September 2000
Abstract Purpose: To apply virtual-endoscopic mode to
display the nasal cavity and to evaluate clinical applica-
tion of virtual endoscopy in comparison with nasal
fiberoptic endoscopy. Materials and methods: Eleven pa-
tients were examined by virtual endoscopy after axial spi-
ral computed tomography (CT) scanning was performed.
In addition, 9 out of these 11 patients underwent fiberop-
tic endoscopy. Spiral CT scanning was performed with
1-mm collimation, 0.5- to 1-mm increment, 25–250 mA s,
and pitch 1–2. Virtual endoscopy was performed by Ex-
plorer software package. Results: Virtual endoscopy could
clearly demonstrate anatomic structures in the nasal cav-
ity, septal deviation, stenosis and obstruction of the mid-
dle meatus, turbinate hyperplasia, and pathological
masses larger than 3 mm in diameter. However, “false ad-
hesions” may appear in virtual endoscopy. The main lim-
itation of virtual endoscopy was the inability to evaluate
mucosa and its surface. Conclusion: Virtual endoscopy of
the nasal cavity is a new and noninvasive method. It can
demonstrate normal and pathological structures in the
nasal cavity. Its ability to visualize is comparable with
fiberoptic endoscopy except for evaluating mucosal sur-face
and secretions. In the future, this method will proba-bly be a
basic instrument of computer-assisted surgery in the
midfacial region.
Keywords Low-dose spiral CT · Virtual endoscopy ·
Fiberoptic endoscopy · Nasal cavity
P. Han · J. Görich · R. Sokiranski ( ) Kreiskliniken
Traunstein – Trostberg GmbH, Department of
Radiology, University of Ulm, Cuno-Niggl-Str. 3,
83278 Traunstein, Germany e-mail: khts-
sokiranski@kktstb.mhn.de,
Tel.: +49-861-7051291, Telefax: +49-861-7051787
W. Pirsig · F. Ilgen
Department of Otorhinolaryngology, University of Ulm,
Traunstein, Germany
© Springer-Verlag 2000
Introduction
Spiral computed tomography (CT) overcomes some short-
comings of conventional CT. The advantage of volume
acquisition makes it possible to retrospectively recon-
struct overlapping images at arbitrary increments, which
can provide high-quality three-dimensional (3D) recon-
struction [1]. The most recent development in spiral CT is
the use of 3D endoscopic mode to display endoluminal
diseases, for example, simulation of bronchoscopy, an-
gioscopy, and cystoscopy [2–6]. Virtual endoscopy (VE)
allows the operator to demonstrate dynamically the inte-
rior of the lumen organs using “fly” function. Therefore, it
may assist in surgical planning, especially in computer-as-
sisted surgery (CAS), and medical education and training
[7–9]. In order to elucidate the ability of VE to display
structures within the nasal cavity, we applied virtual-en-
doscopic mode to display the nasal cavity and evaluated
clinical application of VE in comparison with nasal
fiberoptic endoscopy (FE) in the study.
Materials and methods
Patients
Eleven patients (9 men and 2 women) were examined between August
1996 and December 1996 in the Department of Radiology of the
University of Ulm. One was an newborn girl, one was a 75-day-old
infant, and 9 patients were 23–71 years old. Nine pa-tients suffered
from chronic sinusitis and/or polyposis, two pa-tients had nasal tumor
diagnosed by clinical examination and CT scan. Six patients with
sinusitis underwent endoscopic sinus sur-gery, and two patients with
tumors underwent tumor excision.
Informed consent was obtained from all patients in accordance with
the 1964 Declaration of Helsinki; the examined patients agreed that the
data derived from them was to be analyzed without personal
identification.
Axial spiral CT scans
Eleven patients (22 nasal cavities) were scanned with a CT double-
detector scanner (Elscint CT-Twin Flash; Elscint, Haifa, Israel) be-
 579

fore FE was performed after 1–2 days. Patients were placed supine on – Score 0. Structures could not be seen.
the scanner bed. The scan plane was parallel to the hard palate and the In addition, artifacts in VE images were also analyzed by compar-ison
study coverage extended from the frontal sinus to the max-illary teeth. with sectional CT images. If a tissue defect was not demon-strated in
Scan time varied between 20 and 51 s. A 120-kV volt-age was used for the original CT images, but it appeared in VE images, this tissue defect
adults; 90 kV was used for the newborn and the infant. Patients were was considered as an artifact (pseudoforamen).
scanned with 1-mm collimation (1.3-mm slice thickness), 0.5- to 1-mm The VE images were independently evaluated by two radiolo-gists
increment. Four patients (including children) were scanned with 25– and a rhinologist. A video of FE was also separately assessed by a
100 mA s and pitch 2, seven pa-tients with 200–250 mA s and pitch 1. radiologist and two rhinologists.
Contrast medium (Ultravist 300) was intravenously adminis-tered
in two children with tumors. The newborn was injected with a total of
8 ml at a rate of 0.2 ml/s with a power injector. A 40-s delay was used Results
between the onset of the injection and the start of scanning. The
contrast medium was administered to the infant at a rate of 0.5 ml/s, for Anatomical structures displayed by VE and FE
12 ml with a 24-s delay. Nose drops were used for three patients prior
to CT scanning to eliminate reversible nasal congestion and mucus. Nine patients with chronic sinusitis were examined by both
VE and FE. Three patients (four nasal cavities) had a history
of endoscopic sinus surgery. Two children with tu-mors
Virtual endoscopy
were only examined by VE; therefore VE was only
Spiral CT data were used to produce VE images. VE imaging con-sists compared with FE in nine patients (Table 1).
of the following four steps: (1) image acquisition; (2) data transfer; (3) VE was performed in all 11 patients. It clearly demon-
VE imaging; and (4) real-time 3D rendering. strated the septum in 16 of 18 nasal cavities and the
After patients were scanned, digital CT images were trans-ferred
from the spiral CT scanner to a graphics workstation (Indigo 2; Silicon epipharynx in 17 of 18 nasal cavities (Table 1). The
Graphics) via a network. The time for the transfer of CT images took turbinates and nasal meatuses were clearly displayed in
approximately 15 min. During transfer, window width and window
level of view for CT images were respectively set at 700–1000 HU and
70–100 HU. Table 1 Numbers and scores of the observed structures by fiberoptic
VE imaging and real-time 3D rendering (i.e., VE) were per-formed endoscopy (FE) and virtual endoscopy (VE) in 18 nasal cavities
using a volume-visualization program, IRIS Explorer (Sil-icon
Graphics). Explorer used a simple threshold method and a “marching
cubes” algorithm [10] to form an isosurface (a surface of equal voxel FE VE
value) of an organ. The threshold was selected for each patient by 2 1 0 2 1 0
comparing an original CT image with the modified image (“secondary
image”), which would comprise the VE image. The time taken for VE Septum 17 1 0 16 2 0
imaging with Explorer was approximately 2 min, which depended on
2D data volume. Inferior meatus 17 1 0 15 3 0
After a VE image was created, its color and light were adjusted by Middle meatus 11 5 2 12 4 2
using editing functions. The interior of the nasal cavity was dy- Inferior concha 16 1 1 15 2 1
namically demonstrated using the fly function. The turbinates, nasal Middle concha 14 3 1 14 3 1
meatus, and ostiomeatal complex of both nasal cavities and their Uncinate process
a 8 2 4 9 0 5
pathological changes were examined. The VE of all patients was a 5 2 7 8 0 6
Ethmoidal bulla
analyzed real-time on the computer monitor and recorded to videotape,
Maxillary ostium 6 0 12 4 0 14
so that it could be compared with FE. Important anatomic structures
Epipharynx 18 0 0 17 1 0
and characteristic findings of pathology were photographed.
a The structure in four nasal cavities was operatively resected
Preoperative endoscopy
Table 2 Pathological changes detected in nine patient (18 nasal
Nine patients underwent endoscopy before surgery in the Depart-ment cavities) on fiberoptic endoscopy (FE) and virtual endoscopy (VE)
of Ear, Nose and Throat in the University of Ulm. The ex-amination
was performed under local anesthesia. The endoscopes were introduced Pathology FE VE
through the nostrils. Each nasal cavity was exam-ined with 0° and 70° Septum (n = 9) Deviation 7 7
scope, and findings were recorded to video-tape. Observed structures Spur 5 5
included the septum, inferior and middle turbinates, inferior and middle
Perforation 1 1
meatus, uncinate process, eth-moidal bulla, ostium of the maxillary
sinus, and pathological changes at these sites. Adhesion 2 3
Meatus (n = 18) Narrowing 6 6
Obstruction 3 3
Imaging analysis a Hyperplasia 6 6
Turbinate (n = 36)
The septum, epipharynx, middle and inferior conchae, middle and Paradox-curvature 1 1
inferior meatus, uncinate process, ethmoidal bulla, and ostium of the Polyps (n = 18) 6 5
maxillary sinus were observed. The visualization of these structures
throughthe VE and FE was classified with scores: b 14 0
Inflammation (n = 18)
– Score 2. Structures could be clearly displayed, i.e., the structures
were seen in their entirety, with sharp, well-defined margins. a Including inferior and middle turbinates in 18 nasal cavities
– Score 1. Structures could be partly visualized or had blurred margins b Inflammation in the nasal cavity, consisting of mucosal
and were not easily recognized. conges-tion and edema
580

a
a

Fig. 1 a, b Septal spur in the right nasal cavity displayed by fiberoptic

endoscopy (FE) (a) and virtual endoscopy (VE) (b). The septal spur b
extends to middle concha and adheres to its inferior edge. (Sp spur, MC
middle concha)
Fig. 2 a, b A small polyp in right nasal cavity displayed by FE (a)
and VE (b). (Po Polyp, MC middle concha, S septum)
12–15 of 18 nasal cavities. VE clearly displayed the unci-
nate process, ethmoidal bulla, and ostium of maxillary si-nus
played: 10 of 14 (71.4%) uncinate processes, 7 of 14 (50%)
in 9 of 14 (64.3%), 8 of 14 (57.1%), and 4 of 18 (22.2%)
ethmoidal bullae, and 6 of 18 (33.3%) ostia of the maxillary
nasal cavities, respectively.
sinuses.
FE clearly displayed the septum in 17 of 18 nasal cav-
ities and epipharynx in all nasal cavities, and could also
reveal the swallowing movement of epipharynx. The Pathological changes in VE and FE
turbinates and nasal meatuses were clearly displayed in 11–
17 of 18 nasal cavities (Table 1). If nasal cavities with VE could display patients’ septal deviation, nasal meatus
scores 2 and 1 are added together, the following were dis- narrowing and obstruction, turbinate hyperplasia, and

b
Fig. 3 a, b A large polyp occupying the right middle meatus dis-played
by FE (a) and VE (b). The VE image was obtained by spi-ral CT
scanning with 25 mA s. (S septum, Po polyp, LW lateral wall)
masses within the nasal cavity with varying frequency
(Table 2).
Septal deviation was detected in seven patients by VE. In
five cases, septal spurs were displayed. In two cases, the
spurs were very large and extended to the turbinates (Fig. 1).
There were adhesions between septum and turbinate at four
sites in three nasal cavities.
581
b
Fig. 4 a, b The nasal tumor in the infant displayed by VE (a) and spiral
CT scanning (b). The VE image shows a mass bulging into right nasal
cavity (Tu), resulting in the obstruction of the inferior nasal meatus.
Black arrow shows the rest of the nasal cavity. Spi-ral CT imaging
shows that the mass is of soft tissue density and presses surrounding
tissue structures (arrow)
Using VE, both nasal tumors and polyps appeared as
pathological masses. When the lesions were small, they
were shown as masses bulging into the nasal cavity (Fig. 2).
If the lesions were large, they made the nasal cav-ity
extremely narrow or obstructed. Nine polyps in four patients
(five nasal cavities) were shown by VE. Two polyps were
approximately 12–20 mm in diameter; they
582

occupied the whole middle meatus and put pressure on

surrounding tissues (Fig. 3). The remaining polyps were in
the middle meatus and were between 3 and 9 mm. In the
infant with a tumor, the frontal part of the right nasal cav-ity
was extremely narrow (Fig. 4) and the left nasal cavity of
the newborn was completely obstructed by a soft tissue
mass.
Eleven polyps in five patients (six nasal cavities) were
displayed by FE. The locations and sizes of the polyps were
equal to those detected by VE. Two polyps that were not
seen under VE were smaller than 3 mm in diameter. Visual
pathological changes in VE and FE are summa-rized in
Table 2.

Artifacts

Artifacts appeared in VE images in three out of the nine

patients; they were easily distinguished. The artifacts
showed up as very small defects (pseudoforamina), with
irregular squares or triangles, and usually appeared at the a site
where the curve of nasal contour changed sharply, such as the
margin of the anterior nasal aperture and walls
of ethmoidal cells. In this study, artifacts did not affect
evaluation of pathological changes because they were of
special shape and very small.

Discussion

Methodology

VE is a noninvasive examination without additional X-ray

load. It can be applied to patients of all ages provided they
can undergo a CT scan. VE is a postprocessing procedure
and it can be performed without the presence of the pa-
tients. The time taken in VE imaging, i.e., time creating a
VE image with Explorer, is approximately 5 min.
VE can demonstrate the macroanatomic structures of
nasal cavity and intranasal masses larger than 3 mm in di-
ameter and can precisely show whether the nasal meatus is
obstructed or narrowed.
The spatial resolution of CT is not dependent on a high
dose of radiation for a high-contrast area such as in the
paranasal sinus [11, 12]. Axial and coronal imaging of si-
nuses [11, 12] and 3D bone reconstruction of maxillofa-cial
complex [13] were explored by conventional CT with a low
radiation dose. The results in these studies showed that the b
low dose of 60–80 mA s had no significant influ-ence on 3D
bone image, but it slightly increased the noise on section Fig. 5 a, b The false adhesion on a VE image. The paradoxical
images in the soft tissue portions. The results with 25-mA s curvature of the middle concha can be seen under FE (a) and on VE
spiral CT scanning in the present study are in accordance image (b), but adhesion between the septum and left middle concha can
with these previous findings. only be seen on VE image (arrow). (MC middle con-cha, Sp spur, S
septum)

Pathway of endoscopy tril and from the choanae. FE can also show the posterior
nose, but only via the mouth [14]. In addition, the virtual
There are at least two pathways to enter the nasal cavity and endoscope can penetrate walls of sinuses without injury.
paranasal sinuses in virtual endoscopy, from the nos- However, it is very difficult or usually impossible for a

fiberoptic endoscope to penetrate paranasal sinuses with-out
injury [14, 15].
Anatomic visualization of the nasal cavity
Data in Table 1 show that VE was comparable with FE in
the ability to display structures of the nasal cavity such as
septum, turbinates, and middle meatus. There are, how-ever,
some differences between VE and FE concerning the ability
to visualize these structures.
It is very difficult for FE to enter a narrow or ob-structed
middle meatus. Thus, the ethmoidal bulla and un-cinate
process cannot or can only partly be seen. VE can “fly” into
an very narrow middle meatus even if the mea-tus is only 1–
2 mm wide. Therefore, in the present study, display rates of
score 2 for the uncinate process (9 of 14) and ethmoidal
bulla (8 of 14) with VE were slightly higher than those (8 of
14 and 5 of 14) with FE. However, if the middle meatus was
completely filled by soft tissue, that is, no air-filled lumen
existed in the VE image, VE can fly through the obstructed
meatus, but an image can-not be seen. This is because
Explorer uses a threshold method and a marching-cubes
algorithm [10] to form an isosurface of an organ. It is able to
extract the air-filled lu-men from a nose where the contrast
difference between air and surrounding soft tissue is on the
order of 1,000 : 1, but it is much more difficult to extract
lymph nodes or tu-mor from surrounding soft tissues where
the contrast dif-ference is less. In the study group, three
nasal cavities were obstructed and four nasal cavities had a
history of endoscopic sinus surgery. Therefore, display rates
of the uncinate process and ethmoidal bulla with VE were
not so high.
Artefacts and pitfalls of VE
VE can clearly display pathologies in the nasal cavity such
as septal deviation and perforation, meatal stenosis and
obstruction, and turbinate hyperplasia, as well as most nasal
masses. There is, however, false positivity, e.g., “false
adhesion” and false negativity in VE.
VE showed adhesions between the septum and the
turbinates at four sites in three nasal cavities, while FE
showed only two adhesions in two nasal cavities, which
were identical in location to that observed by VE. The other
two adhesions, displayed by VE but not by FE, were
considered as false adhesion (Fig. 5). The cause of the false
adhesion was mucus in the nasal cavity. Mucous se-cretion
attached to the surface of turbinate and septum presented as
soft tissue density on CT imaging. Therefore,
583
the mucus was reconstructed as soft tissue in VE images. If
the mucus is attached between the septum and turbinate, a
false adhesion appears on the VE image. One method to get
rid of the false adhesion is to administer de-congestive nasal
spray to patients prior to CT scanning.
VE cannot find masses smaller than 3 mm. FE demon-
strated 11 nasal polyps, while VE found only nine nasal
polyps. Two polyps, whichwere not seen, were smaller than
3 mm in diameter.
Data from Table 2 show a good correspondence of vi-
sualizing of all nasal structures except of mucosal inflam-
mation. Inflammation was present in 18 nasal cavities and
was visualized by FE in 14 cavities, but in no cavity by VE.
References
1. Heiken JP, Brink JA, Vannier MW (1993) Spiral (helical) CT.
Radiology 189:647–656
2. Kimura F, Shen Y, Date S, Azemoto S, Mochizuki T (1996)
Thoracic aortic aneurysm and aortic dissection: new endo-scopic
mode for three-dimensional CT display of aorta. Radiol-ogy
198:573–578
3. Rubin GD, Beaulieu CF, Argiro V, Ringl H, Norbash AM, Feller
JF, Dake MD, Jeffrey RB, Napel S (1996) Perspective volume
rendering of CT and MR images: applications for en-doscopic
imaging. Radiology 199:321–330
4. Davis CP, Ladd ME, Romanowski BJ, Wildermuth S, Kno-plioch
JF, Debatin JF (1996) Human aorta: preliminary results with virtual
endoscopy based on three-dimensional MR imag-ing data sets.
Radiology 199:37–40
5. Vining DJ, Liu K, Choplin RH, Haponik EF (1996) Virtual
bronchoscopy: relationships of virtual reality endobronchial
simulations to actual bronchoscopic findings. Chest 109:549– 553
6. Fenlon HM, Bell TV, Ahari HK, Hussain S (1997) Virtual cys-
toscopy: early clinical experience. Radiology 205:272–275
7. Carney AS, Baldwin D, Patel N, Sandeman D (1994) Virtual reality
surgery in otorhinolaryngology. BMJ 309:1233
8. Coleman J, Nduka CC, Darzi A (1994) Virtual reality and la-
paroscopic surgery. Br J Surg 81:1709–1711
9. Ossoff RH, Reinisch L (1994) Computer-assisted surgical tech-
niques: a vision for the future of otolaryngology head and neck
surgery. J Otolaryngol 23:354–359
10. Lorensen WE, Cline HE (1987) Marching cubes. A high reso-lution
3D surface construction algorithm. Comp Graph 21: 163–169
11. Duvoison B, Landry M, Chapuis L, Krayenbuhl M, Schnyder P
(1991) Low-dose CT and inflammatory disease of the paranasal
sinuses. Neuroradiology 33:403–406
12. Marmolya G, Wiesen EJ, Yagan R, Haria CD, Shah AC (1991)
Paranasal sinuses: low-dose CT. Radiology 181:689–691
13. Gholkar BA, Gillespie JE, Hart CW, Mott D, Isherwood I (1988)
Dynamic low-dose three-dimensional computed tomog-raphy: a
preliminary study. Br J Radiol 61:1095–1099
14. Messerklinger W (ed) (1978) Endoscopy of the nose. Urban and
Schwarzenberg, Baltimore, pp 2–6
15. Maran AGD, Lund VJ (eds) (1990) Clinical rhinology. Thie-me,
Stuttgart, pp 22–25, 54–57