Volume 49 - Issue 1.pdf - Minimal Invasive Neurosurgery 2006 - MMikulicz

O. M. Rygh 1, 2, 4

J. Cappelen 1

T. Selbekk 3, 4

F. Lindseth 3, 4

T. A. N. Hernes 2, 3, 4

G. Unsgaard 1, 2, 4

Endoscopy Guided by an Intraoperative

3D Ultrasound−Based Neuronavigation System

Abstract

Introduction

Objective: We have investigated the feasibility of using 3D ultra−

sound−based neuronavigation for guiding neuroendoscopy.

Methods: A neuronavigation system with an integrated ultra−

sound scanner was used for acquiring the 3D ultrasound image

data. The endoscope with a tracking frame attached was calibrat−

ed to the navigation system. The endoscope was guided based on

intraoperative 3D ultrasound data in 9 operations. In 5 of the

operations, ultrasound angiography data were also obtained. Up−

dated image data (e. g., more than one 3D ultrasound dataset)

were obtained in 6 of the operations. Results: We found that

the image quality of 3D ultrasound was sufficient for image

guidance of the endoscope. Planning of the entry point and

trajectory as well as finding optimal sites for fenestration were

successfully performed. Blood vessels were visualized by 3D

ultrasound angiography. In one procedure of third ventriculos−

tomy, the basilar artery was visualized. Updated image data

were quickly obtained, and in two of the cases, a reduction of

the size of cysts was demonstrated. Conclusions: 3D ultrasound

gives accurate images of sufficiently high quality for image

guidance of neuroendoscopy. Updated 3D ultrasound datasets

can easily be acquired and may adjust for brain shift. Ultrasound

angiography image data are also available with this technology

and can visualize vessels of importance.

Advances in endoscope technology have resulted in more fre−

quent use of neuroendoscopy. A number of different indications

for using this technology exist, such as third ventriculostomy

[1 ± 3], colloid cysts [4, 5], septum pellucidum cysts [6, 7], intra−

ventricular tumors [8, 9], and pituitary tumors [10] among

others. However, even with advances in neuroendoscopy, certain

limitations still exist, such as orientation of the endoscope in

abnormal anatomy, inserting the endoscope into small narrow

ventricles, visualization in opaque fluids and finding small sub−

ependymal tumor masses [11].

Several methods for solving these limitations are reported. In−

traoperative imaging with MRI [12,13] and ultrasound [14 ± 19]

has been used to give improved anatomical orientation in neu−

roendoscopic procedures. Real−time 2D ultrasound has been

used to guide neuroendoscopic procedures since the 1980s and

has been reported to give satisfactory images for guiding such

procedures [16 ± 19]. However, other authors find ultrasound

less suitable for guiding neuroendoscopic procedures [11]. Ste−

reotaxy has been used for finding the optimal entry point and

trajectory for endoscopic procedures in the ventricular system

and for third ventriculostomy [9, 20, 21], but still this method

has its practical limitations.

Key words

Ultrasound ¥ three−dimensional ultrasound ¥ neuronavigation ¥

neuroendoscopy

Neuronavigation integrated with an endoscope seems to be a

natural evolution of neuroendoscopy and is reported to give im−

proved anatomical orientation, and to facilitate the selection of

Affiliation

1 Department of Neurosurgery, St. Olav University Hospital, Trondheim, Norway

2 The Norwegian University of Science and Technology, Trondheim, Norway

3 SINTEF Health Research, Trondheim, Norway

4 National Centre for 3D Ultrasound in Surgery, St. Olav University Hospital, Trondheim, Norway

Correspondence

Ola M. Rygh, M. D. ¥ Department of Neurosurgery ¥ St. Olav University Hospital Trondheim ¥

Olav Kyrres gate 17 ¥ 7005 Trondheim ¥ Norway ¥ Tel.: +47/9284/9775 ¥ Fax: +47/7386/7977 ¥

E−mail: ola.rygh@ntnu.no

Bibliography

Minim Invas Neurosurg 2006; 49: 1±9 Georg Thieme Verlag KG Stuttgart ¥ New York

DOI 10.1055/s−2005−919164

ISSN 0946−7211

Table 1 Summary of the patients treated by endoscopic procedures with neuronavigation based on 3D ultrasound

Patient

Sex/age

Diagnosis

Procedure

No. of 3D

ultrasound

acquisitions

US angiog−

raphy

used

Preoperative 3D

MRI used in

neuronavigation

F/63 years

Unilateral hydrocephalus, post−SAH

Fenestration of septum pellucidum

F/67 years

Intracerebral cysts

Fenestration of cysts

Yes

F/73 years

Colloid cyst, hydrocephalus

Extirpation of cyst, septostomy

Yes

M/19 years

Septum pellucidum cyst

Fenestration of cyst

Yes

M/11 years

Interhemispheric cyst

Fenestration of cyst

Yes

M/3 months

Multiloculated hydrocephalus

Fenestration of cysts

2 years

Multiloculated hydrocephalus

Fenestration of cysts

Yes

M/25 years

Septum pellucidum cyst

Fenestration of cyst

M/33 years

Hydrocephalus, aqueduct stenosis

Third ventriculostomy

Yes

the optimal entry point and trajectory for an endoscopic proce−

dure [7,11, 20, 22, 23]. However, neuronavigation systems based

on preoperative images in general may suffer from inaccuracy

due to registration errors and brain shift [24 ± 28].

A custom−made insert (see Fig. 3C) for the fixation adapter was

used to make the plane of the tracking frame and the endoscope

parallel, necessary for correct calibration for neuronavigation.

Fixation of the endoscope was accomplished with a single−arm

fixation device (Aesculap, Tuttlingen, Germany).

Ultrasound image quality has improved considerably in the last

decades due to technological development and application adap−

tation [29, 30]. The combination of 2D ultrasound and tracking

technology enables freehand 3D ultrasound reconstruction. This

gives a 3D ultrasound dataset that can be used in a similar man−

ner as 3D MRI datasets are used in a conventional neuronaviga−

tion system [29, 30]. In addition, with 3D ultrasound, the inaccu−

racy caused by patient registration is avoided and updated in−

traoperative images can be acquired whenever needed during

surgery. This technology has been reported to solve some of the

challenges of navigated tumor surgery and vascular neurosur−

gery [29 ± 31]. In the present study, we have investigated the fea−

sibility of using intraoperative 3D ultrasound for image guidance

in neuroendoscopy. We here report our experience with this

technology in 9 procedures.

Ultrasound and neuronavigation system

An ultrasound−based intraoperative imaging and neuronaviga−

tion system (SonoWand , Mison, Trondheim, Norway) was

used. This system may be used as an ultrasound scanner, a con−

ventional neuronavigation system, or an integrated, ultrasound−

based neuronavigation system that uses the features of both

technologies [30]. The system is based on optical tracking tech−

nology, and comes with a 4 ± 8 MHz flat phased array ultrasound

probe, precalibrated for image data acquisition and optimized for

neurosurgery and with optimal resolution at depths of 3 ± 6 cm

[32]. A patient reference frame is attached to the head−holder

(Mayfield frame). In one case the patient reference frame was at−

tached to the operating table, because the young age of the pa−

tient (3 months) prohibited the use of a head−holder. For tracking

of the ultrasound probe, a tracking frame is attached to the

probe. Neuronavigation can be performed with a tracked pointer

device, or any surgical instrument with an attached tracking

frame after calibration. The trajectory and position of the pointer

or surgical tool tip is displayed as a line with crosshairs or a dot

(depending on the version of the neuronavigation software) in

the corresponding images. The pointer and surgical tool can be

virtually elongated using an offset feature. The ultrasound probe

is tilted or translated over the area of interest by free hand move−

ment and the 2D ultrasound images acquired are reconstructed,

making a 3D ultrasound dataset ready for navigation (Fig. 1B).

The procedure takes about one minute. The ultrasound probe is

not in the operating field after the acquisition of a 3D ultrasound

volume, unless additional 3D ultrasound datasets are acquired or

real−time 2D imaging is needed. The workflow of this procedure

and set−up of equipment are summarized in Fig. 1.

Patients and Methods

Patients

Between January 2003 and February 2005, 9 procedures in 8 pa−

tients were performed using the described system. (Two proce−

dures were performed on patient 6 in Table 1, at ages 3 months

and 2 years). There were 3 female patients. The age of the pa−

tients ranged from 3 months to 72 years. The patients are sum−

marized in Table 1.

Endoscopic equipment

We used a rigid operating endoscope (Aesculap, Tuttlingen, Ger−

many) of 6 mm outer diameter and one working channel, one ir−

rigation channel, one overflow channel as well as the optic chan−

nel. A single chip video camera was mounted on the eyepiece for

visualization and connected to a monitor.

Navigation based on 3D ultrasound can be performed directly

without any patient registration with fiducials, since the images

are acquired in the same coordinate system as navigation is per−

formed. Therefore, no patient or image registration error will af−

fect the overall accuracy of the system when navigation based on

An optical tracking frame (Mison, Trondheim, Norway), usually

used with the CUSA, was used for tracking the endoscope. The

tracking frame was attached to the fixation adapter for the trocar.

Rygh OM et al. Endoscopy Guided by an Intraoperative º Minim Invas Neurosurg 2006; 49: 1 ± 9

Fig. 1 Neuroendoscopy guided by 3D ul−

trasound. A The probe is placed on the dura.

The anatomy and pathology are visualized

by real−time 2D ultrasound. B During image

acquisition, the probe with tracking frame is

tilted or translated over the area of interest.

A 3D ultrasound dataset is reconstructed

from a stack of 2D images. C The calibrated

endoscope with the tracking frame (arrow)

is tracked by the navigation system and the

tip position and trajectory of the endoscope

are displayed on the neuronavigation

screen. The patient reference frame is

attached to the head−holder.

3D ultrasound is performed [24]. If neuronavigation based on

preoperative MRI also is desired, patient registration of course

will be necessary.

3D ultrasound angiography image data were acquired. The track−

ing frame was attached to the fixation adapter for the trocar,

using the custom−made insert (Fig. 3C). The endoscope was then

calibrated to the navigation system by directing the tip of the en−

doscope at a small hole in the centre of the reference frame (Fig.

3D). Based on the ultrasound images, and using the offset feature

to display the trajectory of the endoscope, the best entry point

and trajectory for the endoscope were decided, and a small open−

ing in the dura was made. The endoscope was then inserted with

image guidance (Fig. 3E). The surgical procedure planned, for ex−

ample, fenestration of a cyst, was then performed with the live

video images from the endoscope, as well as image guidance. If

updated images were required, additional 3D ultrasound data−

sets were obtained during the procedure. In one operation, real−

time ultrasound was also used as a method for confirming the

flow of CSF through the fenestration, using power Doppler.

The display modalities available are: 1) Orthogonal slicing: three

2D slices are oriented as axial, sagittal and coronal slices (Fig.

2A). 2) Anyplane slicing: One slice defined by the axis and rota−

tion of the pointer or custom calibrated tool (for example, the en−

doscope) (Fig. 2B). A plane perpendicular to the anyplane can

also be added (dual anyplane) (Fig. 2C). With a tracking frame at−

tached, the endoscope tip and trajectory defines the slicing of the

image volume (Fig. 2D).

The overall accuracy of 3D ultrasound−based navigation using

the SonoWand

system in a clinical setting may be as good as

2 mm [24].

Operative technique

All the operations were performed with neuronavigation based

on 3D ultrasound. In two of the operations, preoperative MRI

also was available for navigation.

Results

Nine procedures were performed on 8 patients during this study

(Table 1). We found the image quality of ultrasound to be satis−

factory for image guidance in all cases. The ventricles as well as

pathological cysts were clearly outlined and we found that the

ultrasound images gave sufficient information for: 1) deciding

the optimal entry point and trajectory for inserting the endo−

scope, especially in cases with small narrow ventricles; 2) anato−

The position of a small craniotomy (diameter approximately

4 cm) was chosen based on preoperative CT or MR images (but

not neuronavigation) (Fig. 3A). After cleaning the dura for blood

and bone debris, sterile ultrasound gel was applied on the dura,

and 3D ultrasound data were acquired (Fig. 3B). If required, also

Rygh OM et al. Endoscopy Guided by an Intraoperative º Minim Invas Neurosurg 2006; 49: 1 ± 9

Fig. 2 Display techniques. A Orthogonal

slices: Three orthogonal 2D slices from each

3D volume oriented as axial, sagittal and

coronal slices. B Anyplane slices: Anyplane

slices are defined by the trajectory and rota−

tion of the pointer or surgical tool. C Dual

anyplane slices: In dual anyplane mode an

additional plane perpendicular to the first

plane is added. D The endoscope or any sur−

gical tool with a tracking frame, when cali−

brated, works as a pointer, for example, with

orthogonal slices.

mical orientation during endoscopy, thus facilitating the choice

of sites for fenestration of cysts; 3) quality control at the end of

the procedure, excluding bleeding and in two cases verifying

the reduction of size of cysts (patients 2 and 4 in Table 1).

Illustrative Cases

Patient 4

A man aged 19 years at the time of the operation, had episodes of

loss of consciousness. A septum pellucidum cyst was found on CT

and MRI investigations (Fig. 4A). An endoscopic fenestration of

the cyst to the ventricular system using navigated 3D ultrasound

was planned. We did not have preoperative 3D MR images for

this procedure, so only 3D ultrasound images were used for navi−

gational guidance of the endoscope. 3D ultrasound images were

acquired after making a mini−craniotomy. The ultrasound images

were of sufficient quality for inserting the endoscope with navi−

gation guidance into a narrow right ventricle (Fig. 4C). The fenes−

tration was done using conventional technique. The foramen of

Monro was observed to be occluded by the cyst wall. Updated

3D ultrasound acquired after removing the endoscope demon−

strated that the septum pellucidum cyst already was reduced

somewhat in size (Fig. 4D). It also showed the canal after removal

of the endoscope, with no signs of bleeding. A postoperative CT

taken the following day showed that the cyst was reduced in

size (Fig. 4B). The postoperative recovery was uneventful, and at

2 months follow−up, the patient had not had any new episodes of

syncope.

Even though we did not measure the clinical accuracy systemati−

cally, the general impression was that the accuracy was satisfac−

tory. This could be confirmed by approaching a recognizable

structure (for example, the foramen of Monro) with the endo−

scope and by comparing with the position of the endoscope tip

displayed on the navigation screen.

In five of the operations ultrasound angiography also was ac−

quired. We found that blood vessels were clearly visualized in

all these cases. This was found to give particularly useful infor−

mation in two cases: In patient 6, vessels in septa of multiloculat−

ed hydrocephalus were visualized (see Fig. 6B), while in patient 8

the basilar artery could be visualized (see Fig. 6A).

In six operations updated 3D ultrasound image data were ob−

tained, and in two cases (cases 2 and 4), the updated ultrasound

images could demonstrate the reduction of the size of a cyst. Up−

dated ultrasound image data were acquired quickly; it typically

took about a minute.

Seven of the operations were performed with only 3D ultrasound

data for navigation. In two procedures we had preoperative 3D

MRI with fiducials available in addition to the 3D ultrasound

data (patients 5 and 8 in Table 1).

Patient 5

This boy of 11 years of age at the time of the operation had been

investigated because of tics in his face and arms. An interhemi−

spheric cyst was found on investigations with CT and MRI, as

well as an agenesis of the corpus callosum (Fig. 5A and Fig. 5B).

An endoscopic fenestration of the cyst using navigated 3D ultra−

sound was planned. Preoperative 3D MRI with fiducials was

available; consequently patient registration was done in this

case. Ultrasound images were acquired through a mini−craniot−

omy before opening the dura and showed the interhemispheric

cyst and the small ventricles clearly, in our opinion just as good

as the MR images in the navigation system (Fig. 5D and Fig. 5E).

We chose to display both the ultrasound and the MR images on

All procedures were successful, e. g., the goal of the procedure,

such as fenestration of a cyst, was reached. One patient (patient

1 in Table 1) had postoperative meningitis, but recovered from

this. We did not record any other complications at 1 ± 2 months

follow−up. The patients are summarized in the Table 1. Two illus−

trative cases (patients 4 and 5 in Table 1) are presented in detail.

Rygh OM et al. Endoscopy Guided by an Intraoperative º Minim Invas Neurosurg 2006; 49: 1 ± 9

Fig. 3 Operative technique.

A A mini−craniotomy, about 4 cm in dia−

meter is made. B The ultrasound probe with

a tracking frame is placed on the dura for

imaging. C A tracking frame (in this case

without marker spheres) is attached to the

fixation adapter for the endoscope trocar.

A custom−made insert (arrows) ensures that

the plane of the tracking frame and the

endoscope are parallel. D The neuroendo−

scope with the tracking frame attached is

calibrated by directing the tip of the endo−

scope at a small hole in the centre of the

reference frame. E With the tracking frame

attached, the neuroendoscope is inserted

with image guidance, based on the 3D

ultrasound image data.

the navigation screen during the procedure. However, we based

the operation on intraoperative 3D ultrasound knowing that

these volumes were not affected by registration errors, and also

that the inaccuracy due to brain shift was minimal since the 3D

ultrasound was acquired just before the procedure. Neuronavi−

gation was used to guide the endoscope into the small right lat−

eral ventricle (Fig. 5D). We noticed a small difference (approxi−

mately 4 mm) between the position of the endoscope tip dis−

played on the MRI and the ultrasound images on the navigation

screen (Fig. 5E). This difference we assumed to be caused by

brain shift or registration error of the MR images. A fenestration

of the cyst to the ventricle was done, as well as a fenestration

from the cyst to the basal cistern. The postoperative recovery

was uneventful, and a postoperative CT demonstrated that the

cyst was reduced in size (Fig. 5C). On follow−up after two months

the patient’s symptoms had improved considerably and he was

performing better in school.

Discussion

Several ultrasound solutions have been proposed in neuroendo−

scopy. Real−time ultrasound has been used as an adjunct to neu−

roendoscopic procedures [15 ± 18], but it has not gained popular−

ity. This is probably due to old experiences of low image quality,

as well as the fact that the probe has to be kept in the operation

field for acquiring real−time images, a somewhat inconvenient

concept.

More modern ultrasound solutions have also been explored.

Resch et al. [16,17] developed a technique where a small sono ca−

theter (originally developed for intravascular use) is inserted

into the working canal of an endoscope, giving a 360−degree ax−

ial (to the endoscope) real−time view of the anatomy, like a

“mini−CT”. The solution gives good images of the anatomy in a

plane axial to the endoscope, but no image ahead of the endo−

scope. 3D ultrasound datasets acquired on pediatric patients

have been used for simulating virtual neuroendoscopies pre−

Rygh OM et al. Endoscopy Guided by an Intraoperative º Minim Invas Neurosurg 2006; 49: 1 ± 9

Volume 49 - Issue 1.pdf

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: