Chest ACCP Education Calendar
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Guest Access | Sign In via User Name/Password
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Article Archive
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (19)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Solomon, S. B.
Right arrow Articles by Wang, K. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Solomon, S. B.
Right arrow Articles by Wang, K. P.
(Chest. 2000;118:1783-1787.)
© 2000 American College of Chest Physicians

Three-dimensional CT-Guided Bronchoscopy With a Real-Time Electromagnetic Position Sensor*

A Comparison of Two Image Registration Methods

Stephen B. Solomon, MD; Peter White Jr., MD; Charles M. Wiener, MD, FCCP; Jonathan B. Orens, MD and Ko Pen Wang, MD

* From the Departments of Cardiovascular and Interventional Radiology (Dr. Solomon) and Pulmonary Medicine (Drs. White, Wiener, Orens, and Wang), Johns Hopkins School of Medicine, Baltimore, MD.

Correspondence to: Stephen Solomon, MD, Department of Radiology, Blalock 545, Johns Hopkins School of Medicine, 600 N Wolfe St, Baltimore, MD 21287; e-mail: ssolomo{at}jhmi.edu


   Abstract
TOP
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Study objectives: To compare two different image registration methods for accurately displaying the position of a flexible bronchoscope on a previously acquired three-dimensional CT scan during bronchoscopy.

Setting: Bronchoscopy suite of a university hospital.

Patients: Fifteen adult patients scheduled for nonemergent bronchoscopy.

Methods: A miniature electromagnetic position sensor was placed at the tip of a flexible bronchoscope. Previously acquired three-dimensional CT scans were registered with the patient in the bronchoscopy suite. Registration method 1 used multiple skin fiducial markers. Registration method 2 used the inner surface of the trachea itself for registration. Method 1 was objectively assessed by measuring the error distance between the real skin marker position and the computer display position. Methods 1 and 2 were subjectively assessed by the bronchoscopist correlating visual bronchoscopic anatomic location with the computer display position on the CT image.

Results: The error distance (± SD) from known points for registration method 1 was 5.6 ± 2.7 mm. Objective error distances were not measured for method 2 because no accurate placement of the bronchoscope sensor could be correlated with CT position. Subjectively, method 2 was judged more accurate than method 1 when compared with the fiberoptic view of the airways through the bronchoscope. Additionally, method 2 had the advantage of not requiring placement of fiducial markers before the CT scan. Respiratory motion contributed an error of 3.6 ± 2.6 mm, which was partially compensated for by a second tracking sensor placed on the patient’s chest.

Conclusion: Image registration method 2 of surface fitting the trachea rather than method 1 of fiducial markers was subjectively judged to be superior for registering the position of a flexible bronchoscope during bronchoscopy. Method 2 was also more practical inasmuch as no special CT scanning technique was required before bronchoscopy.

Key Words: bronchoscopy • computers • CT • image-guided surgery • stereotaxy


   Introduction
TOP
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Although CT and MRI have helped radiologists and bronchoscopists with prebronchoscopic planning, application of cross-sectional imaging to the actual bronchoscopic procedure has been limited. Although techniques such as virtual bronchoscopy may help the bronchoscopist with prebronchoscopic planning, no technique has effectively linked the bronchoscope with the CT images to aid the procedure itself. Inasmuch as bronchoscopists are primarily guided by the view from the bronchoscope, they are limited to an intraluminal view of the airway. This creates challenges during transbronchial needle aspiration (TBNA) when

the bronchoscopist must take a biopsy of an extraluminal lesion. In fact, the positive diagnostic yield for TBNA is variable, with some reports as low as 37%.1 2 3 4

Another challenge for the bronchoscopist is performing transbronchial forceps biopsies. In these cases, the bronchoscopist relies on C-arm fluoroscopy for guidance. However, because the lesion is in three-dimensional space, using two-dimensional fluoroscopy can be misleading.

Image navigation systems using CT and MRI data have been widely used in neurosurgery.5 6 However, as most of these systems detect the position of the hand-held portion of a surgical tool and then extrapolate to the tip of the instrument, they require an inflexible, rigid instrument. These systems using light emission from the surgical instrument for position determination would not be applicable to flexible bronchoscopy.

We and others7 8 have reported the use of a miniature electromagnetic position sensor that can be placed at the tip of flexible instruments such as endoscopes and catheters. This sensor detects both real-time position and orientation and then displays it on a previously acquired three-dimensional CT scan during bronchoscopy in the bronchoscopy suite. Crucial to the technique is the method of aligning the patient with the CT images taken previously. This method of alignment is called registration, and in this study, two different methods for registration were investigated. One of the methods (method 1) uses skin markers placed on rigid structures. The other method (method 2) relies on touching the wall of the trachea with the bronchoscope. By evaluating these two techniques, we hope to optimize the technique of image-guided bronchoscopy for practical clinical use.


   Materials and Methods
TOP
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Patients
The protocol was approved by our institutional review board, and before bronchoscopy, all patients signed written informed consent

Fifteen patients (7 men and 8 women) scheduled for bronchoscopy were included in the study. Of these 15 patients, there were 12 patients scheduled because of abnormal imaging findings (ie, solitary mass, nodules, and/or adenopathy), 2 patients for post–lung transplant follow-up, and 1 patient for hemoptysis. Seven patients had TBNA, four patients had lavage, two patients had inspection only, and two patients had forceps lung biopsy. The mean body mass index, taken from a standard table using patient height and weight, was 23 kg/m2 with a range of 18 to 31 kg/m2. Patient height and weight data were available only for nine patients.

Bronchoscopy
Bronchoscopy was performed by a pulmonary attending physician with or without the aid of a fellow. The patient signed separate informed consent forms for the study and the diagnostic bronchoscopy before the procedure, which was performed in a dedicated endoscopy unit. Patients were monitored with continuous ECG and finger pulse oximetry. BP was measured every 5 min. The nasal passage, vocal cords, and airways were anesthetized with lidocaine (combination of 2% viscous, 10% spray, and/or 1% or 2% solution). The standard practice for sedation was IV midazolam (typical dose, 3 to 5 mg) and IV fentanyl (typical dose, 50 to 100 µg). Atropine (0.5 to 1 mg) was administered if the patient’s heart rate was < 100 beats/min. The bronchoscopist passed the bronchoscope via a naris and performed an airway inspection. After assessment of the registration methods for this study, the bronchoscopist performed the clinically necessary procedure (ie, lavage, TBNA, forceps biopsy, or inspection).

CT Scans
On the morning of the bronchoscopy, the patients had approximately 10 metallic 2-mm "nipple markers" placed on their chests. These markers were placed in each axilla and from the level of the clavicle to the xiphoid (Fig 1 ). As best as possible, markers were placed in areas unlikely to change with patient position. Therefore, bony surfaces (eg, clavicles and sternum) were preferred to flexible soft tissue. In some large patients, the axilla to axilla distance was wider than the mapping field. For these patients, markers were placed on one side of the thorax.



View larger version (154K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Metallic skin markers were placed on the patients’ chests before CT scanning. These markers remained until after bronchoscopy. They were used to register the images using method 1. Markers were placed on relatively rigid structures such as bony structures.

 
Spiral CT scans (Somatom Plus Four; Siemens; Iselin, NJ) were performed using 3-mm collimation, 6 mm/s table speed, and 3-mm reconstruction. Scans were performed with IV contrast injected at approximately 2 mL/s. The patients kept their arms down to best simulate their position during the later bronchoscopy. Also, scans were performed while the patient held a breath after exhalation. Images were sent electronically to the navigation computer in the bronchoscopy suite.

Navigation System
A miniature 1.5-mm-diameter position sensor (Biosense, a subsidiary of Johnson & Johnson Inc; Tirat HaCarmel, Israel) was placed at the tip of a flexible bronchoscope. The sensor and wires were placed within the sheath of a sheathed bronchoscope (Vision Sciences; Natick, MA; Fig 2 ). External to the patient were three electromagnetic emitting sources, each with temporal and signal specificity (Fig 3 ). These source coils each emit an ultra-low-strength electromagnetic field (5 x 10-6 to 5 x 10-5 T) that codes a high-accuracy mapping region measuring approximately 20 x 20 x 20 cm. The sensor can detect position in six degrees of freedom (both position and orientation).



View larger version (76K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. The 1.5-mm-diameter position sensor was placed at the tip of a flexible bronchoscope. The sensor was fixed in place under a disposable sheathed bronchoscope.

 


View larger version (158K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. The black triangle suspended over the patient’s chest contains three electromagnetic field sources at each of the triangle’s points. These sources each emit a specific signal that is used to determine the bronchoscope tip position.

 
A second miniature sensor was fixed to the skin above the sternum with adhesive. This second sensor maintained the frame of reference once the image correlation was complete.

Image Registration Method 1 (Skin Markers)
Registration method 1 used the external markers to register the image. Each of approximately eight markers were touched with the sensor while simultaneously indicating the corresponding marker seen on the computer CT display during expiration (Fig 4 ). This procedure took approximately 5 min before the bronchoscopy, and on completion, the system was actively tracking the position and orientation of the sensor.



View larger version (122K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4. Registration method 1 required touching each of the metallic markers with the sensor and then simultaneously indicating the structure touched to the computer. This procedure took approximately 5 min. Note that attached to the patient’s chest is a second sensor used to track respiratory motion.

 
Image Registration Method 2 (Internal Trachea Mapping)
After assessment of registration method 1, registration method 2 was performed. Registration method 2 did not use any external markers but, instead, used the patient’s own airway. This method was performed after the bronchoscopy began. First, the sensor at the scope tip was brought to the origin of the right upper lobe bronchus, then to the carina, and finally to the origin of the left upper lobe bronchus. At each of these anatomic points, the corresponding CT position was indicated to the computer. After these anatomic approximations, the sensor was dragged along the anterior and lateral walls of the trachea. The computer accepted approximately 30 of these wall points to form an approximation of the trachea position. Once the structure of these points was indicated to the computer, the system was actively tracking (Fig 5 ). This registration method also took approximately 5 min to perform.



View larger version (106K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5. The computer display of the navigation system provided coronal (top left), sagittal (top right), axial (bottom left), and three-dimensional (bottom right) views. Crosshairs indicate the real-time position of the sensor on the two-dimensional views. The bronchoscope tip is in the airway just above the carina. The three-dimensional view shows the trachea in blue and the anterior nodal mass for biopsy in yellow. The green bar indicates the orientation of the bronchoscope tip as it is flexed anteriorly, prepared for biopsy.

 
Accuracy
After registration, but before bronchoscopy, registration method 1 was assessed for accuracy in five patients. This was done by touching the approximately 10 surface markers with the sensor. At each marker, the position on the CT scan was noted. The distance between the actual position and the position of the marker on the CT scan was measured. This method was not used for registration method 2 as the position of the surface markers was far from the region used for registration (ie, the trachea to the skin).

Subjective assessments for the two methods of image registration were made by directing the bronchoscope to the carina and the left and right upper lobe bronchi branches and assessing CT position accuracy. CT position and orientation was also correlated during the inspection of the entire airway. Respiratory variation was also noted by measuring the position distance between normal inspiration and expiration during sedation for three successive respiratory cycles.


   Results
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
In the final five patients, the error distance between the actual marker position and the position noted on the CT scan was measured. This was performed in the final five patients to allow for a learning curve related to the use of the navigation system. Using registration method 1, this error was on average (± SD) 5.6 ± 2.7 mm (range, 1.3 to 10.4 mm). The error seen for markers placed on flexible soft tissue was higher than that for rigid points such as the clavicle.

Subjectively, both methods provided accurate registration. Occasionally, registration was inaccurate, with the CT scan indicating a position outside the airway. On these occasions, the registration was repeated to correct image correlation. When using fiberoptic bronchoscopy as the "gold standard," method 2 was judged more accurate than method 1.

When patients coughed gently, position was maintained by the secondary reference sensor. However, when one patient coughed vigorously, lifting her chest off the table, image registration had to be repeated.

Respiratory position variation between inspiration and expiration during the normal tidal volume under sedation measured 3.6 ± 2.6 mm. This reflects partial compensation by the second external position reference sensor. Last, precision measurements taken repeatedly at the same point showed a variation of 2.3 ± 1.6 mm.


   Discussion
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
Currently, bronchoscopic procedures are performed using fiberoptic or video images. However, optics offer limited assistance for obtaining biopsies of extraluminal lesions. Virtual bronchoscopic techniques aid in prebronchoscopic planning, but are not actively part of the biopsy procedure. In this study, a miniature electromagnetic position sensor was placed at the tip of a flexible, fiberoptic bronchoscope in an attempt to bring three-dimensional CT imaging to the bronchoscopic procedure suite.

Critical to the application of real-time electromagnetic position sensing is the ability to align the previously acquired images with the patient’s position on the bronchoscopy table. In this study, two different methods for alignment or registration were evaluated.

The advantage of method 1, the external marker method, was that the registration procedure could be performed before bronchoscopy or even sedation was begun. However, the disadvantage of this procedure was that it required repeating the CT scan on the day of or the day before bronchoscopy so that the skin markers could be placed before the scan. Also, it was difficult to find fixed skin locations for marker placement in obese patients with movable soft tissue.

The advantage of method 2, the internal method, was that no skin markers were necessary. This meant that the CT scan could have been performed several days to weeks beforehand. The scans could even have been performed at outside locations and transmitted electronically to the bronchoscopy suite. Obese patients were not as difficult to register as they were with the external method approach. Also, the registration was subjectively judged better than the external method. This was attributed to choosing registration points directly in the vicinity of the desired mapping area. The disadvantage of this internal method was that the registration process occurred while the bronchoscope was in the patient, thus extending the total bronchoscopy time.

While in fixed phantoms (ie, nonmoving objects that are useful for testing the position sensor), the electromagnetic position sensor showed accuracy and precision of approximately 1 mm; the accuracy seen here in a moving chest was on average 5 mm while the precision was on average 2 mm. Because most target lesions for TBNA or forceps biopsy are > 1.5 or 2.0 cm, respectively, the system should offer improved guidance. Error is most likely the result of respiratory motion. In this study, we had a single respiratory sensor placed on the patient’s chest. This primarily tracks motion in the vertical dimension and does not accurately track tracheal or bronchial motion. Placing the respiratory tracking sensor in the trachea itself would allow for more accurate tracking, although it may be clinically impractical.

Image-guided bronchoscopy may be helpful in the future. The work presented here helped to make the technique more clinically practical. This study showed that the internal method (method 2) of registration was possible and preferable. This means that a special CT scan with skin markers will not be necessary. It also means that the difficulty of finding rigid skin sites for skin markers, especially in obese patients, may be avoided.

In this study, clinical feasibility was shown. A method of image registration that can be easily and reproducibly applied was determined. Using the currently optimized technique of image registration, a randomized study is planned to test for improved diagnostic yield compared with conventional blind TBNA.


   Footnotes
 
Abbreviation: TBNA = transbronchial needle aspiration

Dr. Solomon has been a consultant to Biosense Inc.

Received for publication November 22, 1999. Accepted for publication June 8, 2000.


   References
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Haponik, EF, Cappellari, JO, Chin, R, et al (1995) Education and experience improve transbronchial needle aspiration performance. Am J Respir Crit Care Med 151,1998-2002[Abstract]
  2. Wang, KP, Brower, R, Haponik, EF, et al (1983) Flexible transbronchial needle aspiration for staging of bronchogenic carcinoma. Chest 84,571-576[Abstract/Free Full Text]
  3. Gay, P, Brutinel, W (1989) Transbronchial needle aspiration in the practice of bronchoscopy. Mayo Clin Proc 64,158-162[ISI][Medline]
  4. Schenk, DA, Chambers, SL, Derdak, S, et al (1993) Comparison of the Wang 19-gauge and 22-gauge needles in the mediastinal staging of lung cancer. Am Rev Respir Dis 147,1251-1258[ISI][Medline]
  5. McDermott, MW, Gutlin, PH (1996) Image-guided surgery for skull base neoplasms using the ISG viewing wand: anatomic and technical considerations. Neurosurg Clin North Am 7,285-295[Medline]
  6. Golfinos, JG, Fitzpatrick, BC, Smith, LR, et al (1995) Clinical use of a frameless stereotactic arm: results of 325 cases. J Neurosurg 83,197-205[ISI][Medline]
  7. Solomon, SB, White, P, Acker, DE, et al (1998) Real-time bronchoscope tip localization enables three-dimensional CT image guidance for transbronchial needle aspiration in swine. Chest 114,1405-1410[Abstract/Free Full Text]
  8. Ben-Haim, SA, Osadchy, D, Schuster, I, et al (1996) Nonfluoroscopic, in vivo navigation and mapping technology. Nat Med 2,1393-1395[CrossRef][ISI][Medline]



This article has been cited by other articles:


Home page
 ChestHome page
Y. Schwarz, J. Greif, H. D. Becker, A. Ernst, and A. Mehta
Real-Time Electromagnetic Navigation Bronchoscopy to Peripheral Lung Lesions Using Overlaid CT Images: The First Human Study.
Chest, April 1, 2006; 129(4): 988 - 994.
[Abstract] [Full Text] [PDF]

Home page
 CirculationHome page
J. Sra, D. Krum, A. Malloy, M. Vass, B. Belanger, E. Soubelet, R. Vaillant, and M. Akhtar
Registration of Three-Dimensional Left Atrial Computed Tomographic Images With Projection Images Obtained Using Fluoroscopy
Circulation, December 13, 2005; 112(24): 3763 - 3768.
[Abstract] [Full Text] [PDF]

Home page
 ChestHome page
H. Hautmann, A. Schneider, T. Pinkau, F. Peltz, and H. Feussner
Electromagnetic Catheter Navigation During Bronchoscopy: Validation of a Novel Method by Conventional Fluoroscopy
Chest, July 1, 2005; 128(1): 382 - 387.
[Abstract] [Full Text] [PDF]

Home page
 CirculationHome page
T. Dickfeld, H. Calkins, M. Zviman, R. Kato, G. Meininger, L. Lickfett, R. Berger, H. Halperin, and S. B. Solomon
Anatomic Stereotactic Catheter Ablation on Three-Dimensional Magnetic Resonance Images in Real Time
Circulation, November 11, 2003; 108(19): 2407 - 2413.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Article Archive
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (19)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Solomon, S. B.
Right arrow Articles by Wang, K. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Solomon, S. B.
Right arrow Articles by Wang, K. P.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS