| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Department of Respiratory Medicine (Drs. Asano and Matsuno), Gifu Prefectural Gifu Hospital, Gifu; the First Department of Medicine (Drs. Shinagawa and Yamazaki), Hokkaido University School of Medicine, Sapporo; the Departments of Internal Medicine (Dr. Suzuki), National Health Insurance Sekigahara Hospital, Gifu; the Department of Respiratory Medicine (Dr. Ishida), Fukushima Medical University School of Medicine, Fukushima; and the Department of Radiology (Dr. Moriya), Ohara General Hospital, Fukushima, Japan.
Correspondence to: Fumihiro Asano, MD, PhD, Department of Respiratory Medicine, Gifu Prefectural Gifu Hospital, 4-6-1 Noishiki, Gifu 500-8717, Japan; e-mail: asano-fm{at}ceres.ocn.ne.jp
Abstract
Study objectives: We performed ultrathin bronchoscopy for pulmonary peripheral lesions using a system that displays virtual bronchoscopy (VB) images to the lesion simultaneously with actual images and navigates the bronchoscope to the target bronchus. We then evaluated the system with regard to its usefulness and problems.
Design: A pilot study.
Setting: A tertiary teaching hospital.
Patients: The subjects were consecutive patients with small pulmonary peripheral lesions (
30 mm).
Interventions: Using this system, the rotation, advancement, and retreat of VB images were possible, and the bronchus into which the bronchoscope was to be advanced was displayed. VB images were displayed along with actual images, and the ultrathin bronchoscope was advanced to the target bronchus under direct vision. Under CT and radiographic fluoroscopy, a pair of forceps was inserted into the lesion via the bronchoscope. Thin-section CT images were obtained; after confirming the advancement of the bronchoscope into the target bronchus and the arrival of the forceps at the lesion, a biopsy was performed.
Results: Study subjects included 37 patients with 38 lesions. VB images to a median of the sixth- (third- to ninth-) order bronchi could be produced. Using this system, the ultrathin bronchoscope could be advanced into the planned route for 36 of the 38 lesions (94.7%). The system was used for a median of 2.6 min, and the median examination time was 24.9 min. The biopsy forceps could be advanced to the lesion in 33 of the 38 lesions (86.8%), and diagnosis was possible for 31 lesions (81.6%).
Conclusions: This navigation system is useful for ultrathin bronchoscopy for pulmonary peripheral lesions.
Key Words: lung cancer • navigation • small peripheral pulmonary lesion • transbronchial diagnosis • ultrathin bronchoscope • virtual bronchoscopy
Due to recent advances in CT apparatuses and their widespread use, particularly with the introduction of low-dose helical CT to the examination of lung cancer,1 the detection rate of pulmonary peripheral lesions has been increasing. Bronchoscopy is a routine diagnostic method for these lesions, with few related complications. However, the reported bronchoscopic diagnosis rate for pulmonary peripheral lesions ranges from 30 to 77%,2345678910 for the following reasons. Since a bronchoscope with an external diameter of 5 to 6 mm, which is a commonly used size at present, can be advanced only to approximately the third-to fourth-order bronchi, the forceps and curette should be guided for some distance toward the lesion. The forceps and brush themselves are not flexible, and their guidance to the target bronchus past many bronchial branching sites is difficult. This procedure is also difficult even when using a curette.
The ultrathin bronchoscope11 that has been used in clinical practice can be advanced to more peripheral bronchi than the conventional bronchoscope. Since the bronchoscope can be advanced close to peripheral lesions, subsequent guidance of the forceps is easy.12 In addition, the ultrathin bronchoscope can be advanced to lesions in places such as the mediastinal side of the lung apex that are difficult to reach with a conventional bronchoscope,13 making it useful for the diagnosis of small pulmonary peripheral lesions.1213 However, it is difficult to identify the route to a pulmonary peripheral lesion under direct vision within the limited bronchoscopic examination time. Therefore, before performing ultrathin bronchoscopy, confirmation of the route to be used for the advance of the bronchoscope by thin-section CT (TSCT) is very important. However, it is difficult to obtain a three-dimensional understanding of the complicated tracheobronchial tree based on axial images. Virtual bronchoscopy (VB) can be used to three-dimensionally display images resembling actual images based on helical CT images,14 but VB has rarely been used in bronchoscopy for pulmonary peripheral lesions.15 This is partly because the routinely used bronchoscope is thick, allowing observation of only limited areas, and more peripheral VB images cannot be obtained.
We have previously reported performing ultrathin bronchoscopy with VB navigation in which VB images to the target bronchus are produced from TSCT images and are used as a guide map at the time of advancement of the ultrathin bronchoscope.16 This method is useful for the diagnosis of small pulmonary peripheral lesions1718 and marking before thoracoscopic surgery.19 However, our previous method has some limitations in terms of the actual bronchoscopy.17 First, since the bronchoscope is advanced while being rotated, the rotation causes a shift of the observed actual image from the virtual image. The bronchial branching pattern includes many bifurcations into two bronchi with similar sizes. Mistakes therefore tend to be made with regard to the bronchus into which the bronchoscope will be advanced when the shift of the observed image from the VB image is marked. Second, when the bronchoscope moves unexpectedly due to patient movements such as coughing, immediate identification of the virtual bronchus branching image corresponding to the observed bronchial branching is difficult. To solve these problems, we developed an apparatus that displays VB images in comparison with actual images and evaluated its usefulness as a navigation system in ultrathin bronchoscopy for pulmonary peripheral lesions.
Materials and Methods
Subjects
The subjects were consecutive patients with pulmonary peripheral lesions ( 30 mm) encountered at the National Health Insurance Sekigahara Hospital between December 2002 and December 2004. Lesions with a CT-confirmed pure ground-glass opacity pattern ( 10 mm) were excluded because their diagnosis is considered to be difficult by biopsy that allows the collection of only a part of the lesion. The Institutional Review Board for Human Research approved this study protocol. All patients were given detailed descriptions of the examination and were informed that a new approach was being evaluated. Informed consent was obtained from all patients.
VB
First, the CT examination was performed using a helical CT scanner (HighSpeed Nx/I; General Electric Medical Systems; Tokyo, Japan) with the following parameters: 120 kilovolt, 1-mm collimation, dual detectors, pitch 3; and rotation time, 0.8 s. Helical CT data sets were acquired from an area centering on the lesion during single breath-hold inhalations so that the entry of the lobar bronchus involved by the lesion could be contained. The bronchus involved by the lesion was evaluated on axial TSCT images obtained by reconstruction at 0.5-mm intervals using the chest algorithm from helical CT data. Based on the helical CT data, VB was performed using software (Navigator, Advantage Windows 2.0; General Electric Medical Systems), and VB images to the lesion were produced. When the bronchus involved by the lesion was unclear, VB images to the bronchus that was closest to the lesion were produced.
VB Navigation System
Digital imaging and communications in medicine data on produced VB images were input into the navigation system (prototype; Olympus; Tokyo, Japan) developed in cooperation with Olympus. VB images of each bronchial branching on the route to the lesion were selected, and the bronchi for bronchoscope advancement were marked. This system has the following characteristics in terms of VB image display: (1) VB images between branching sites are displayed as animated images and can be peripherally advanced using a foot switch as the bronchoscope is actually advanced; (2) at each branching site, VB images can be rotated as the bronchoscope is actually rotated; and (3) the thumbnail of VB images at each bronchial branching is displayed as a catalogue.
Ultrathin Bronchoscopy With the Navigation System
Bronchoscopy was performed using an ultrathin bronchoscope (BF-type XP40 or XP260F; Olympus: external diameter, 2.8 mm; channel diameter, 1.2 mm) in a room equipped with CT and radiographic fluoroscopy apparatuses. Each patient was premedicated with 25 mg of hydroxyzine and 0.5 mg of atropine sulfates. Local anesthesia of the upper respiratory tract was performed using 2% lidocaine. After advancement of the ultrathin bronchoscope into the involved lobar bronchus, this system was used. First, the VB image was rotated to make it consistent with the observed actual image and was then advanced to the next branching. Next, the ultrathin bronchoscope was advanced to the next branching in the same manner as the VB image. Whenever the observed bronchoscopic image shifted from the VB image due to the rotation procedure during advancement of the bronchoscope, the VB image was rotated until it became consistent with the actual image. This procedure was repeated; using VB images as a navigator, the ultrathin bronchoscope was advanced to the target bronchus as far as possible under direct vision (Figs 12345
). After recording the time of use of the system, a pair of forceps (FD-44D-1, FB-56D-1; Olympus) was guided to the lesion under radiographic fluoroscopy and, if necessary, CT fluoroscopy (tube voltage, 120 kilovolt; tube current, 30 mA; collimation thickness, 5 mm). TSCT was performed under conditions the same as those for VB, except for the tube voltage (50 mA), to determine whether the ultrathin bronchoscope was sufficiently advanced into the target bronchus and whether the forceps had reached the lesion. When the ultrathin bronchoscope was guided to the target bronchus, navigation using this system was considered to be successful. Subsequently, biopsy was repeated until adequate specimens were collected. When necessary, we also performed cytodiagnosis of discharge attached to the biopsy forceps, substances brushed off, and bronchial lavage fluid.
|
|
|
|
|
Thirty-seven patients with 38 small pulmonary peripheral lesions ( 30 mm) were entered into the study. These included 23 male and 14 female patients with a median age of 72.5 years (30 to 85 years). The median lesion size was 18.5 mm (6 to 30 mm). The lesion size was 1 cm in 5 lesions, 1 to 2 cm in 21 lesions, and > 2 cm in 12 lesions. Fourteen lesions were located on the right superior lobe, 4 were on the middle lobe, 9 were on the inferior lobe, 7 were on the left superior lobe, and 4 were on the inferior lobe. Fourteen lesions could not be visualized by radiographic fluoroscopy. The involved bronchus was not clarified by TSCT in 10 lesions.
VB images could be produced to a median of the sixth- (third- to ninth-) order bronchi. Ultrathin bronchoscopy could be performed without complications in all subjects, and the median examination time was 24.9 min (range, 11.6 to 58.2 min). The median time of the use of the virtual navigation system was 2.6 min (range, 1.1 to 8.3 min). The bronchial branchings on VB images were in good accordance with the actual branchings. For 28 lesions, the ultrathin bronchoscope could be advanced under direct vision into the bronchus for which VB images could be produced. For eight lesions, the ultrathin bronchoscope could be advanced only one to two branchings before the bronchus for which VB images could be produced, although the bronchoscope remained on the planned route. Therefore, for 36 lesions, the ultrathin bronchoscope could be guided through the planned bronchial route using this system under direct vision. For the other two lesions, since the branching on VB images differed from the actual branching, the bronchoscope was advanced into a wrong bronchus. In these cases, the bronchoscope was guided to the planned route using CT fluoroscopy. The biopsy forceps could be advanced to the lesion in 33 of the 38 lesions (86.8%), and diagnosis could be made for 31 of the 33 lesions. Lung cancer was diagnosed in 17 lesions (adenocarcinoma, n = 14; small cell carcinoma, n = 1; squamous cell carcinoma, n = 1; and non-small cell carcinoma, n = 1); metastasis of ovarian cancer in 1 lesion; tuberculosis in 1 lesion; nontuberculous mycobacterial disease in 2 lesions; and inflammation in 10 lesions. The lesions determined to be inflammation decreased in size or disappeared during the subsequent observation period, which was consistent with the clinical characteristics of inflammation. Of the two other lesions in which a diagnosis was not made, one lesion showed no malignant findings and hamartoma was diagnosed by operation, and the other lesion was suggested to contain a few malignant cells by cytodiagnosis but a diagnosis could not be made by histologic examination, and lung cancer (adenocarcinoma) was subsequently diagnosed by surgery. The five lesions to which the forceps could not be advanced showed no bronchus involvement by the lesion on TSCT images. For these lesions, the ultrathin bronchoscope could be guided to the target bronchus that was the closest to the lesion, but the forceps could not be guided to the lesion. Of the five lesions, lung cancer was diagnosed (large cell carcinoma) in one lesion by surgery, lung cancer (adenocarcinoma) by conventional bronchoscopy performed as re-examination in one lesion, metastatic lung tumor (metastasis of colorectal cancer) based on the clinical course in one lesion, and inflammation based on the clinical course in one lesion; and the other lesion is still under observation without diagnosis. The diagnosis rate of this procedure was 81.6% for the 38 lesions examined by ultrathin bronchoscopy: 80.8% for 26 lesions 20 mm in size, and 83.3% for 12 lesions > 20 mm. In the 28 lesions for which the involved bronchus was clear, the forceps arrival rate was 96.4% and the diagnosis rate was 89.3%. In the 10 lesions for which the involved bronchus was unclear, the forceps arrival rate was 60.0% and the diagnosis rate was 50.0%. For the 37 lesions that could be finally diagnosed, the sensitivity, specificity, negative predictive value, positive predictive value, and accuracy for malignant disease were 81.8%, 100%, 78.9%, 100%, and 89.2%, respectively.
Discussion
VB is a novel technique for the noninvasive evaluation of the tracheobronchial tree.14 VB has been used for the evaluation of trachea and bronchi in children,22 bronchial stenosis,23 study of anatomic malformation and variants,24 guidance of lymph nodal transbronchial biopsy,25 and preoperative and postoperative assessment,26 but its use has been limited to the central airway. We have previously reported the usefulness of VB as a navigator in the bronchoscopic examination of pulmonary peripheral lesions but have also found new problems to be solved.17 Unlike VB, the tip of the actual bronchoscope can only be moved up or down. Therefore, for advancement of the bronchoscope to the target bronchus, the bronchoscope itself should be appropriately rotated at each bronchial branching when necessary. In our previous method, the observed images shifted from VB images produced in advance due to the rotation procedure. The use of VB images was difficult, and advancement into a wrong bronchus was observed in 6% of cases.17 In addition, when the bronchoscope tip was unexpectedly moved by patient’s coughing or respiratory movements, the location could become unclear, and the bronchoscope should be withdrawn and readvanced from the central side after confirming the branching.
In this study, VB images were rotated at each bronchial branching to achieve consistency with actual images. As a result, there was no mistake of advancement into a wrong bronchus due to the bronchoscope rotation procedure, and bronchoscope navigation was possible for 36 of the 38 lesions. This method may be particularly useful for lesions in areas such as S6, which require large rotation of the bronchoscope at the time of advancement. In addition, since a catalogue of the thumbnail of each virtual bronchial branching is displayed simultaneously with actual images, the order of the branching of the observed branching could easily be confirmed, even when the bronchoscope tip shifts or the visual field is transiently hidden by discharges. Due to these improvements, the navigation time and examination time were shortened in this study. For the 36 lesions to which navigation was possible, VB images were consistent with the tracheobronchial branching pattern of actual images to a median of the sixth- (third- to ninth-) order of branching, which confirmed the clinical applicability of VB images even in the peripheral region.
Attempts have recently been made to display VB images in comparison with actual images. In one study,27 actual images were superimposed on VB images or three-dimensional CT images produced in advance by detecting the accurate position of the bronchoscope from outside the body. A method using an electromagnetic sensor is promising but requires a special system and high superimposition sensitivity.28 Since the branching orifice displayed on VB images is wider than the actual orifice, and there are movements due to respiration and heart beats, the superimposition of these images may be difficult, particularly in peripheral bronchi. Another method involves the autoanalysis of the branching pattern on actual images and the display of corresponding VB images.29 This method, however, is not practical. In contrast, though our method is not automatic, the procedure itself is easy and practical. Since the system accepts digital imaging and communications in medicine data and audio/video interleaved data, VB images produced based on CT images at each institution can be directly used, and the input of the data into the system takes only several minutes. This system can therefore be used in general hospitals. Simulation by VB has been reported to be especially useful for the bronchoscopy training of beginners.30 To improve technical skills using this system before bronchoscopic examination, bronchoscopy simulation is also possible. In addition, since there are patients with branching anomalies or complicated branching patterns of even the segmental or subsegmental bronchi, this system may be also useful for routine bronchoscopy.
The problem is that the production of VB images depends on the performance and appropriate imaging conditions of CT, features of the software, and the experience of the technologist who produces VB images. In this study, for two lesions, the adjustment of the segmentation threshold was inadequate at the time of production of the VB images, and a bronchial orifice that was present on the route could not be visualized. For this reason, the VB image was not consistent with the actual image, and the bronchoscope was advanced to a wrong bronchus. To improve this, we are now developing software for the production of VB images by automatic adjustment of segmentation parameters and automatic search of the route to the target.
The diagnosis rate by conventional bronchoscopy for small pulmonary peripheral lesions ranges from 21 to 76.5% for those 20 mm in diameter.23789 In this study, the overall diagnosis rate was 81.6%, and that for lesions 20 mm in diameter was 80.8%. Considering that 14 of the 38 lesions could not be visualized by radiograph fluoroscopy and were not indicated for conventional bronchoscopy, the diagnosis rate by our method may be high. In addition, in the lesions for which the involved bronchus was clear, both the diagnosis rate (89.3%) and the forceps arrival rate (96.4%) were markedly high. Gaeta et al4 reported a significantly higher diagnosis rate in lesions for which the involved bronchus was clear (59%) than in those for which the involved bronchus was unclear (18%), but a significantly lower diagnosis rate in lesions for which the involved bronchus was the fifth order or more (33%) than in lesions for which the involved bronchus was the fourth order (90%). In our method, the ultrathin bronchoscope can be accurately advanced very close to the lesion using a navigation system. Therefore, the subsequent guidance of the forceps was easy, and a high arrival rate and a high diagnosis rate could be obtained. Therefore, even small pulmonary peripheral lesions are good indications for this method when the involved bronchus could be clarified by TSCT.
In this study, one of the reasons why a diagnosis could not be made was the lack of clarity of the involved bronchus on TSCT images. In general, the bronchoscopic diagnosis rate has been reported to be low in cases in which the involved bronchus is not clear.24 Even when the involved bronchus was not clear, the ultrathin bronchoscope could be accurately guided to the target bronchus near the lesion using this system. However, the subsequent guidance of the forceps was difficult even using CT, the forceps arrival rate and the diagnosis rate were low. In such cases, transbronchoscopic needle biopsy or percutaneous biopsy or operation may be necessary. The second reason for not being able to make a diagnosis was that adequate specimens could not be obtained in two lesions. A disadvantage of the ultrathin bronchoscope is that adequate specimens sometimes cannot be obtained and the obtained specimens are small. At our hospital, this problem is overcome by increasing the number of samplings and the combined use of cytodiagnosis. Further improvement of the forceps is also necessary.
For bronchoscopic examination of peripheral lesions, this system that allows display of VB images as a navigator in comparison with actual images is useful in bronchoscopy of peripheral lesions. For the widespread and routine use of this method for pulmonary peripheral lesions, further improvements in the system and a randomized study to clarify its usefulness are necessary.
Footnotes
Abbreviations: TSCT = thin-section CT; VB = virtual bronchoscopy
The authors report that they have no conflicts of interest related to this paper.
Received for publication September 27, 2005. Accepted for publication January 27, 2006.
References
This article has been cited by other articles:
|
|
 |
|
  M. Y. Dolina, D. C. Cornish, S. A. Merritt, L. Rai, R. Mahraj, W. E. Higgins, and R. Bascom Interbronchoscopist Variability in Endobronchial Path Selection: A Simulation Study Chest, April 1, 2008; 133(4): 897 - 905. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
(fifth-order bronchi), and x (sixth- or more order bronchi), and those that branch inferiorly, anteriorly, and medially are expressed as ii, ß, and y, respectively.