Lung cancer is a leading cause of cancer death in the Western world. More than 150,000 solitary pulmonary nodules are reported annually in the United States,1 each of which could indicate the presence of lung cancer. In addition, the increased use of computed tomography (CT) for screening could result in substantial increases in this number. Finally, the proportion of peripheral nodules compared with that of central lesions is also increasing, probably from changes in tobacco consumption. The presence of peripheral lesions frequently requires tissue diagnosis.2,3 The evidence-based guidelines of the American College of Chest Physicians advise against bronchoscopic biopsy of solitary pulmonary nodules4 as a result of the low likelihood of conventional technology to obtain diagnostic tissue specimens. Unfortunately, alternative methods such as transthoracic needle aspiration (TTNA), video-assisted thoracoscopy (VAT), or diagnostic thoracotomy have a greater risk of complications.
Primary surgical resection through thoracotomy or VAT is often recommended for malignant lesions. However, in a retrospective analysis of our 1467 operated cases, 720 (49%) lesions were benign.5 Moreover, in screening programs, the number of benign lesions has exceeded 90%.6 Benign nodules do not require surgical resection.3 Thus, the ability to obtain biopsies of isolated, peripheral lung lesions with bronchoscopy could prevent the complications of more invasive biopsy techniques and perhaps reduce the number of unnecessary surgeries by ruling out malignancy.
Unfortunately, conventional bronchoscopy under fluoroscopic guidance is unreliable in obtaining biopsies from lesions in the periphery of the lung, especially if the lesions are less than 2 cm in diameter and when they are hidden behind optical barriers such as the heart, infiltrates, or pleural effusion.7 In these situations, TTNA is considered, yet the pneumothorax rate could be as high as 30%.8 Bronchoscopy under CT fluoroscopic guidance has a success rate of 70% in obtaining diagnostic tissue9; however, the drawbacks such as radiation exposure to the patient and the medical personnel and the necessity of finding CT time for the procedure are considerable.
A new system using an endobronchial ultrasound (EBUS) probe has proved superior in obtaining biopsies, especially in lesions less than 3 cm in diameter10,11; however, the ultrasound technology is demanding, and the ultrasound probes cannot be easily steered beyond the visible parts of the airways.
Given the limitations, an advanced technology that would provide real-time imaging without fluoroscopic guidance12 would be able to track the path of the biopsy tool as it approaches the peripheral lesion and would allow the biopsy tool to be navigated within the periphery of the lung, which is highly desirable. A new electromagnetic guidance system using virtual bronchoscopy and 3-dimensional CT images, in combination with a steerable probe, has been successful in animal experiments and appears to meet these 3 criteria.13
The aim of this study was to determine the use and safety of this new electromagnetic navigation system in getting biopsy tools to the peripheral lung lesions, with the final goal of increasing the diagnostic yield of transbronchial lung biopsy.
MATERIALS AND METHODS
The ethics committee approved the protocol. Informed consent was obtained from all patients before the procedure.
Electromagnetic Navigation System
The electromagnetic navigation system or SDBS (superDimension/Bronchus system; superDimension Ltd., Herzliya, Israel) is an image-guided localization tool that assists in placing bronchoscopic accessories in the periphery of the lung. It has the following components:
1. Electromagnetic Location Board: The system uses low-frequency electromagnetic waves, which are emitted from a 1-cm thick, 47 × 56-cm magnetic board placed under the cephalad end of the bronchoscopy table mattress (Fig. 1).
2. A retractable sensor probe (locatable guide [LG]), 1 mm in diameter and 8 mm long, mounted on the tip of a flexible metal cable (Fig. 1B), is the main feature of the system. Inside the electromagnetic field, its position on the x, y, and z axes as well as motion (roll, pitch, and yaw) can be displayed on a monitor in real-time at a rate of 166 images/s and superimposed on previously acquired CT images. An added feature of the sensor probe is the 360° steerability of its tip, which is provided by 4 wires positioned inside the cable and attached to a rotating knob and a lever on its proximal handle. A socket at the handle provides the electronic connection to the computer (Fig. 1C).
3. The retractable sensor probe is inserted through a 130-cm long and 1.9-mm diameter flexible catheter that serves as an extended working channel (EWC).
4. The computer software creates a graphic depiction of the sensor probe's position in relation to preregistered anatomic landmarks that are superimposed on 3-dimensional CT images of the chest. An additional feature is the “tip-view,” which provides precise information to steer the probe to the lesion by turning the rotating knob on the handle.
1. Spiral CT imaging: To guide the probe to the target, the CT images and the patient's anatomy need to be aligned. A spiral CT scan of the chest (Siemens Somatom Volume Zoom, 4-liner, collimation 1 mm, increment 1 mm, table feed 6 mm/s, 120 kVp, 70 mA) was performed using the conventional breathhold technique in all patients.
2. Radiologic mapping: The digitized information from the CT scan was downloaded into the SDBS software in DICOM (Digital imaging and communications in medicine standards were created by the National Electrical Manufacturers Association to aid the distribution and viewing of medical images such as CT scan, magnetic resonance images, and ultrasound.) format. This information was then used to reconstruct graphic axial, coronal, and sagittal views of the chest and virtual images of the bronchial tree. Between 5 and 7 prominent anatomic landmark “fiducial points” (usually the main carina, the right and left upper lobe carina, the middle lobe carina, and in some cases, the lower lobe branching) were marked as coordinates on the corresponding CT images as well as on the virtual bronchoscopy image. Also, the center of the target lesion was identified and labeled in a similar fashion.
3. Endobronchial mapping: Bronchoscopy was performed with the patient under general anesthesia and high-frequency jet ventilation with a combination of rigid and flexible technique to avoid excessive motion produced by breathing, coughing, or body movement. Three reference sensors were fixed on the patient's thorax to compensate for respiratory movements and possible movement on the table. A flexible videoscope (Olympus EXERA 1T160 with a 2.8-mm diameter biopsy channel) was inserted through the rigid bronchoscope. The sensor probe was inserted through the working channel of the flexible bronchoscope. The “radiologic landmarks” selected on the virtual bronchoscopy image were identified in vivo and touched with the probe to register their location in the SDBS software. Registration of all this information into the computer software automatically synthesized a guidance scheme for approaching the lesion.
4. Real-time navigation: After the mapping was completed, the bronchoscope was advanced toward the bronchus leading to the lesion, with the sensor probe projecting from its distal end. The 3-dimensional CT images were displayed for the corresponding CT slice according to the actual position of the sensor. Once reaching the wedge position, the sensor probe and the EWC were steered to the target under guidance of the 3-dimensional CT images and especially of the “tip-view” orientation (Fig. 2A). Fluoroscopy in the anteroposterior and 30° oblique views confirmed that the sensor probe had reached the target. Then the EWC was fixed at the entrance of the bronchoscope's biopsy channel by a lock, and a miniaturized endobronchial ultrasound probe (UM-BS20-26R; Olympus, Tokyo) was inserted through the EWC (after retracting the sensor probe) to confirm the position of the tip of the EWC inside the lesion. Subsequently, flexible forceps, a brush, or a curette was inserted through the EWC to obtain a tissue specimen.
The accuracy of the navigation system was assessed with an “average fiducial target registration error” (AFTRE), which was defined to express the registration quality. It was calculated as an average of fiducial point errors, ie, the error of a fiducial point after correlation of the rest of the points. The AFTRE could be optimized (reduced) by repositioning the misplaced landmark or by eliminating the landmarks with the greatest deviation.
From July to December 2003, patients who had been diagnosed with peripheral lesions that were beyond the field of vision of the flexible bronchoscopy were invited to participate in the study. Lesion size was not considered in the eligibility criteria because use and safety were the primary objectives of the study.
Thirty patients were enrolled in the study (23 men, 7 women; ages 37-78 years; mean 65 years). The lesions ranged from 12 mm to 106 mm in diameter, 10 to 30 mm in 13 patients, mean 24 mm, standard deviation (SD) +1-5.4 mm; more than 30 mm in 17, mean 51.9 mm, SD ± 19.6 mm. The mean distance between the lesion to the pleura was 1.9 mm (range, 0-40 mm; SD, 10.3 mm). Lesions were in the right upper lobe in 11 patients, in the middle lobe in 2, in the right lower lobe in 3, in the left upper lobe in 9, and in the left lower lobe in 5.
Of the 30 patients, 29 were evaluable. In 1 patient, the sensor probe and the other instruments could not be advanced to the periphery of the apical segment of the left upper lobe because the airways were distorted by previous surgery.
A definite diagnosis was established with help of the SDBS system in 20 of the 29 (69%) patients. Among these 20 lesions, 15 were malignant (13 nonsmall cell lung carcinomas, 2 small cell lung carcinomas) and 5 were benign (active tuberculosis 1, tuberculoma 1, inflammatory infiltrates 2, round atelectasis 1). Because the patients in the latter group were not eligible for surgical procedures, the benign nature of their disease was established by observation of definite resolution under conservative treatment. Nine (31%) of the biopsies were false-negative as proved by surgical biopsy. The false-negative biopsies were from the right upper lobe in 4 patients, from the left upper lobe in 4, and from the right middle lobe in 1.
Navigation with the system was performed as planned in 25 of the 29 patients. In 5 patients (4 with lesions smaller than 3 cm), navigation was difficult or impossible as a result of: 1) mismatch in registration, 2) insufficient reference points on the chest wall, 3) insufficient endobronchial landmarks, and 4) anatomic distortion from previous surgeries.
Accuracy of Registration
Differences between the marked structures on the planning CT images and the registration points recorded during bronchoscopy (AFTRE) ranged from 2.85 mm to 9.53 mm (mean, 6.12 mm; SD, 1.7 mm).
Distance to the Target
The distance from the sensor probe to the target center was 1 to 11 mm in lesions smaller than 3 cm (mean, 5.8 mm; SD, 3.7 mm) and 3 to 28 mm in lesions 3 cm and larger (mean, 10.4 mm; SD, 7.8 mm).
Confirmation with Endobronchial Ultrasound
In 25 patients (86%), the lesion could be visualized with the EBUS probe inserted through the EWC. However, because the transducer is located 5 mm proximal to the tip of the probe, the ultrasound waves are directed radially and not axially; thus, in 4 of the 9 false-negative lesions, even though the lesion could be located beside the bronchus by the EBUS, they could not be reached with the biopsy tools.14
Radiologic mapping took less than 5 minutes (mean, 3 minutes). Registration with the SDBS during bronchoscopy added a mean of 2 minutes (range, 1-3.34 minutes) to the regular procedure time and navigation to the target added a mean of 7.3 minutes (range, 1.3-14.1 minutes).
One patient experienced pneumothorax after the lung biopsy and required chest tube drainage. Three other patients experienced minor, self-limiting bleeding that did not require any intervention. One patient died 4 days after the procedure from extensive metastases of small cell lung cancer. The death was deemed to be unrelated to the procedure.
We experienced no serious complications related to the use of this novel device and thus confirm its safety from animal studies. Although the aim of the study did not include improving the diagnostic yield of bronchoscopy, the increment over fluoroscopy-guided biopsy (69% vs. 30-50%) was striking. Thus, SDBS is feasible not only from a technical point of view, but from a clinical one as well.
The time needed, in addition to the conventional bronchoscopy, was considerable in the first few patients, because there was a learning curve in handling the equipment and navigating using newer information. However, with experience, the procedure time was reduced to 15 minutes.
However, several technical issues need to be addressed if SDBS is to be applied as a first-line biopsy tool for small peripheral lesions.
Planning and Registration
The mismatch in registration in some procedures is the result of several factors. The first is the difference between virtual bronchoscopy and the real bronchoscopy. In virtual bronchoscopy, spurs frequently appear wider than in reality; thus, the exact centered positioning of the registration marker is somewhat arbitrary. In some patients with extensive airway distortion from various reasons, including prior surgery, certain landmarks cannot be found or their locations are grossly misjudged.
Second, the planning CT image is acquired during a single breathhold. During registration, breathing and pulsation movements cause some displacement, especially in patients with chronic bronchitis, the bronchial spurs are unstable, distorted, or displaced. This displacement can be overcome to some extent by repeated breathholds under local anesthesia or by using continuous jet ventilation under general anesthesia. Under these conditions, only vascular and cardiac pulsations cause slight displacement, and registration error can be reduced to 1 to 2 mm.
Third, a good frontal registration triangle must be created by placing the chest wall markers well apart, which in some patients is essential. For example, in patients with an enlarged heart, the carina is widened and the main bronchi run horizontally. Thus, endoscopic registration of the main and upper lobe carina forms a straight line instead of a triangle and fall outside the frontal registration and cause inaccuracy.
Navigation is the most essential part in the procedure. Choosing the center of the target for navigation to smaller lesions is appropriate, yet in lesions over 3 cm in diameter, navigating to the target center might result in substantial error. By their mere size, larger lesions frequently occlude the bronchus leading to the center, and the sensor probe cannot be advanced. This inability explains the wider range in distances from the target center in larger lesions. Thus, in larger lesions, it might be better to navigate to the margin closest to the bronchus that leads to the lesion rather than to the target center.
Approaching peripheral lesions during conventional bronchoscopy is more or less a trial and error procedure. With electromagnetic navigation, the course of the sensor probe is followed on the computer screen and, if it deviates from the target, it has to be retracted back to the closest registration point and approach in a different direction is attempted. To avoid repeated insertion into the wrong bronchus, an electronic track can be left behind on this screen after each retraction. If, however, smaller lesions have to be approached, planning should start from the lesion going toward the closest reference point in a retrograde fashion. An electronic track should be created (distal to proximal) along which the sensor probe can be advanced. Eventually, the steering of the probe could be drawn on the screen, computer-assisted, or perhaps computer-driven. Currently, planning the track from the periphery to the central airways is too difficult because distal structures are too complex to differentiate between the bronchial and vascular system. We feel that improved electronic separation of the different structures during the planning CT can augment navigation to the target.
Steering the Sensor Probe
The steerability of the sensor probe is 1 of the most important features of the SDBS. Even behind almost completely obstructed central airways, we could maneuver the probe toward the target. However, because the extended working channel ended close to the tip of the sensor probe, the distal end became stiff, reducing the range of deflection. In addition, the angulation of certain bronchi limit the range of freedom of the tip, making the sensor probe “flip” to a different direction and making reaching peripheral lesion impossible. The flexibility of the catheter needs to be improved to overcome these obstacles.
Improved Control of Position
Fluoroscopy is comparatively unreliable in indicating the position of the sensor probe. As explained previously, the EBUS ultrasound probe does not allow axial viewing, and it can only confirm the exact position if the extended working channel is placed well inside the lesion. Preliminary experience with ultrathin endoscopes 1.4 mm in diameter (Polydiagnost Co., Germany) as adjunct tools has been promising and could improve positioning and the diagnostic yield.
In both animal and human trials, the SDBS guidance system proved feasible and safe for obtaining bronchoscopic biopsies of peripheral lung lesions. In addition, although improvement in diagnostic yield was not the aim of this study, we obtained diagnostic biopsies in 69% of our patients (20 of 29), which is greater than the 50% at maximum established using fluoroscopic control. We expect that with suggested improvement, the SDBS navigation system will be a valuable tool in the bronchoscopic diagnosis of peripheral lung lesions.
Heinrich D. Becker, Armin Ernst, Felix Herth, and Yehuda Schwarz are members of the Medical Advisory Committee of Super Dimension. They have received no financial remuneration for their effort. Study in part supported by Super Dimension equipment and accessories.