2.1.

OCT-AFI System

We used a swept-source OCT-AFI system developed previously by our group.²⁵ Briefly, a 50-kHz swept-source laser (SSOCT-1310, Axsun Technologies Inc., Billerica, Massachusetts) with 20-mW output power feeds a single-mode fiber 90/10 sample/reference split Mach–Zehnder OCT interferometer. Light for fluorescence excitation from a 445-nm semiconductor laser is coupled into the core of the polyimide coated double-clad fiber (DCF, FUD-3489, Nufern, East Granby, Connecticut) with the OCT light using a wavelength division multiplexer. The sample arm consists of a fiber-optic rotary joint that connects the stationary optics to the rotating core of the imaging stylet. A custom-built rotary-pullback drive allows rotation speed up to 33 Hz and pullbacks up to a maximum length of 160 mm performed at $1\mathrm{mm}/\mathrm{s}$. The OCT interferogram is acquired using a fast digitizer (ATS9350, Alazar Technologies Inc., Pointe-Claire, Quebec) in “k-clock” acquisition mode with custom data acquisition software. Autofluorescence light is collected in the inner cladding of the DCF, detected by a photomultiplier tube (H10723-20, Hamamatsu, Japan), and digitized by the second channel of the fast digitizer.

The OCT modality has a measured resolution of 24 um in the longitudinal direction and 14 um in the circumferential direction, and an imaging depth of $\sim 1$ to 2 mm in tissue. The AFI modality has a measured spot size at the surface of the stylette of 18 um in the longitudinal direction and 12 um in the circumferential direction. The AFI beam does not penetrate more than 1 mm into tissue and its spot size increases quickly with distance from the imaging stylet.

2.2.

OCT-AFI Imaging Stylet

The needle-compatible imaging stylet is based on an existing OCT-AFI catheter design developed by our group. The design was modified to be compatible with the commercial biopsy needle (FleXNeedle, 18G, Broncus Medical Inc, San Jose, California) selected for this work. The stylet included with the biopsy needle had an outer diameter (OD) of 0.69 mm. The OD of our existing catheter had to be reduced from 0.90 mm to match and pass easily through the needle lumen. Additionally, the insertable length of the catheter design was increased from 130 to 175 cm to span the entire length of the needle. The entire device was deployed through the instrument channel of a standard bronchoscope (Olympus BF Type P180A—EVIS EXERA II bronchovideoscope).

By changing the diameter and length of the catheter, mechanical performance can be impacted, as characterized by non-uniform rotational distortion (NURD) at the distal tip of the rotating optical assembly. Imaging stylets were tested in 3D printed channels with eight equally spaced markers placed along their length to highlight any deviations in a scan pattern from expected patterns.²⁶ This was done both with the imaging stylet unconstrained and contorted to match a typical path it would take through a bronchoscope and into human lung.

For the reduced-diameter window tube, Pebax^® 70D, Pebax^® 72D, and PET were considered and tested as potential tube materials due to their optical clarity and flexibility.

The distal tip of the imaging stylet is shown in Fig. 1.

Fig. 1

Distal end of the imaging stylet. (a) Schematic showing the OCT and excitation light path in red. (b) Imaging stylet loaded in the biopsy needle and retracted slightly for puncture and (c) stylet extended for imaging.

To reduce the OD of the stationary window tube to fit inside the biopsy needle, the diameter of the rotating optical assembly needed to be reduced to maintain clearance between the two. The torque cable [Fig. 1(a)] determines the size of the rotating assembly and runs the length of the probe, protecting the optical fiber while delivering the proximally-driven torque to the distal tip. A high torsional stiffness is desirable to avoid NURD²⁷ but often comes at the cost of decreased flexibility that limits the assembly’s ability to perform well in low-radius bends when conforming to lung anatomy. We ordered and tested custom double-wound torque cables (Asahi Intecc Co. Ltd., Aichi, Japan) with an OD of 0.33 mm and an inner diameter of 0.17 mm. Measurements made to estimate bending and torsional stiffness of the cable were both significantly lower than existing cables, but measurements of NURD performance showed torque cable-induced distortion comparable to our existing probes in straight orientations, and slightly superior in tortuous conditions emulating the path the bronchoscope would take in clinical conditions.

The stationary window tube [Fig. 1(a)] surrounds the rotating optical assembly and the water that acts as lubricant and optical coupling medium to reduce reflections due to index of refraction changes. The window tube outer surface makes contact with the entire inside length of the biopsy needle and tissue, when extended. Optical clarity, low friction, and resistance to buckling are all requirements of acceptable tubing. Of the materials tested, Pebax^® offered better resistance to buckling, whereas PET is stiffer and therefore, more easily passed through the narrow lumen of the biopsy needle (i.e. higher “pushability”). A probe was constructed using Pebax^® 70D, which was found to stick inside the biopsy needle occasionally despite having low friction over small distances. Another probe was made using PET to compare and was found to not stick even when the needle was contorted into tortuous orientations to simulate clinical conditions. The PET tubing was used for all future window tubes.

For flushing water, a T-junction luer lock connector was added to the proximal end of the biopsy needle handle (Fig. 2), such that a seal could be formed around the imaging stylet and water irrigated between the stylet and needle using an attached syringe. This also allowed for the application of vacuum pressure while the imaging stylet was still loaded (but partially retracted) in the needle to collect aspiration biopsy samples.

Fig. 2

The proximal handle of the Bronchus FleXNeedle, with attached T-junction for vacuum suction, and releasable luer seal around the imaging stylet.

2.3.

Image-Guided Needle Puncture Procedure

The protocol for image-guided biopsy was developed through ex vivo experimentation with porcine lung tissue. This led to the following puncture and pullback protocol (with reference to Figs. 3 and 4), which was followed during the subsequent in vivo imaging studies: (1) navigate the bronchoscope (with needle preloaded) to the desired airway; (2) verify bronchoscope position with fluoroscopy; (3) extend 10 mm of needle from the bronchoscope and angle the needle tip toward the desired site; (4) puncture by advancing the needle forward such that the needle penetrates the airway wall by 10 to 20 mm (Fig. 3); (5) irrigate the puncture site with saline [Fig. 4(a)]; (6) retract only the needle, leaving the imaging stylet in place [Fig. 4(b)]; (7) collect a pullback with imaging stylet [Fig. 4(c)]; and (8) re-advance the needle around imaging stylet [Fig. 4(d)].

Fig. 3

Videobronchscope images of the needle being (a) maneuvered into view; (b) advanced out of the tubing by 1 cm; and (c) used to puncture the airway wall.

Fig. 4

Puncture and pullback procedure. The needle and imaging stylet puncture the tissue together (a) before the needle is retracted by 10 mm while the stylet is held in place (b). The stylet is then used to capture a helical scan of the punctured tissue (c) and the needle is then re-advanced over the stylet into the tissue being scanned (d), at which point the stylet can be withdrawn from the needle for TBNA.

Where biopsy aspiration is indicated by image confirmation of a diagnostic site, the imaging stylet may then be fully retracted such that transbronchial needle aspiration (TBNA) can be performed at the same site.

Sites in the cranial right lobe were punctured to a depth of 10 mm. Sites were flushed with sterile water prior to imaging to displace blood, but two sites were imaged immediately after puncturing and prior to flushing to compare image quality. Figure 5 shows cross-sectional B-scans taken from pre- and post-flushed puncture sites with histograms of OCT intensity values for subsets of the image at roughly the same depth into tissue. The change in mean OCT intensity at site 1 ($<5$ gray levels, 1.6%) is significant ($P=0.007$), whereas the change at Site 2 ($<2$ gray levels, 0.6%) in not significant ($P=0.5082$).

Fig. 5

Pre and post-flush histograms of the OCT signal. Histograms were calculated from pixels in the red boxes, with the mean intensity shown at each site, and standard deviation in parentheses. There was $<5$ gray level difference between the mean OCT signal at site 1 and $<2$ gray level difference at site 2, indicating minimal improvement in penetration and contrast.

2.4.

Ex Vivo Imaging of Needle Puncture Site

The operation of the imaging stylet with the commercial biopsy needle was validated using six non-inflated excised porcine lungs. Organs were donated by researchers at the UBC Centre for Comparative Medicine after conducting unrelated studies covered by existing Animal Care protocols and ethics approvals.

Three lungs were also acquired and inflated using both positive internal and negative external pressure to inflate the alveoli as much as possible and mimic the OCT-AFI appearance of peripheral lung tissue in vivo.

In both cases, we navigated to the lung site of interest using the bronchoscopes white-light video and pierced with the needle up to a depth of 20 mm. Next, the needle was retracted fully into the needle sheath while the stylet and the needle sheath were kept stationary. The stylet, now embedded in tissue, was used to scan the length of the puncture. These pullbacks were acquired at 33 Hz, $1\mathrm{mm}/\mathrm{s}$ over 90-mm total length. AFI gain for each case was adjusted to achieve the highest contrast between the brightest and darkest features in a representative pullback.

2.5.

In Vivo Imaging of Needle Puncture Site

Image-guided biopsy was performed on three live swine using the commercial biopsy needle and the OCT-AFI imaging stylet during bronchoscopy based on standard instructions for use of transbronchial biopsy needles.^28,29 Procedures were performed in the Animal Care Facility at the Jack Bell Research Center and the study was approved by the UBC Animal Care Committee (A16-0029). The swine were sedated with acepromazine and ketamine followed by intravenous propofol. Animals were intubated and ventilated with blood pressure, ${\mathrm{O}}_2$ saturation, and heart rate monitored throughout the procedure.

Owing to anatomic differences (swine have longer trachea and airways that branch more slowly compared to humans³⁰), image-guided biopsy was performed in the cranial lobe as the other lobes were out of reach using a human bronchoscope. The cranial lobe branches from the trachea just prior to the main carina, and as such, involves a sharp turn similar to upper-lobe anatomy in the human lung. At each biopsy site, after puncture with the needle, sterile water was flushed around the stylet through the aspiration channel. Flushing was done to displace any blood present at the site, which may otherwise reduce penetration of OCT light and absorb the autofluorescence excitation and/or emission light. At two locations, pullbacks were acquired before and after flushing to compare image quality and inform whether regular flushing is necessary. The mean OCT intensities from pullbacks collected before and after a saline flush were compared using a two-tailed student’s $t$-test.

2.6.

Discrimination of Trans- and Endobronchial Pullbacks

Endobronchial pullbacks were collected by navigating the biopsy needle into the lumen of the target airway, advancing the imaging stylet from the needle, maneuvering the imaging stylet such that it was in contact with the airway wall along the desired pullback length, followed by acquisition of the pullback. Transbronchial pullbacks were collected by puncturing the airway wall (with stylet retracted), retracting the needle leaving the stylet in the puncture site, followed by acquisition of the pullback. OCT and AFI features were identified in the pullbacks that could discriminate between the endobronchial and transbronchial pullbacks. Endobronchial pullbacks were collected in RB1Cr and RB1Cd (cranial lobe) and compared with transbronchial pullbacks collected from puncture sites in the same airway.

2.7.

Nodule Confirmation

Following a similar protocol for the transthoracic deposition of artificial nodules in human lungs in vivo performed by Tsuchida et al.,³¹ powdered agarose and acridine orange were dissolved at 1.5% and 0.002% w/v, respectively, in water at 60°C. When cooled to body temperature, this mixture yields a solid mimicking the stiffness of lung parenchyma. The concentration of acridine orange was chosen to slightly exceed the highest level of fluorescence expected in normal lung tissue. Artificial fluorescent nodules could then be created by injecting the agarose/acridine orange/water mixtures into the parenchyma of one swine using an 18G needle designated for nodule placement.

Nodule deposition and detection were performed in vivo following process development on one of the ex vivo porcine lungs. We created two nodules in the cranial lobe and two nodules in a lower lobe.

At each nodule site, we first collected a transbronchial OCT pullback as a negative control. The solution and delivery needle were kept at 60°C until immediately prior to injection to keep the mixture fluid. The warmed needle, loaded with $\sim 0.5\mathrm{ml}$ of the agarose solution, was advanced quickly down the bronchoscope and used to puncture the airway at a depth of 5 mm. Positive pressure at the proximal port was applied and held for 30 s to allow the semi-viscous agarose to flow into the tissue. The needle was then unloaded from the bronchoscope and prepared for the next site while the nodule was left for at least 2 min to set before attempting to re-puncture and collect a second pullback for nodule detection. The bronchoscope was kept in position throughout to keep the puncture site in view of the white-light camera so that the needle could more accurately be maneuvered to the same airway wall section that was just punctured.

3.1.

Transbronchial Pullback

A transbronchial pullback from an airway in the cranial lobe is shown in Fig. 6. The autofluorescence information from the pullback is shown as a $\theta z$ surface plot [Fig. 6(a)] since AFI is not depth-resolved. Volumetric OCT information is plotted in longitudinal cross sections, $r\text{-}z$ [Fig. 6(b)] and circular cross-sections, $\theta r$ [Figs. 6(c)–6(f)].

Fig. 6

Example transbronchial pullback from an airway in the cranial right lobe showing a blood vessel, with the distal end of the scan on the left: (a) en-face AF projection; (b) OCT azimuthal pullback showing depth into tissue at the angle marked by the horizontal dashed line in (a); and four cross-sections (c)–(f) taken at the positions marked with vertical dashed lines. A vessel is visible in the AF projection, and in cross section in (b) and (e), marked with *. The needle outer tube is also visible at the proximal end of the scan, indicated with an arrow. The horizontal band with intense vertical streaks at the bottom of the AFI image is an artifact.

Figure 6(a) shows the AFI plot where the mucosa at the distal (leftmost) end of the puncture site is non-fluorescent and appears dark, and a 0.4-mm diameter blood vessel is seen forking at Fig. 6(e). The bright horizontal broadband at the bottom of the AFI plot with brighter vertical streaks increasing in number toward the right is an artifact due to specular reflections of the excitation light from the window tube being coupled back into the imaging catheter. This artifact was observed at most sites in the cranial lobe owing to the rotating optical assembly resting non-coaxially within the window tube in the non-rigid puncture cavities.

The OCT longitudinal cross section [Fig. 6(b)] shows the depth information at the horizontal cut (dashed white line) indicated in the AFI plot. The needle sheath is visible on the right side of the pullback (indicated by the white arrow). The vertical white lines indicate the position of the four cross-sections in the bottom panels. Figure 6(c) shows the probe entirely embedded in homogenous structural tissue, with no significant adjacent features, compared to Fig. 6(d), which has a grainier appearance suggesting unfilled alveoli from nearby airways. Figure 6(e) shows the medium-sized vessel near the puncture (indicated with an asterisk), and alveoli that are filled with air (dark semi-round cavities, especially on the left side).

OCT image quality, which is somewhat reduced due to multipath interference (ghosting) in DCF-based multimodal OCT systems, is comparable to the quality observed in our larger-diameter OCT-AFI catheters upon which the smaller-diameter imaging stylet is based. Features such as alveoli, mucus, airway epithelium, and adjacent airways are all clearly discernible. The needle sheath is visible at the end of the pullback, providing a physical reference point for distance measurements.

In the AF images, alveolar patterns are visible in the smaller distal airways, and superficial blood vessels of various sizes can be traced in some cases. We see a relatively uniform distribution of tissue autofluorescence, owing to the structural proteins collagen and elastin that normally provide mechanical support for the airways.

3.2.

Discrimination of Trans- and Endobronchial Pullbacks

Figure 7 compares cross-sectional imaging from transbronchial and endobronchial pullbacks at distal and proximal locations taken from non-inflated excised porcine lungs. At the distal position, the transbronchial cross-section shows homogenous backscatter that gradually decreases in intensity, whereas the endobronchial cross-section shows stratified intensity with bright structural tissue around the stylet more sharply transitioning to less scattering tissue, and alveoli present and separated from the stylet. At the proximal position, the transbronchial cross-section still has tissue surrounding the stylet, with minimal structure and bright features (possible un-filled alveoli) distributed both in contact with and separated from the stylet; in the endobronchial cross-section, the stratified tissue structure has expanded smoothly to reveal a lumen larger than the stylet, and a structural cartilage band (CR) is visible and concentric with the airway.

Fig. 7

Cross-section comparison of OCT features observed in transbronchial and endobronchial imaging: (a) transbronchial distal scanning shows alveoli in most directions, and only a thin membrane separating the probe from alveoli; (b) transbronchial proximal scanning still shows some alveoli, but also the more solid tissue of airway wall in the upper right; (c) endobronchial distal scanning shows some alveoli, but separated from the probe by airway wall or supportive tissue; and (d) endobronchial proximal scan showing an airway much larger than the probe, with CRs concentric with the airway. CR: cartilage band.

3.3.

Nodule Confirmation

We deposited four artificial nodules in one of the swine and then attempted to identify the nodules based on their fluorescent contrast two minutes later after the nodules had solidified.

The time between readying the solution and injection varied only by about 10 to 30 s for each nodule, but it was apparent that some nodules were fluid enough to leak out of the puncture site, whereas others solidified too much in the needle to be ejected. Another complication was the heat from the agarose needle caused the bronchoscope lens to fog up (despite using anti-fogging solution), which made confirmation of nodule deposition and relocation of the nodule site even more difficult.

One of the four artificial nodules provided a strong AFI signal, one provided a weak signal, and two nodules provided no signal above the background level of normal tissue autofluorescence. The puncture site that returned the clearest response is shown in Fig. 8, with the nodule located on the right, visible as a brightly fluorescent mass in the enface AFI image [Fig. 8(a)], and a semi-homogenous layer in OCT [Fig. 8(b)], with lower-fluorescing “normal” tissue in the distal direction. Differentiation of tissue and lesion in the figure is possible in OCT but much clearer in AFI. The needle becomes visible on the proximal side of the scan, obscuring tissue on the proximal side of the nodule in both the AFI and OCT images. As such, the imaging stylet was able to positively identify whether a nodule had been punctured with a single 20 second-long scan in one of the four attempts.

Fig. 8

Transbronchial pullback through an artificial lesion: (a) enface AFI and (b) OCT azimuthal pullback showing depth into tissue at the angle marked by the horizontal dashed line in (a). The distal (left) half of the pullback is in normal tissue, followed by the bright positive-fluorescence artificial lesion, and finally the needle (indicated with arrows in both views).