Optical coherence tomography-guided laser marking with tethered capsule endomicroscopy in unsedated patients

: Tethered capsule endomicroscopy (TCE) is an emerging screening technology that comprehensively obtains microstructural OCT images of the gastrointestinal (GI) tract in unsedated patients. To advance clinical adoption of this imaging technique, it will be important to validate TCE images with co-localized histology, the current diagnostic gold standard. One method for co-localizing OCT images with histology is image-targeted laser marking, which has previously been implemented using a driveshaft-based, balloon OCT catheter, deployed during endoscopy. In this paper, we present a TCE device that scans and targets the imaging beam using a low-cost stepper motor that is integrated inside the capsule. In combination with a 4-laser-diode, high power 1430/1450 nm marking laser system (800 mW on the sample and 1s pulse duration), this technology generated clearly visible marks, with a spatial targeting accuracy of better than 0.5 mm. A laser safety study was done on swine esophagus ex vivo , showing that these exposure parameters did not alter the submucosa, with a large, 4-5x safety margin. The technology was demonstrated in living human subjects and shown to be effective for co-localizing OCT TCE images to biopsies obtained during subsequent endoscopy.


Introduction
Tethered capsule endomicroscopy (TCE) is a new technology that implements optical coherence tomography (OCT) using a swallowable capsule that can be administered in unsedated patients [1,2]. Once swallowed, the device traverses the gastrointestinal (GI) tract, continuously collecting OCT images, creating a three-dimensional microscopic map of the esophagus in its entirety [1,2]. The procedure takes only 5-6 minutes and patients can return to their daily activities immediately after it is over [3,4]. Multiple studies have demonstrated the promise of using TCE for screening for esophageal diseases in patients [1][2][3][4][5][6][7], for other upper GI tract organs [8], and the use of esophageal TCE has been reported in a primary care clinic [3]. As opposed to endoscopy, TCE does not require sedation or a specialized setting and can be conducted by nurses or technicians. Following the procedure, the device can be disinfected an features of O Esophagus (B The next m diagnostic ac biopsies. Req whole organ m (VLE) [10,11 power laser l patient's esop these marked caused by pa [12,13]. Stud marking is hi spatial targeti now being use Laser mar device in vivo balloon. Our p showed that exposure dura time required system that p artifact.
In addition into the capsu important ben (NURD) artif Here, we rep micromotor-b nd reused ma CT-TCE poten BE), which here major step in th ccuracy agains quisite co-loca microscopy tec 1] and TCE. U light solves thi phagus that co sites are inher atient/device m dies using dri ighly effective ing error of ab ed clinically in rking in TCE o, as it is not previous laser laser marks w ation (Fig. 1). T d to generate a provided highe n, it has been ule has many a nefits of micro facts [15] that c port the deve based laser mar 1. Endoscopic ima a duration of 2 se rted due to tissue m any times (> 1 ntially make it etofore has not he clinical tran st the gold sta lization of OC chniques such Using of the s is issue by cre orrespond to s rently co-local motion that oc iveshaft-based, for co-localiz bout 1 mm. Ba n multiple cente devices has ad easily held in marking study were irregular The instability a laser mark be er power to sh recognized tha advantages ove omotor-based can become es lopment, valid rking TCE dev age of TCE laser m econds. Blue arro motion during the e 10 times), dri t useful for scr t been well add nslation of TCE andard, histopa CT images and as balloon-ba same catheter eating endosco specific OCT t ized to the targ ccurs between , centering ba ing esophagea alloon-based V ers to good eff dditional chall one place as y in human subj and ineffectiv of the capsule e reduced from horten the expo study is rotocol is y subjects are recruited from patients with biopsy-proven BE, aged 18 and older, who are undergoing an upper endoscopy. After the capsule was swallowed by an unsedated subject, the operator started OCT imaging. Images were recorded while the capsule passively traveled down the esophagus. Once the OCT image indicated that the capsule had passed through the gastroesophageal junction and reached the stomach, the capsule was manually pulled back across the lower esophageal sphincter and through the esophagus using the tether. When a target of interest was identified during pull-back imaging, the tether was held in place, and the tether tick mark reading at the incisors was recorded. The operator subsequently stepped on the foot pedal and enabled the marking laser system by pressing the enable button. Then, the operator clicked on the target displayed on the screen and pressed the activation button to place the laser mark(s). One or two marks may be placed, depending on study protocols and user input prior to the procedure. After marking, the motor automatically resumed spinning to acquire post-marking images. Following a short period of post-marking imaging, the tether tick marks at the incisor were recorded again and imaging and pullback resumed. This process was repeated for the next region of interest (up to two marks per centimeter and a maximum of 6 marks per subject). Most laser markings were performed at the diseased area, which is typically at distal portion of the esophagus, near the gastroesophageal junction. After all targets in the subject were laser marked, the capsule was pulled out from the subject and a questionnaire regarding tolerability of the procedure was administered. Then, the subject was sedated and underwent standard of care endoscopy with biopsy. During the endoscopic procedure, the laser marks were identified and biopsies were acquired in between or adjacent to the laser marks. Histology from the biopsy samples were then correlated with the corresponding OCT images. When multiple biopsies were taken, the histology was further correlated with OCT data by the corresponding tether tick mark recorded at the incisors.

The targeting accuracy of TCE laser marking
We quantified the cross-sectional targeting accuracy in human subjects in vivo by subtracting the location of the selected target position in the pre-marking image from the location of hyper-reflective mark in the post-marking image. Pre-and post-marking images were first coregistered using anatomical landmarks (e.g. tissue fold, crypt, and glands etc.) shared by both images.

1 Laser thermal injury tests on ex vivo swine esophagus
Using 800 mW of power at 1430/1450 nm, the exposure time between 1 and 4 seconds created cauterized marks that were visually identifiable ( Fig. 7(A)). Longer exposure times created larger cauterized marks. Figure 7(B) shows a representative NBTC histology image at a marking site (800 mW, 1s). wed that there w the margin to exposure durat on an excised sw 0/1450 nm, 800 m A representative he non-stained re the injury width an he mark to the subm extent of the la ce between the ayer (Fig. 7(B Fig. 7(B), rks to the ws that the and 4s. In ry did not C)). These produces standard ze motion in injury or distal  Figure 9 show motor TCE d generated hig

TCE lase
A total of 5 deviation) ye enrolled but capsule. Anot made it diffic laser marking including seda deviation) m successful (la attempts) of th applied. There Figure 11 esophageal sq marking OCT placed. These esophageal sq seen as superf orange arrow performed aft (Fig. 11(D), r confirmed tha shows the his roduced using ± 0.09 mm (sta or between the 0. Histogram of in it is scanned using er marking cli subjects (2 m ars old have u excluded from ther excluded s cult to identify g process was ation took 9.6 minutes for the aser mark pla he attempts. Th e were also no shows an exa quamous muco T image (Fig. 1  (standard ects were allow the on, which the entire procedure (standard king was out of 12 hen it was E patient) n the prended to be arance of er marks, ) and (C), ndoscopy esophagus ig. 11(E)) . . (E) Histology from a biopsy taken between the two laser marks confirms that the tissue between the marks is squamous epithelium. Figure 12 shows an example of TCE laser marking guided biopsy of Barrett's esophagus, in vivo. The pre-marking OCT image (Fig. 12(A)) shows BE as tissue without squamous layering, heterogeneous backscattering, and an irregular mucosal surface. For this case, only one laser mark was applied on the BE region at the location denoted by the red line ( Fig.  12(A)). After the laser mark was applied, the post-marking image showed a highly reflecting laser cautery mark near the intended location (Figs. 12(B), (C)). White light endoscopy performed after the TCE procedure demonstrates a clearly visible laser mark on the surface of salmon-colored esophageal mucosa (Fig. 12(D), red arrow). Histology from an endoscopic biopsy obtained adjacent to the mark (Fig. 12(E)) shows specialized intestinal metaplasia, the most common form of BE.

The accuracy of TCE laser marking in vivo
We were able to successfully co-register 9 pre-and post-marking images from the 5 subjects. After subtracting intent-to-mark and post-marking locations, the cross-sectional difference between the intended target and the actual marking positions was 0.95 +/− 0.53 (standard deviation) mm.

Discussion
The field of TCE is growing with many studies now using tethered capsules for GI tract diagnosis [1,2,7]. Since TCE does not have biopsy capability, it is essential that new technology become available to allow TCE images to be correlated to histopathology. Without histopathologic validation, this promising technology cannot be used for patient care. Laser marking using a tethered capsule is the most attractive way to conduct these validation studies and thus TCE laser marking technology presented in this study is a critical advancement. Moreover, we have demonstrated that using TCE laser marking to guide biopsy is feasible and safe in human subjects. After validation, the TCE device can be used independently for screening unsedated patients and for placing marks at lesions that may require follow-up during a subsequent endoscopy.
The main challenge that we encountered when conducting TCE-based laser marking was marking precision artifacts caused by motion of the capsule in the esophagus, which is greater than that of balloon-catheter VLE. By increasing the power to 800 mW and decreasing the laser exposure time to 1 second, we found that these artifacts can be overcome to produce tightly localized marks on the esophagus in living human subjects (Fig. 11(D) and 12(D)). Besides shortening the laser exposure time, we further mitigated this potential issue by performing laser marking only during pullback (Fig. 6). This procedural step allows us to fix the proximal end of the capsule to the leading edge of the peristaltic contraction. In addition, the size of targets, such as BE, dysplasia and adenocarcinoma are typically > few mm [17] and the accuracy of endoscopic biopsy is > 1 mm, which are roughly on the same scale as the ~1mm marking error at the dynamic gastroesophageal junction (GEJ) as shown in Fig. 12(A) and (B). Our overall cross-sectional targeting accuracy is 0.95 ± 0.53 mm (standard deviation from 5 subjects, 9 marks). Finally, since the hyper-reflective laser marks can be visualized in OCT images, we can correlate histology with post-marking OCT images for validating the diagnostic accuracy against histology. Given these considerations, the accuracy of TCE laser marking should be more than sufficient for most regions of interest in the esophagus.
In addition to optimizing the laser marking exposure parameters, we improved the device by incorporating a distal stepper micromotor in the capsule to scan the OCT beam. The A-line clock was used for synchronizing the A-line acquisition and motor rotation, so that a stationary target would not rotate and would remain stationary in the imaging window for high-precision marking. Once receiving the coordinate of target, the motor stopped at one of the 40 dwelling points that was closest to the target. The maximum discrepancy between the intended position and the actual dwelling position was half of the space interval between two neighboring dwelling points. Figure 10 shows the histogram of the size of space intervals. The maximum interval at the capsule's surface was 1 mm. Therefore, the maximum discrepancy between the intended position and the actual dwelling position (targeting error) was ~0.5 mm. Under same number of steps, the targeting accuracy scales inversely with diameter, as the diameter goes up, the accuracy goes down and vice versa. We have initially used 11-mm-diameter TCE devices as this is a standard set forth by predicate devices such as video capsule endoscopes. Unsedated TCE targeting accuracy is similar to sedated VLE targeting accuracy and thus both technologies may be used to guide targeted biopsy acquisition. The aim of this study was to demonstrate the feasibility of performing laser marking on unseated patients with TCE devices. In addition, since the stepper motor does not require an encoder for absolute position determination, the device can be less expensive and complex, which is consistent with the low-cost requirement of a capsule-based screening technology.
We used ex vivo animals to test our marking laser parameter set and froze those parameters for the clinical product. All subjects were tested under the same parameters. These parameters and the described clinical device satisfy the requirements for effective TCE-based laser marking in human subjects. For our exposure parameters (800 mW, 1430/1450 nm), the deposited laser energy, as determined by exposure duration (1-4 s), diffused predominantly in the lateral direction. From 1 to 4 seconds, the lateral injury width (Fig. 8(A)) increased by a factor of two, but the depth (Fig. 8(B)) only increased by ~20%. Therefore, it is not surprising that the laser thermal injury does not penetrate beyond muscularis mucosae even with a 4s exposure time (Fig. 8(C)). The extent of injury on the swine esophagus is comparable to that of standard biopsy forceps. For minimizing the motion artifact, we used a 1 s exposure time, which has a large 4x safety margin.
Results obtained in 5 subjects so far demonstrate that TCE laser marking can be a safe and effective way to correlate OCT TCE images to co-localized biopsies and corresponding histology. The missing mark (1 out of 12 attempts) was successfully identified in OCT image, but could not be identified by video endoscopy. Due to an uneven tissue surface, video endoscopy was not able to visualize 100% of the esophagus. In the future, we can increase the number of marks per target to avoid this potential issue. The co-localization capability is highly important now to validate the diagnosis of TCE images in histopathologic correlative studies. In addition, guided biopsy via TCE could also find similar utility to real-time targeting in VLE, enabling the sampling of aberrant tissue that is missed by random endoscopic biopsy, the current standard of care. TCE laser marking has advantages over balloon-based VLE targeting as the former can be conducted in unsedated patients at the point of care. Larger clinical studies using TCE laser marking should be conducted to validate TCE and determine its clinical utility for guiding biopsy.
Looking towards the future, it is also possible to consider a scheme where TCE diagnosis, targeting, and laser marking are performed automatically, based on advanced signal processing [18,19] or machine learning methods [20]. Such an advance could be facilitated by use of a pulsed Raman fiber laser [21], coupled into the inner cladding of a double clad fiber, recently demonstrated for ablating tissue in real time while OCT imaging [22]. This strategy would also be consistent with the screening case use scenario of tethered capsule technologies, obviating expert image interpretation during the TCE procedure.

Conclusion
OCT-based tethered capsule endomicroscopy is an up and coming, minimally-invasive technique for upper GI tract microscopic screening in unsedated patients. Histologic validation is a critical step that is required for the adoption of TCE. Here, we have described a TCE laser marking system and a stepper micromotor, beam scanning capsule that allows TCE OCT images to be correlated to endoscopic biopsies taken from the patient. Animal and human data using this technology in vivo shows that TCE laser marking is an effective and safe method for making this correspondence. Beyond histopathologic validation, TCE laser marking also has the potential to become a useful tool for marking microscopically abnormal tissue found during an outpatient capsule screening procedure that can subsequently be biopsied endoscopically. The convenience and cost-profile of such a strategy could significantly improve the care of patients with BE and other gastrointestinal diseases.