Schematic of the imaging set-up for the techniques used in the study. Imaging of excised vocal fold specimens was performed through a coverslip (CS) using several benchtop optical imaging instruments. Near-infrared laser light was focused into and scanned onto the sample through different optical assemblies to generate either large cross-sectional fields of view (FOVs) or stacks of small en face images. The cross-sectional FOV for optical frequency domain imaging (OFDI) and the en face FOVs for full-field optical coherence microscopy (FF-OCM) and spectrally encoded confocal microscopy (SECM) are represented with approximate dimensions relative to the specimen. FF indicates false vocal fold; TF, true vocal fold; and V, ventricle.
Images of a cadaveric vocal fold specimen taken from the body of a 23-year-old man. Abbreviations used in more than 1 panel: Ep indicates epithelium; FF, false vocal fold; LP, lamina propria; M, muscularis mucosa; TF, true vocal fold; and V, ventricle. A, In the gross photograph, the dashed line represents the cutting plane for the histologic sections (B and C) as well as the location of optical frequency domain imaging (OFDI) (D). B, Hematoxylin-eosin–stained histopathologic specimen (original magnification ×40). Scale bar indicates 100 μm. C, Trichrome-stained histopathologic specimen (original magnification ×100). Scale bar indicates 100 μm. D, For the OFDI image, vocal fold location was assessed by placing a fiducial marker of masking tape (FM) on the coverslip (CS) over the FF. The Ep and the junction between the Ep and the lamina propria (Ju) are clearly seen. The dashed arrow points at a signal-poor area that appears to be consistent with the location of the hypocellular middle LP. The scale bar indicates 500 μm (0.5 mm).
Images of a cadaveric specimen taken from the body of a 10-month-old boy. Abbreviations used in more than 1 panel: CS indicates coverslip; Ep, epithelium; LP, lamina propria; and S, shadow of the masking tape used as a fiducial marker. In all of the panels containing a solid arrow, the solid arrow points to the Ep-LP junction. All scale bars represent 100 μm. A, Hematoxylin-eosin stained image (original magnification ×100). FF indicates false vocal fold; TF, true vocal fold; and V, ventricle. B, Trichrome stained image (original magnification ×100). The rectangle delineates the area magnified in panel C. C, Greater magnification of the indicated portion of panel B (original magnification ×140). D, Optical frequency domain imaging (OFDI). E, Angle-resolved OFDI. The dashed arrows in panels C and E indicate image features suggestive of superficial blood vessels.
Images of a cadaveric specimen taken from the body of a 3-year-old boy. Abbreviations used in more than 1 panel: Ep indicates epithelium; LP, lamina propria; and TF, true vocal fold. A, Hematoxylin-eosin stained image (original magnification ×40). Scale bar indicates 1.0 mm; FF, false vocal fold; and V, ventricle. B, Trichrome stained image (original magnification ×100). Scale bar indicates 0.5 mm; M, muscularis mucosa. C, Corresponding full-field optical coherence microscopy cross-section with 4 different depths indicated: (1) 25 μm; (2) 100 μm; (3), 125 μm; and (4) 175 μm. Scale bar indicates 100 μm. D, En face images taken along the 4 planes indicated in panel C show the transition between the Ep and the LP. The dashed arrows in panel D(2) point to thin fibers; the solid arrows in panel D(3) point to individual nuclei. All scale bars indicate 100 μm.
Spectrally encoded confocal microscopy mosaic (top) and depth sequences (bottom rows) of a surgical pediatric vocal fold specimen. Imaging depths for each sequence are indicated in micrometers at the top left corners of the smaller images. The color-coded squares on the large mosaic indicate the locations of acquisition of the respective color-coded smaller images. Ep indicates epithelium; G, glandular structures. All arrows point to individual nuclei, and both scale bars indicate 100 μm.
Portable spectrally encoded confocal microscopy system comprising the cart (A), laser assembly (B), and handheld probe (C). The probe is built in 2 sections: the handle and the 13-mm-thick optical tube, which are separated for sterilization of the tube preoperatively. The parts are then reassembled in the operating room after the handle and cables are covered with sterile bags. SOA indicates semiconductor optical amplifier.
Two nonconsecutive frames from the 10-frame-per-second video acquired in the operating room, in vivo, during clinical spectrally encoded confocal microscopy of the larynx. Ep indicates epithelium; FB, image fly-back; N, nucleus; R, reflection artifact; and SLP, superficial lamina propria. Arrows point to individual epithelial cells.
Boudoux C, Leuin SC, Oh WY, Suter MJ, Desjardins AE, Vakoc BJ, Bouma BE, Hartnick CJ, Tearney GJ. Optical Microscopy of the Pediatric Vocal Fold. Arch Otolaryngol Head Neck Surg. 2009;135(1):53–64. doi:10.1001/archoto.2008.518
To compare and contrast 4 optical imaging techniques for evaluating the developing microstructure of the pediatric vocal fold and to identify the optimal strategy for in vivo imaging.
Academic medical center.
A total of 6 laryngeal specimens: 5 pediatric (ages 10 months to 16 years) (4 from cadavers and 1 from a living patient immediately after laryngectomy) and 1 cadaveric young adult specimen (age, 23 years).
Sequential noninvasive optical imaging of pediatric vocal fold specimens using optical frequency domain imaging (OFDI), angle-resolved OFDI (AR-OFDI), spectrally encoded confocal microscopy (SECM), and full-field optical coherence microscopy (FF-OCM), followed by fixation, sectioning, and histologic analysis of the same specimen for comparison.
Main Outcome Measure
Correlation between the microstructure observed using the 4 noninvasive optical imaging techniques and with the results of histopathologic analysis for the same specimen.
A successful in vivo imaging technique for developmental assessment of the pediatric vocal fold would include visualization of distinct layers (epithelium, lamina propria, and muscularis mucosa) and allow for identification of the individual cells composing the layers. The OFDI and AR-OFDI techniques provide a global assessment of the microstructure of the pediatric vocal fold to a depth of 1200 μm but lack the ability to distinguish cellular and subcellular structures. The FF-OCM technique allows for visualization with improved cellular detail (1-μm resolution), but the image acquisition speed is too slow for clinical use. The SECM technique has a faster acquisition rate and shows good cellular and subcellular detail to a depth of 250 μm.
The OFDI and SECM techniques were identified as promising and complementary candidates for in vivo cellular and subcellular imaging of the epithelium, basement membrane, and lamina propria of the pediatric vocal fold. To further validate the clinical potential of these techniques, a handheld SECM probe has been developed and demonstrated for in vivo evaluation of the pediatric vocal fold.
The treatment of pediatric vocal fold disease requires a complete understanding of pediatric vocal fold structural development and differentiation. Most of our current understanding of pediatric vocal fold structure is extrapolated from studies of adult vocal folds. Hirano’s1 seminal work evaluated the histopathologic characteristics of adult cadaveric laryngeal specimens, which allowed for a new understanding of the layers of the mature vocal fold. The complete description of the vocal fold includes a superficial epithelium, a lamina propria, and a vocalis muscle. The lamina propria can be further subdivided into 3 layers: the superficial (SLP), middle (MLP), and deep (DLP) layers, each characterized by its relative elastin and collagen content.1 The SLP, or Reinke space, is relatively hypocellular and contains low levels of collagen and elastin. The MLP is predominantly composed of elastin fibers, and the DLP contains collagen fibers.2,3
On a functional basis, the fully developed vocal fold can be divided into a cover-body ensemble.1- 3 The cover is composed of the surface epithelium and SLP, and the body is defined as the vocalis muscle. The MLP and DLP make up the transition zone between the cover and the body. As phonation requires the mobile cover layer to freely slide over a rigid body layer, preserving the lamina propria intact and uninjured is crucial. Adult vocal fold microsurgery requires a full trilaminar differentiation to allow subepithelial dissection with sparing of the SLP, thus preserving phonation.4- 8 While this structure is completely defined in an adult vocal fold, our knowledge of the pediatric vocal fold development is limited. It is currently unknown exactly how and when transitions occur from a relatively acellular monolayer at birth9 into a more complex 2-layered structure during childhood,10 and finally the mature adult trilaminar organ. It is thus difficult to assess at what age adult microsurgical techniques can be applied to the pediatric population without permanently affecting phonation.
Preliminary assessment of this pediatric vocal fold development was performed through histologic evaluation of pediatric cadaveric larynges. The differentiation into a bilaminar structure was seen to begin by age 2 months. The anatomic transition to a trilaminar structure occurs between ages 11 months and 5 years and is completed by age 7 years; however, it is not until age 13 years that the molecular composition (elastin and collagen contents) begins to resemble that of the adult vocal fold.11 While cadaveric studies permitted a first delineation of the different maturation stages, the statistical significance of the findings has been limited by the relative scarcity of cadaveric specimens as well as by artifacts caused by intubation and postmortem fixation.
To further increase our understanding of the developmental processes involved in vocal fold maturation, an array of in vivo imaging techniques are being developed and investigated. Over the last decade, particular attention has been given to optical imaging techniques for their ability to noninvasively acquire images with very high resolution (1 to 10 μm) and contrast at depths up to several millimeters. Both optical coherence tomography (OCT)12 and reflectance confocal microscopy (RCM)13,14 have transitioned from laboratory instruments to clinical tools, particularly in the fields of ophthalmology,15- 18 gastroenterology,19- 21 cardiology,22 and dermatology.23,24
In laryngology, OCT has allowed identification of the superficial epithelium, basement membrane, lamina propria, and thyroarytenoid muscle in both porcine and human specimens.25,26 In addition, in vivo studies of OCT images taken using a small catheter showed the ability of OCT to distinguish the epithelial layer from the SLP27 using collagen's birefringence as a natural contrast mechanism. Small catheters were also used to image microstructures of the larynges of 15 pediatric patients in vivo.28 In these 2 studies, imaging depths of OCT reached 2 mm and 1.6 mm, respectively. In comparison, RCM was shown to provide micrometer-resolution images of excised head and neck tissue types, showing for each specimen type its characteristic histologic features up to a depth of 400 μm.29 Reflectance confocal microscopic endoscopes were also developed to successfully image the uterine cervix30 and the gastrointestinal tract in vivo.31
The ideal optical imaging technique to allow for a comprehensive analysis of the developing pediatric vocal fold, and in particular its lamina propria structure, would image with enough depth and resolution to enable a clear understanding of when the layered structure of the lamina propria develops and what cells and fibers compose each layer. According to the findings of Boseley and Hartnick,11 who probed the depths of the lamina propria over the stages of development, such an ideal optical imaging system would be able to image to a depth approximating 1.2 to 1.3 mm. In addition, the instrument should be small enough for easy and safe passage either down the side port of a laryngoscope or through a standard suction catheter for in vivo imaging. This limits the diameter of the rigid part of the imaging instrument to 6 mm for infants (smallest laryngoscope) and up to 16 mm for older teenagers (using a small adult laryngoscope). Alternatively, very thin fiberoptic-based catheters (1-2 mm in diameter) could be used transnasally.
Our group32 recently performed a comparative study of 4 optical microscopy techniques that were shown either to fit the dimensional criteria or to have the potential to be miniaturized: optical frequency domain imaging (OFDI), angle-resolved OFDI (AR-OFDI), full-field optical coherence microscopy (FF-OCM), and spectrally encoded confocal microscopy (SECM). Preliminary imaging of the porcine vocal fold allowed visualization of architectural, cellular, and subcellular features of the vocal apparatus at depths up to 2 mm (OFDI and AR-OFDI) and 400 μm (FF-OCM and SECM).32 While the preliminary results showed great promise, 2 questions remained. First, can these technologies be used to delineate the different layers of the lamina propria? While cellular changes were observed with the highest magnification imaging techniques, the absence of a clear trilaminar structure in the porcine model indicated the need to pursue our investigation using human laryngeal cadaveric specimens. Second, given that OFDI could be performed using a small catheter,33 could SECM be adapted for in vivo imaging of the vocal folds in children? Could it be modified to conform to the size constraints specific to this population, and an acquisition rate be achieved sufficient to avoid motion artifacts? Some SECM probe designs have been proposed,34,35 and promising in vivo imaging of the skin has been shown through a handheld probe,36 but to date, no SECM system has been adapted to fully satisfy the constraints of the operating room.
In the present study, we present a comparative analysis of these 4 clinical optical microscopy techniques as tools for in vivo assessment of pediatric vocal fold development. First, we evaluate the potential of these optical microscopy techniques for noninvasive, high-resolution imaging of the pediatric vocal fold by comparing images of cadaveric and surgical pediatric specimens with corresponding histologic sections. Results from this comparative study constitute the final validating step toward conducting in vivo studies on a larger scale. Then we present a proof-of-feasibility study regarding the transfer of SECM into the operating room for in vivo imaging of the human pediatric vocal fold. Using a novel SECM handheld probe coupled to an integrated clinical cart comprising a compact wavelength-swept laser and an acquisition platform displaying images in real time, we obtained, to our knowledge, the first RCM images of pediatric vocal fold in vivo.
A variety of optical microscopy technologies have been developed and refined for use in vivo: OCT37 and its more recent implementation, OFDI,38,39 FF-OCM,40,41 and endoscopic RCM implemented as SECM.42,43 The 4 techniques offer different imaging capabilities in terms of field of view, depth of penetration, image acquisition speed, and resolution (Table 1). Over the course of 6 months, using 4 optical microscopes built in-house, we imaged 5 pediatric cadaveric specimens from children aged 10 months to 16 years and 1 surgical specimen.1 All protocols were approved by our institutions' internal review boards.
Prior to imaging, laryngeal specimens were dissected vertically between the thyroid ala into hemilarynges. Each specimen included 1 ala of thyroid cartilage, half of the cricoid cartilage, 1 arytenoid cartilage, 1 false vocal fold, 1 ventricle, and 1 true vocal fold. Following dissection, specimens were kept in phosphate-buffered saline solution to preserve tissue hydration, which was observed to play a critical role in image contrast. Imaging was performed though a 170-μm-thick coverslip used as a fiducial marker of depth for en face imaging with SECM and FF-OCM. Using the coverslip, we applied a slight compression to the sample. Compression has the disadvantage of distorting the specimen's morphologic characteristics compared with a histologic view, but it has the advantage of keeping the sample within the instrument focal range where both resolution and contrast are optimal. A piece of masking tape was placed on the coverslip to identify the false vocal fold. The tape, which is opaque to the laser light, acted as a fiducial marker on the OFDI image for the location of the false vocal fold and allowed identification of the ventricle and true fold areas.
The orientation of the OFDI imaging site was marked with India ink. Imaging with SECM and FF-OCM as well as histologic evaluation were performed at this site. Immediately following imaging, samples were fixed in a solution of formalin. Histologic evaluation was performed on 6-μm-thick slices using either hematoxylin-eosin or Masson trichrome stain (a 3-stain process used to identify collagen in tumors). In the case of the laryngectomy specimen, the contralateral vocal fold was processed for comparative histologic analysis.
Figure 1 shows a schematic of the configurations used for imaging the vocal fold specimens. For each of the 4 techniques used in this study, a laser beam was focused on the sample via a lens or a microscope objective and was scanned on the tissue to provide either a cross-sectional image (OFDI and AR-OFDI) or an en face image (ie, an image parallel to the specimen's surface: SECM and FF-OCM). As depicted in Figure 1, the relative sizes of the field of view for each device changes, a change that reflects the variety of optics and scanning mechanisms used in the respective set-ups. Light back-scattered from the sample was collected through the same lens in an epicollection mode, as is required for noninvasive imaging in situ. Three of the clinical microscopy instruments investigated in this study (OFDI, AR-OFDI, and SECM) used near-infrared illumination, which provided greater tissue penetration through decreased tissue scattering.
Optical coherence tomography is an interferometric, noninvasive optical tomographic imaging technique offering millimeter penetration with approximately 5- to 15-μm axial and lateral resolution.12 In an OCT system, light from a broad-bandwidth light source is split into 2 paths: one is a sample path illuminating the specimen and the other is a reference path, where light is typically reflected from a mirror. Interference between light reflected from the sample and reference paths provides information on the sample's reflectance as a function of depth. A 2-dimensional cross-sectional image is obtained by scanning the laser beam across the specimen.
Optical frequency domain imaging uses a variation of the optical technology used in OCT: while OCT uses a broad-bandwidth light source with a moving reference arm, OFDI (sometimes called swept-source OCT) uses a wavelength-swept light source with a stationary reference arm.38,39,45 The advantage of OFDI over OCT is that while the type of image obtained is essentially identical, OFDI provides 100-fold improvement in signal to noise ratio.38,39 This amelioration can be used to increase the imaging speed by 2 orders of magnitude. In addition to increasing image acquisition rates, AR-OFDI reduces speckle noise through angle-compounding of the different depth profiles.46 Although the usefulness of OFDI catheters has been demonstrated,20,33 the present procedures were performed using an optical fiber–based benchtop system, the specifications of which are detailed in Table 1.
We chose to include OFDI and AR-OFDI instruments in this comparative study because they offer both resolution (15 × 15 × 7 μm) and penetration depth (1-2 mm) that should allow for visualization of the epithelium and most layers of the lamina propria (typically 1.2 mm deep11) at acquisition rates compatible with in vivo imaging (ie, free motion artifacts) and through an instrument sufficiently narrow (1-2 mm in diameter) and flexible to examine the youngest patients.
Coupling low-coherence interferometry with high numerical aperture objective lenses to attain resolution at the subcellular level in 3 dimensions, FF-OCM is another important technique for imaging biological tissue. In contrast to OFDI, which obtains cross-sectional views (ie, as in typical histologic sections), FF-OCM visualizes en face planes, planes parallel to the surface of the sample. Figure 1 schematically illustrates both cross-sectional and en face imaging along with the relative sizes of fields of view for all techniques. While OFDI is capable of observing large areas (3-5 mm wide and 1-2 mm deep), FF-OCM typically images smaller areas (˜700 × 700 μm) and has a lower depth (<1 mm) over which high-quality images can be obtained. An advantage of FF-OCM is that its resolution (˜1 μm) is much better than that of OFDI, and it is approximately isotropic (ie, the same in every direction). These characteristics allow for reconstruction of stacks of en face images into cross-sectional views, facilitating correlation with histologic slides.
The FF-OCM system in this study41,47 used 2 identical microscope objective lenses and spatially incoherent broadband light from a xenon arc lamp in a Linnik interference microscope configuration (Table 1).41,47 While it is unlikely that the imaging depth of FF-OCM will allow for the visualization of all layers of the lamina propria, its exquisite resolution should allow for examination of the SLP with very high magnification to obtain understanding of vocal fold development at a cellular and subcellular level. In the current stage of development, FF-OCM is not yet suitable for clinical use owing to its relatively slow acquisition rate. The purpose of incorporating this imaging technique in our comparative study was to assist in correlating microscopic structures seen on stained histopathologic slides with those seen using novel optical tools. In addition, as FF-OCM images can be reconstructed in any given orientation (ie, en face or cross-sectionally), the technology can be used to facilitate the histopathologic interpretation of cross-sectional images obtained with OFDI and en face images acquired using SECM.
An optical fiber–based RCM technique, SECM has the advantage that its mechanism may be miniaturized to fit into handheld instruments34 and endoscopes.35 As with other RCM instruments, SECM is capable of obtaining high-resolution, high-contrast images of thin sections of bulk tissue by using optical sectioning, a technique involving pinholes and high numerical aperture microscope objective lenses. Traditionally, confocal microscopes have been hard to miniaturize because they require very rapid scanning mechanisms to move a focused laser beam across the sample to form an image. To obviate the need for a fast scanning mechanism within the probe, SECM instead projects different wavelengths onto distinct locations on the sample.42 Analysis of the spectrum returned from the sample yields information about the sample's reflectance at different spatial locations, thereby removing the need of scanning the laser in that dimension. Recent technical improvements have made it possible to image at 10 to 30 frames per second with SECM, rates sufficiently high to image in vivo without motion artifacts. This study was performed on an optical fiber–based benchtop SECM system43 comprising the optical elements detailed in Table 1.
As with FF-OCM, SECM produces en face images at different depths, measured with respect to the coverslip placed against the sample's epithelial surface. With a slightly coarser resolution than FF-OCM, SECM was nonetheless included in this study for its ability to provide cellular information on the SLP both at acceptable frame rates and through the confines of a laryngoscope. While the resolution of SECM is better than that of OFDI, the size of the SECM probe (˜10-15 mm in diameter) is too large to image the vocal folds of the youngest patients.
Figure 2A shows a gross vocal fold cadaveric specimen from a 23-year-old man who died after lung transplantation for cystic fibrosis. The hemilarynx shown in sagittal view contains the false vocal fold, ventricle, and true vocal fold. The dashed line indicates the cutting plane for the histologic sections. The hematoxylin-eosin (Figure 2B) and the trichrome stain (Figure 2C) images show a clear demarcation between the epithelium and the SLP and between the DLP, which is richer in collagen, and the muscle layer. On the trichrome section, an area of transition between the SLP and the DLP is seen (dashed arrow), possibly corresponding to the MLP, which is richer in elastin. Superficial fibrosis artifacts are observed on both sections, possibly corresponding to scarring from prolonged intubation.
Figure 2D shows an OFDI image acquired along the same dashed line seen in Figure 2A but under slight compression using a coverslip (shown as 2 parallel lines above the sample). The fiducial marker placed on the coverslip above the false vocal fold is seen on the OFDI image and allowed identification of the ventricle and true fold areas. Below the coverslip, a 50-μm-thick epithelium is seen to become thinner to the right of the true vocal fold, which corresponds to what is observed on the histologic sections. As was observed by other groups using OCT,26,27 the junction between the epithelium and the lamina propria is well seen with OFDI. While this contrast could be due to the collagen present in the basement membrane, this structure is itself much thinner than the resolution limit of the instrument. Such a strong contrast may be better explained by the large difference between the epithelium, which exhibits a relatively low scattering, and the matrix-rich tissue beneath it, an effect also observed on porcine models.32 Below the epithelium is a dark layer about 150 μm thick consistent with the fibrosis artifact observed on histologic images and possibly resulting from prolonged intubation. The dashed arrow points at a signal-poor layer. The depth of this layer is consistent with the location of the transition between the SLP and the MLP. The transition between DLP and the muscular layer was not observed on this excised sample because contrast became negligible at a depth of 1200 μm (1.2 mm).
Figure 3 shows histologic sections and OFDI images of a pediatric cadaveric specimen (10-month-old boy). Two of the histologic sections (Figure 3A and B) show the false vocal fold, ventricle, and true vocal fold. The transition between the epithelium and the lamina propria is seen at a depth of about 75 μm. A differentiation of the lamina propria into a bilaminar structure comprising the SLP and the DLP is observed at a depth varying between 200 and 400 μm. The inset of Figure 3B is magnified in Figure 3C to allow better identification of the structure seen in OFDI (Figure 4D) and AR-OFDI (Figure 4E). Masking tape was placed on the coverslip over the false vocal fold as a fiducial marker. The shadow of the tape is seen at the left of both the OFDI and AR-OFDI images (Figure 3D and E). The tip of the true vocal fold is indicated by an arrow and appears as a thin epithelial layer curved and pressed against the ventricle as the structures were slightly compressed by the coverslip. The transition between the epithelium and lamina propria is clearly seen at a depth of about 75 μm, which is consistent with histologic findings. The angle compounding provided by AR-OFDI (Figure 3E) allows for the observation of smaller structures that were hidden by speckle noise in OFDI (Figure 3D). The dashed arrow (Figure 3E) points to a small signal-poor area that appears to correlate with blood vessels in the histologic images.
Figure 4 shows images of a cadaveric specimen taken from the body of a 3-year-old boy with sulfate oxidase deficiency causing aspiration and airway obstruction. In Figure 4A, the superficial epithelium and the bilaminar structure of the true vocal fold are evident, and the superficial and deep layers are apparent. Figure 4B shows the increasing density of collagen fibers when moving from superficial to deeper layers. A superficial and deep layer of the lamina propria are evident, with a hint of a middle layer as well. Figure 4C shows an FF-OCM cross-sectional reconstruction from a series of depth images, some of which are shown in Figure 4D. The cross-sectional image (Figure 4C) shows an epithelium of varying thickness (30-80 μm) atop a bright thin layer that separates the epithelium from a lower heterogeneous structure. The en face images of Figure 4D vary in content as depth increases. Images at 25 μm show a rather nondescript epithelium (Figure 4D, while images at 100 μm reveal very bright and thin fibers (Figure 4D, dashed arrows). These fibers disappear at a depth of 125 μm (Figure 4D) and are replaced at 175 μm by coarser fiberlike structures(Figure 4D). Individual nuclei may be observed at different depths (eg, 125 μm: Figure 4D, solid arrows).
Figure 5 shows images of the laryngectomy specimen acquired with the SECM instrument. Since this system has a much faster image acquisition rate, multiple frames may be acquired rapidly and then stitched together to provide a comprehensive mosaic view of the specimen (Figure 5A). Imaging was performed under compression at a depth of 100 μm with respect to the coverslip. Because the sample was curved, compression (and thus the imaging depth) was greater at the center of the sample than on the sides. As a result, the edges of the sample were imaged in a cross-sectional fashion rather than en face. The epithelium is observed at the left edge of the mosaic separated from the basement membrane, a very reflective structure composed of thin fibers (orange frames). Individual nuclei (arrows) are observed as are well as larger structures reminiscent of glandular structures invaginating from the lamina propria and possibly resulting from prolonged intubation. Other areas of the mosaic (yellow frames) show a reticular pattern, consistent with cell membrane contours. Depth sequences taken at different locations on the sample show nuclei (arrows) and transitions from thin superficial structures to fibrous structures to coarser, deeper ones. Contrast became negligible at a depth of 250 μm.
To assess the feasibility of using SECM in the operating room to visualize pediatric vocal folds at the subcellular level, we built a prototype handheld instrument and used it in a pilot study of a pediatric patient under general anesthesia for airway evaluation using panendoscopy. Per the approved institutional internal review board protocol, patient informed consent and assent were obtained before the procedure.
A 16-year-old adolescent underwent suspension laryngoscopy to allow visualization and stabilization of the larynx at the same time freeing both of the surgeon's hands to perform the bronchoscopy procedure and to operate the SECM probe. During this time, the anesthesiologist ventilated the patient using spontaneous ventilation, which had the advantage of oxygenating the patient without using an endotracheal tube allowing an unobstructed view of the larynx. The vocal folds were first anesthetized topically using a local anesthetic (ie, lidocaine hydrochloride spray) to avoid vocal fold spasms, and the SECM probe was then gently placed onto the medial surface of the true vocal folds under direct visualization. The entire process of image acquisition took approximately 3 minutes.
Figure 6 shows the portable SECM instrument that was developed for intraoperative visualization of pediatric vocal folds. The SECM cart (Figure 6A) contains the custom wavelength-swept laser packaged for clinical use into a 300 × 300 × 140-mm unit and stored in the cart atop the detection and acquisition electronics. A computer acquires the SECM signal and displays it in real time on a monitor during the imaging session. Figure 6C shows the handheld SECM probe. It consists of a handle containing the scanning device and the illumination optics. Laser light is delivered to the probe via a single-mode optical fiber (Figure 6C, yellow cable) and relayed to the miniaturized 0.75 numerical aperture (NA) microscope objective lens, magnification ×40 (resulting in a resolution of 1 μm), through a series of telecentric telescopes housed in a stainless steel tube protruding outside of the handle. These telescopes serve 2 purposes. They ensure that the microscope back pupil is sufficiently illuminated to provide the best resolution, and they extend the instrument so that the vocal folds can be placed within the limited working distance of the objective, located at the distal tip of the probe. Prior to imaging, the tube is separated from the handle and gas sterilized according to standard protocols for surgical optical instruments. The handle and the cables (optical and electrical) are covered by a sterile plastic bag and reassembled with the sterile tube in the operating room minutes before imaging. The tip of the probe is dipped in sterile immersion medium that provides index matching between the lens and the vocal fold to minimize back reflections induced by abrupt changes in index of refraction. Video images of the procedure were acquired at 10 images per second and displayed continuously during the 3-minute procedure. While the imaging rate was sufficient to avoid motion artifacts, the probe was not equipped with a depth selection mechanism to allow quantification of imaging depths as the probe moved closer to the surface of the vocal fold.
Figure 7 shows 2 frames from the video generated from the imaging session. The images are unprocessed and reflect the quality of imaging available to the surgeon during the procedure. Back-reflections from various components were subtracted in real time, with the exception of a sharp artifact in the center of the image coming from an imperfection in the probe optics. The image on the left shows individual cell plasma membranes (arrows) of the upper cellular layers of the vocal fold. Bright structures within the boundaries of these cell membranes are seen in many images, especially in the right-hand image, with their shape being reminiscent of large nuclei. Fibers, which are more easily seen in the dynamic video, can be seen in the lower left corner of Figure 7, left, which was presumably acquired at a different depth owing to the curved geometry of the vocal fold.
The present findings support the conclusions of other groups regarding the relevance of optical imaging techniques for the noninvasive imaging of pediatric human vocal folds at the architectural and cellular levels. A unique aspect of this work is a side-by-side comparison of optical microscopy technologies that were developed for clinical imaging in vivo. The ultimate goal of our research is the development of an imaging technique for the study of the developing pediatric vocal fold, and our results are evaluated with this application focus in mind.
Through descriptions of ex vivo and in vivo imaging of pediatric human vocal folds, the present report highlights the relative merits of OFDI, FF-OCM, and SECM (Table 2). The OFDI technique shows increased sensitivity over traditional OCT systems and offers several advantages for laryngology imaging. Through its ability to reduce speckle noise via angle compounding, AR-OFDI allows for detailed imaging of the layered structure of the lamina propria. The imaging depth for excised study samples was limited to 1.2 mm, possibly owing to dehydration, which is known to change the optical properties of excised specimens compared with in vivo tissue. Endogenous signal from the junction between the epithelium and the lamina propria was very strong for all samples imaged, providing a potentially strong fiducial marker for thickness measurement of the lamina propria and epithelium in vivo. In its current implementation, OFDI does not have the resolution to provide cellular and subcellular imaging of laryngeal tissue, but it shows great potential to identify the boundaries of the different layers of the lamina propria.
The FF-OCM technique provides volumetric imaging and allows comprehensive imaging with exquisite cellular and subcellular detail. While it offers lower penetration depth than OFDI, its very high axial resolution allows for volumetric reconstructions and segmentations at very high magnification and contrast. Furthermore, successful FF-OCM at longer wavelengths has been demonstrated, with a significant improvement in imaging penetration depth.48 In its current implementation, FF-OCM is too slow for use in clinical imaging, but it nonetheless fills an important gap when used ex vivo because it provides a valuable tool for capturing the 3-dimensional microstructure of cadaveric specimens and aids in the understanding of the origin of the scattering signal seen with OFDI and SECM.
We identified SECM as a promising candidate for in vivo cellular and subcellular imaging of the epithelium, basement membrane, and SLP. While image quality in terms of resolution, contrast, and penetration were lower than that of FF-OCM, the images obtained with SECM conveyed the variable architecture of the layers of the vocal fold cover. An advantage of this technology is that it can be used at imaging speeds compatible with use in living patients.
Herein, we also present what we believe are the first in vivo RCM, or SECM, images of pediatric vocal folds. While this preliminary work has given the first clues to the layered structure of the vocal fold, we are not able to fully delineate the cellular and subcellular architecture of each layer. In addition, while cells were imaged at different depths, the current probe does not allow us to control or measure the depth. Technological advances will be required before SECM's full potential for in vivo imaging is realized. First, the size of the probe must be dramatically reduced to fit in the smallest laryngoscopes. Efforts to that end will focus on developing miniaturized microscope objective lenses, as achieved by other groups.49 Second, optical or mechanical mechanisms must be designed to allow for in vivo evaluation of depth of the en face imaging plane. This imaging orientation, which is orthogonal to typical histologic sections, presents an additional challenge that will require “reorienting” the surgeon to view vocal fold layers in this manner. A solution to these challenges might be to create a combined probe with both OFDI and SECM technologies. This combined instrument would make possible (1) imaging over a larger portion of the lamina propria (OFDI) at magnifications comparable to ×4 (OFDI) and ×40 (SECM); (2) easier correlation of cross-sectional views with en face views; and (3) calculation of depth of imaging (SECM) without bulky mechanical systems.
In conclusion, a better understanding of the pediatric vocal fold structure, development, and pathology is important because childhood voice disorders are a common but little understood medical ailment that can limit a child's ability to interact with and communicate with others during critical periods of development. Previous studies on cadaveric pediatric laryngeal specimens10 identified some critical time points in the stages of vocal fold maturation, but these studies were significantly limited by the small sample size and the presence of sample artifact. The ideal study would use noninvasive, in vivo evaluation of the pediatric vocal fold and include a larger pediatric sample size.
While the early results using these newly developed imaging methods have been promising, work is needed to optimize the techniques for laryngeal imaging in vivo and to determine whether a single OFDI probe, a single SECM probe, or a combination of the two would work best. Our next step is to image a population of pediatric patients in vivo using such probes while designing the next generation of instruments that will capitalize on the respective advantages of each technology. This has the potential to increase our understanding of the normative atlas of pediatric vocal fold structure and to identify the structural changes seen with common abnormalities, including vocal fold nodules and cysts.
Correspondence: Christopher J. Hartnick, MD, Department of Otolaryngology, Massachusetts Eye and Ear Infirmary, Harvard Medical School, 243 Charles St, Boston, MA 02114 (Christopher_Hartnick@meei.harvard.edu).
Submitted for Publication: January 31, 2008; final revision received April 24, 2008; accepted May 28, 2008.
Author Contributions: Dr Hartnick had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Boudoux, Leuin, Oh, Bouma, Hartnick, and Tearney. Acquisition of data: Boudoux, Leuin, Oh, Suter, Desjardins, Vakoc, and Hartnick. Analysis and interpretation of data: Hartnick and Tearney. Drafting of the manuscript: Boudoux, Leuin, Hartnick, and Tearney. Critical revision of the manuscript for important intellectual content: Leuin, Oh, Suter, Desjardins, Vakoc, Bouma, Hartnick, and Tearney. Statistical analysis: Tearney. Obtained funding: Tearney. Administrative, technical, and material support: Boudoux, Leuin, Oh, Desjardins, Vakoc, Hartnick, and Tearney. Study supervision: Oh, Bouma, Hartnick, and Tearney.
Financial Disclosure: None reported.
Funding/Support: This work was supported by funds provided by the Wellman Center for Photomedicine (Dr Tearney) and the Wellman Center for Photomedicine graduate program (Dr Boudoux).
Previous Presentation: This study was presented at the 2008 American Society of Pediatric Otolaryngology Scientific Program; May 2, 2008; Orlando, Florida.