Optical Coherence Tomography of the Larynx: Normative Anatomy and Benign Processes

Optical coherence tomography (OCT) is a minimally invasive, light-based imaging modality which produces high-resolution, three-dimensional (3D) images of biological tissues. A major focus of biophotonics research over the last 10 years is OCT-based evaluation of subepithelial laryngeal microanatomy, vocal fold vibration, and differentiation between benign and malignant disease processes of the larynx. With utility in both operative and office-based settings, OCT has potential as an adjunct diagnostic imaging modality to assist clinicians in the evaluation of normative and benign laryngeal processes and vibration parameters of the vocal cords. This chapter describes OCT-based delineation of vocal fold microanatomy, evolution of OCT technology, and OCT research to study normative anatomy, benign pathology, and functionality of the larynx.


Introduction
In patients presenting to the otolaryngologist with throat or voice complaints, a comprehensive laryngeal examination is necessary.While no standardized diagnostic workup applies to all laryngology patients, each established technique has advantages and limitations.In the awake, offi ce-based head and neck examination, the most commonly used technique to evaluate the larynx is indirect mirror laryngoscopy.While effi cient and least invasive of all examination techniques, this method does not allow for image capture and is limited to a surface view of laryngeal tissues.Flexible fi beroptic or rigid endoscopy with or without videostroboscopy allows for indirect evaluation of the larynx, but is also restricted to a superfi cial examination of laryngeal lesions.Conventional imaging modalities such as computed tomography (CT), magnetic resonance imaging (MRI), and ultrasound have limited spatial resolution, precluding their ability to characterize subepithelial microanatomy and structural integrity of the vocal folds.Without a precise assessment of depth of penetration and integrity of the basement membrane, surgeons cannot differentiate between benign and malignant lesions of the larynx in patients presenting with identical symptoms.At present, the gold standard for the assessment of malignant processes is microlaryngoscopy with excisional biopsy.This invasive technique is associated with increased healthcare costs, risks of general anesthesia and, most importantly, risk of disruption of vocal fold structure resulting in permanent dysphonia.Hence, there exists a need for a less invasive diagnostic modality which can assess the microanatomy of laryngeal tissues, while preserving the structural integrity of the vocal folds.

Anatomy
The layered microstructure of the vocal fold was described by Hirano et al. based on images from a scanning electron microscope [ 1 ].This information formed the basis of the "body-cover" theory of voice production.At the luminal surface, the superfi cial "cover" consists of stratifi ed squamous epithelium and the superfi cial (SLP) layer of the lamina propria (LP), as depicted in Fig. 1 .Immediately below, the "transition zone" contains the intermediate and deep layers of the LP (combining to form the vocal ligament), while the deeper "body" contains the thyroarytenoid muscle.During phonation, as air fl ows through the larynx and tension develops within the vocal folds, the contrasting properties of the "cover" and "body" cause a functional separation of these layers and vibration at different rates [ 1 , 2 ].Penetrating laryngeal lesions which disrupt the mucosal "cover" or cause tissue loss can disrupt the "bodycover" model, affect VF vibrational parameters, and may lead to signifi cant changes in voice quality.
OCT is capable of characterizing the layered microstructure of the larynx with optical penetration (approximately 1-2 mm) just short of the vocalis muscle.In a cross-sectional image, gray-scale representations of individual tissue layers are based upon their respective backscattering properties.The vocal folds have a weakly scattering epithelium while the SLP, known as Reinke's space, has a loosely organized fi brous composition.Deeper layers of the LP Fig. 1 Cross-sectional schematic of true vocal fold, depicting the epithelium and superfi cial layer of the lamina propria (LP, "cover"), intermediate and deep layers of the LP ("transition zone"), and the thyroarytenoid muscle ("body") can be differentiated based on their varying density of collagen and elastic fi bers.Hence, OCT can assess depth of penetration of laryngeal lesions and basement membrane and SLP integrity, thus offering a noninvasive means to differentiate benign versus malignant lesions of the larynx.

Principles of OCT
OCT uses a spectrally broadband light source coupled with a low coherence interferometer (e.g., Michelson) to produce highresolution images of biological tissue.Light from an optical source is split into two arms: a "sample" arm (containing the tissue of interest) and a "reference" arm (stationary or moving mirror).Light back-refl ectance from the "sample" is dependent on unique optical backscattering coeffi cients of biological tissue layers.Both optical paths are recombined and detected to form an interference profi le as a function of tissue depth.As the sample beam is scanned across the tissue surface, adjacent depth profi les (A-lines) are compiled to construct two-dimensional or 3-D images with micrometer resolution (~10 μm), millimeter depth (1-2 mm), and video-rate imaging speed [ 3 , 4 ].
The two principal OCT schemes are time domain OCT (TD-OCT) and frequency or "Fourier" domain OCT (FD-OCT).In 1991, Fujimoto's group published the fi rst report of real-time low coherence tomographic imaging in a biological system [ 3 ].Using a super luminescent diode as a light source, their TD-OCT system used a "reference" beam back-refl ected from a moving mirror coupled with a Michelson interferometer to construct an interference pattern.In vitro, cross-sectional images of the retina and coronary artery were produced by mechanically adjusting the distance to the "reference," thereby limiting the image acquisition speed [ 3 ].At the turn of the century, Fercher's group developed the fi rst FD-OCT systems [ 5 , 6 ].FD-OCT uses a wavelengthswept light source and with a stationary reference arm to acquire information in the spectral domain, per Fourier transform of the combined spectra at the output of the interferometer.FD-OCT provides improved signal-to-noise ratio, increased image acquisition rates, and higher sensitivity compared to TD-OCT [ 7 -9 ].Long-range or "anatomic" OCT (LR-OCT), fi rst pioneered in 2003 by Sampson as a derivative of TD-OCT, and later as version of FD-OCT, extends the axial imaging range of traditional systems without sacrifi cing resolution [ 10 , 11 ].
Polarization-sensitive OCT ( PS-OCT ) enhances the diagnostic potential of conventional OCT by recording both the intensity and birefringence of backscattered light.Birefringent tissue (e.g., collagen, muscle, cartilage) changes the polarization state of refl ected

Variations in OCT Technology
Optical Coherence Tomography of the Larynx: Normative Anatomy... light to provide an additional degree of contrast and specifi city to fi brous tissue layers.In imaging the vocal folds, increased collagen content in the vocal ligament can be contrasted with the relatively acellular, overlying SLP [ 12 -14 ].Furthermore, PS-OCT allows for differentiation between areas of vocal fold scarring and normal epithelium.
In 1993, the fi rst in vivo OCT studies were completed independently by groups at the Massachusetts Institute of Technology (MIT, Boston, MA) and Medical University of Vienna (Vienna, Austria) to image the retina [ 15 , 16 ].Later, the introduction of endoscopic OCT systems led to applications in cardiology and gastroenterology [ 17 -21 ].Further evolution into ultrahigh-resolution OCT (2-3 μm axial resolution) and spectral domain OCT has expanded the diagnostic potential of OCT within ophthalmology [ 5 , 22 -24 ].Comprehensive review of OCT principles and research in biomedical interferometry are described in the literature [ 4 , 24 , 25 ].

Evolution of OCT in Laryngology
OCT research within laryngology traverses ex vivo and in vivo imaging in animal and human models.Given the range of applications and imaging techniques (pediatric vs. adult airways, offi cebased vs. intraoperative imaging, static vs. vibrating vocal folds), OCT systems and optical probe design have evolved to optimize functionality and data quality.
The fi rst systems developed for in vivo scanning of the human larynx utilized near-contact endoscopic probes [ 19 , 26 ].These TD-OCT systems (central wavelength λ = 830 nm, 30 nm bandwidth) integrated the OCT sampling arm into standard endoscopic devices for transverse scanning (30 cm/s) along the plane of laryngeal tissue to identify mucosal structural changes in precancerous and cancerous states.Future iterations of these systems led to the Niris Imaging System (Imalux Corp., Cleveland, OH) which was the only commercially available OCT system for imaging of the upper aerodigestive tract (UADT).Designed as a portable TD-OCT system, this unit includes a 2.7 mm diameter probe to acquire realtime 2D images (200 × 200 pixels) with a maximum frame rate of 0.7 Hz.The Niris spatial depth resolution is 10-20 μm with scanning depth of 1.5 mm; lateral resolution is 25 μm with lateral scanning range of 1.5-2.5 mm [ 27 ].In 2010, Rubinstein et al. used the Niris system to obtain intraoperative images of normal laryngeal tissue, transition zones, and pathology [ 27 ].A fl exible probe held within a modifi ed suction handpiece was inserted through a surgical laryngoscope for controlled, accurate positioning at the area of interest (Fig. 2 ).The same year, Brenner's group used the Niris system to identify layered tissue microstructure of recurrent respiratory papillomatosis in the trachea [ 28 ].
At present, most laryngeal imaging has been accomplished using research systems designed and constructed in-house, customized for specifi c applications of interest.Imaging systems can be adapted for short-or long-range imaging and stationary or rotational scanning.Sampling probe designs may vary in outer diameter, focal length, and direction of light propagation (0° vs 90°).

Ex Vivo Studies
With the advent of faster, minimally invasive OCT systems, research has trended towards in vivo OCT imaging of the human larynx.However, much of our understanding of OCT-based visualization of laryngeal microstructure is derived from studies on harvested tissue.Ex vivo OCT in animal and human models has served to provide anatomical standards for normal laryngeal microanatomy and lay the foundation for OCT-based differentiation of benign versus malignant processes.Furthermore, direct comparison of OCT images with corresponding histological section allows scientists to measure OCT sensitivity and understand limitations associated with the technology and image quality.
In 2004, Bibas et al. used PS-OCT to image 10 tissue samples from a single human postlaryngectomy specimen in both longitudinal (B-scan) and transverse/ en-face (C-scan) modes.Stacks of C-scans were used to construct 3D images to identify refl ectivity patterns within the layered microstructure and to correlate OCT with histological sections [ 29 ].Similarly, de Boer's group used conventional OCT and PS-OCT to image human cadaveric larynges to identify patterns of optical backscattering and tissue birefringence associated with the layered microstructure of normal

Human Studies
Fig. 2 Specially designed handheld device coupled with NIRIS OCT system to guide a fl exible probe into the larynx [ 27 ] laryngeal mucosa [ 30 ].These reports helped to establish standards for OCT-based visualization of laryngeal microanatomy and provided a framework for future OCT data analysis.
A team led by Luerssen, Lubatschowski, Ptok, and colleagues has been a forerunner in investigating OCT applications in laryngology since the early 2000s.In 2005, Luerssen et al. performed high-resolution TD-OCT (central wavelength λ = 1350 nm, 5 μm axial resolution) of porcine larynges with an optical fi ber tip in direct contact with sample tissue.Their data demonstrated a clear distinction between epithelium and underlying LP, as well as direct correlation of layered microstructure with histologic section [ 31 ].Additional trials by this group in porcine and primate models using high-resolution or fi ber-based systems (5-10 μm resolution) in contact mode have been described [ 32 -34 ].In these reports, the authors report OCT-based identifi cation of mucosal substructure, laryngeal mucosa micrometry, and comparable sensitivity to histological examination.
Injury to laryngeal mucosa secondary to gastroesophageal refl ux, prolonged endotracheal intubation, or laser therapy for VF lesions triggers a wound healing cascade which, if undiagnosed, terminates in granulation and fi brosis.However, occult subepithelial pathology cannot be identifi ed by operative endoscopy or CT/ MRI.Multiple ex vivo animal studies have demonstrated OCT capability of identifying laryngeal and subglottic histopathology following simulated morphologic injury to the VF or subglottis [ 35 -37 ].Karamzadeh et al. studied OCT of a variety of simulated subglottic injuries (collagen injection, dehydration, rehydration/ edema, and repeated intubation) in harvested rabbit larynges [ 35 ].Larynges were suspended vertically in an OCT imaging stage with a probe positioned over the cricoid cartilage to acquire images vertically, in a cephalocaudal direction.OCT-based micrometry of mucosal tissues demonstrated an increase in LP thickness following submucosal injury, as well as unique signal intensity patterns correlating with levels of subepithelial collagen or edema [ 35 ].
Operative treatment of glottic lesions is often limited by the surgeon's inability to precisely gauge the depth of disease.Given the critical interaction of adjacent layers of the laryngeal mucosa during phonation, the "body cover" model must be respected during excision of glottic lesions to prevent vocal fold scarring or permanent dysphonia.In 2007, Wisweh et al. performed simultaneous OCT and femtosecond laser (fs-laser) cutting on extracted porcine larynges [ 38 ].An OCT system developed by Optimec Ltd. (Nizhny Novgorod, Russia) with 15 μm spatial resolution was used to identify sites of therapy and provides a reference for positioning the sample; the OCT scanner was swiveled out prior to fs-laser ablation at defi ned volumes and depths.This was the fi rst report which investigated the potential of OCT as an intraoperative imaging tool to guide laryngeal microsurgery.

Animal Studies
Giriraj K. Sharma and Brian J.-F.Wong

Clinical Studies
In the operative setting, the larynx is exposed under direct laryngoscopy and visualized with an operating microscope and fl exible or rigid endoscope (0°-90°).Intraoperative OCT of the larynx has been studied using a fl exible endoscopic probe or a probe integrated with an operative microscope.In 1997, Sergeev et al. reported the fi rst in vivo imaging of the larynx.Under direct laryngoscopy, a fl exible sampling arm was advanced through the working channel of a standard endoscope and positioned 5-7 mm away from the tissue of interest.The distal fi ber tip was swung by a galvanometric plate to image healthy laryngeal tissue and differentiate epithelium and LP based on optical scattering properties [ 19 ].
In 2005, Wong et al. reported laryngeal OCT during surgical endoscopy in 82 patients using a custom-built handheld probe.A fl exible sampling fi ber encased in a transparent, fl exible plastic sheath was supported by an outer metal tube (2 mm diameter) and manually guided through the laryngoscope for near-contact or gentle contact imaging of laryngeal mucosa (axial resolution 9 μm, depth 2.6 mm).Images were analyzed for epithelial thickness, layered structure of the mucosa, and microstructural features (e.g., glands, microvasculature).Benign pathology, including Reinke's edema, papillomatosis, polyps, mucous cysts, and granulation tissue were identifi ed based on unique patterns of signal backscattering [ 39 ].In 2008, Kraft et al. used a commercial, contact-mode OCT device (Optimec Ltd.; central wavelength λ = 980 nm, 15 μm spatial resolution) to compare the diagnostic accuracy of microlaryngoscopy with OCT compared with microlaryngoscopy alone in a series of 217 benign and malignant laryngeal lesions [ 40 ].They found that microlaryngoscopy and OCT provided an accurate diagnosis in 93 % of benign lesions and had a higher sensitivity (78 %) than microlaryngoscopy alone (66 %) in predicting epithelial dysplasia; specifi city and accuracy were comparable in both methods.
Additional reports on OCT under direct laryngoscopy or microlaryngoscopy have been described [ 27 , 40 -46 ].Later, spectral domain systems boasting faster imaging speeds (up to 18.5 frames/s) demonstrated improved effi ciency and allow for 3D endoscopic imaging [ 47 ].While imaging through endoscope working channels or handheld probes allow for contact or near-contact imaging, certain factors have been found to limit image quality.Motion artifact due to the movement of the probe tip, enhanced by surgeon's hand tremor, can signifi cantly affect OCT image quality given its high-resolution scale.Furthermore, placement of a probe within the laryngoscope limits the surgeon's visualization of the operative fi eld and interferes with insertion and manipulation of microlaryngeal instruments [ 48 ].To answer this challenge, Vokes et al. fi rst described a hands-free noncontact OCT system, with the

Intraoperative Imaging
Optical Coherence Tomography of the Larynx: Normative Anatomy... sampling probe integrated with a surgical microscope.They integrated a custom built TD-OCT system with a novel interface device attached to an operating microscope by acrylic housing.The interface device consisted of a lens to adjust the focal length of the OCT beam, a galvanometer-mounted mirror to allow for coronal scanning and a second fi xed mirror to redirect the path of light towards the sample tissue.Image frames were acquired at 1.6 mm depth and 6 mm width, with axial resolution of approximately 7 μm [ 48 ].Additional reports of OCT integrated with microlaryngoscopy have been described in the literature [ 49 , 50 ].
OCT probes can be affi xed to fl exible or rigid endoscopes to perform awake, offi ce-based laryngeal imaging.Conjoining the sampling arm with a fl exible endoscope allows for imaging (contact or near-contact) with a fi xed working distance from the vocal folds, easier manipulation of the instrument, and a side-view of the vocal cords.However, direct contact with the vocal folds may precipitate severe cough, gagging or, in rare cases, laryngospasm.In rigid endoscopes cantilevered in the oropharynx approximately 5-8 cm above the vocal folds, LR-OCT systems can scan the larynx in a less invasive manner which may be better tolerated by patients.However, this method requires manual adjustment of the working distance with any movement.
In 2005, Luerssen et al. fi rst described laryngeal OCT in awake patients under local anesthesia.A fi ber-based endoscopic system (Institute of Applied Physics, Nizhny Novgorod, Russia) acquired images with the distal end of the sampling probe (2 mm diameter) in direct contact with laryngeal tissues.OCT data demonstrated clear distinction between epithelial mucosa and the loose, subepithelial collagenous tissue of the LP [ 31 ].Additional studies by Luerssen's group describe a laryngoscope-integrated OCT probe for noncontact, transoral imaging with a low-profi le design to optimize practicality and patient comfort [ 32 -34 ]. Figure 3 demonstrates OCT images from healthy false vocal folds (supraglottic tissues), acquired from a research system (Fig. 3a ) and a commercially available system (Fig. 3b ).This data demonstrates respiratory epithelium, consisting of pseudostratifi ed columnar epithelium with mucous-secreting goblet cells and ciliae.In contrast, OCT of the healthy true vocal folds depict nonkeratinized, stratifi ed squamous epithelium (Fig. 4 ).
Chen and Wong's group constructed a TD-OCT system with a sampling probe fi xed onto a laryngoscope to perform transoral, noncontact OCT [ 51 ].This fi rst-generation system had a slow scanning mechanism (1 frame/s) and produced images with signifi cant motion artifact secondary to patient movement (breathing, swallowing, and refl exes) and physician hand tremor.Later, the same research group developed an FD-OCT system with a "double barrel" handheld carriage integrating a gradient-index (GRIN) lens-based probe

Offi ce-Based Imaging
Giriraj K. Sharma and Brian J.-F.Wong and rigid video endoscope.This instrument acquired cross-sectional images at 8 frames/s, thus reducing the degree of motion artifact [ 52 ].Further pushing the frontier in dynamic imaging of the VF, a swept source (central wavelength λ = 1310 nm) FD-OCT system was designed (Fig. 5a ) and constructed to acquire transoral endoscopic images of the larynx at 40 frames/s using an enhanced long GRIN lens-based probe integrated with a rigid endoscope in a similar double-barrel apparatus (Fig. 5b ) [ 53 ].Lubatschowski's group integrated a swept-source FD-OCT system with a rigid laryngoscope for noncontact OCT and synchronous video imaging of the vocal folds through a singular beam path [ 54 ].Their compact imaging apparatus offered a more practical and well-tolerated means for transoral offi ce-based imaging of the vocal folds.These noncontact, transoral endoscopic systems were the fi rst to provide synchronous OCT and video imaging of the larynx during respiration and phonation without the use of topical anesthesia.Furthermore, noncontact imaging prevents morphometric distortion which may result from direct tissue compression from the tip of the probe.Awake, offi ce-based OCT offers comparable results to contact or near-contact OCT during surgical endoscopy, however, with lower axial resolution (20 μm).Limitations of these modifi ed laryngoscope-based systems include Fig. 3 OCT images of the normal false vocal folds acquired from a research OCT system ( a ) and a commercially available OCT system ( b ).E Epithelium (psuedostratifi ed columnar), BM basement membrane locus, LP lamina propria Fig. 4 OCT images of the normal true vocal fold acquired from a research OCT system ( a ) and a commercially available OCT system ( b ).SS stratifi ed squamous epithelium, BM basement membrane locus, LP lamina propria motion artifact caused by patient movement during the exam (e.g., breathing, swallowing, refl exes) and physician's hand tremor.Furthermore, because the endoscope-OCT systems were cantilevered within the limited space of the pharynx, an extended focal length (compared to systems integrated with surgical endoscopes) was required for laryngeal imaging, limiting the lateral resolution and signal intensity [ 31 , 32 , 51 , 55 ].Currently, long-range swept source systems are being designed for high-speed transoral imaging without compromise of resolution.
In 2006, Klein et al. described laryngeal OCT and PS-OCT in awake subjects using a probe integrated with a fl exible transnasal endoscope.The OCT probe, encased in Tefl on tubing, was advanced through the operating channel of the endoscope until the distal end was visible and placed in contact with the glottic mucosal surface.Images were analyzed for epithelial micrometry, gray-scale intensity variation secondary to birefringent tissue (i.e., collagen content), and unique features of benign processes (e.g., papilloma, cysts, scarring) [ 41 ].Additional reports of awake, offi ce-based, fl exible endoscopic OCT are described in the literature [ 56 ].

Vocal Fold Vibration
Vocal fold vibration has been widely studied under normal and pathologic circumstances using laryngeal videostroboscopy and high-speed video.Lohscheller et al. utilized phonovibrography to translate vocal fold vibration frequency, velocity, and acceleration into 2D diagrams for visualization and analysis [ 57 , 58 ].In examination of patients with laryngeal lesions, clinicians routinely correlate endoscopic fi ndings with characteristics of vocal fold vibration noted on stroboscopy.This information helps to predict the consequence of lesions or interventions on phonation.However, the diagnostic sensitivity and specifi city of endoscopy with videostroboscopy is low, as examination is limited to a surface view of the vocal folds.Thus, noninvasive diagnostic technology is needed to characterize the laryngeal mucosal wave in 3D during phonation.Correlation of high-resolution cross-sectional OCT images with vibratory characteristics of the laryngeal mucosa (e.g., frequency, amplitude) may lead to better understanding of how lesions or interventions affect the mechanical properties of laryngeal mucosal tissue and alter phonation.
In 2006, Luerssen's group fi rst reported in vivo TD-OCT (central wavelength λ = 1300 nm) of the vocal folds during phonation [ 32 ].A fi ber-based OCT probe was integrated into a beam path of a laryngoscope to capture images in noncontact mode with a scanning rate of 10 Hz (resolution 10-20 μm).By measuring oscillation peaks in relation to scanning time, they were able to calculate vibrational frequency.Later, Yu et al. used a swept-source FD-OCT system to image vocal fold oscillation at 40 frames/s [ 53 ].The higher frame rate allowed for minimization of motion artifact and dynamic vibration of the vocal folds.In seated, nonanesthetized patients, an OCT probe attached to a laryngoscope was inserted through the oral cavity and centered above the larynx.Their device allowed for dual-channel endoscopic and video-rate OCT imaging of the vocal folds.In both studies, vibration parameters such as frequency and amplitude were calculated from individual OCT frames.In 2011, Wong and Chen's group demonstrated OCT and optical Doppler tomography (ODT) of vibrating focal folds using a swept-source FD-OCT system (center wavelength λ = 1050 nm) with an imaging speed of 100 frames/s [ 59 ].Their system extended the limit for high-speed functional OCT with image frames rate near that of physiologic fundamental frequencies (females 200 Hz, males 120 Hz).
Typically, however, imaging speeds of OCT systems are insuffi cient to directly capture vocal fold oscillation in the audio frequency range.To answer this challenge, Chang et al. reported motion-triggered laser scanning to capture four-dimensional (4D) images of vibrating ex vivo calf larynges [ 60 ].Their modifi ed OCT system acquired multiple A-lines over a single oscillation cycle at a frequency of 100 Hz before shifting to the adjacent transverse location.This method of data acquisition allows for temporal and spatial registration of A-lines to yield phase-aligned snapshots of tissue oscillation over a complete vibratory cycle.Hence, the imaging frequency range is determined by the A-line rate (up to 200 kHz) instead of the OCT system frame rate.Additional studies of triggered laser scanning of vibrating laryngeal tissue are reported in the literature [ 61 ].

Neonatology and Pediatrics
The neonatal and adult larynges differ signifi cantly in both structural and microanatomic aspects.In the newborn airway, the larynx sits higher up and more anteriorly while the circumferential cricoid results in a cross-sectional narrowing of the airway at the subglottis [ 62 ].Newborn VF mucosa consists of a uniform, monolayered LP composed of ground substances (hyaluronic acid, fi bronectin, fi broblasts, collagenous and elastic fi bers), less fi brous components, and no vocal ligament [ 63 ].During the adolescent years, the LP matures into a layered microstructure.The delicate mucosa of the newborn larynx, coupled with the unique anatomical confi guration of the airway, make the newborn laryngeal and subglottic tissues susceptible to injury following long-term intubation or gastroesophageal refl ux.
Boseley et al. used pediatric cadaveric larynges to describe the maturation of the vocal fold.They noted a differentiation from the monolayered LP into a bilaminar structure beginning by 2 months of age.Transition into a trilaminar structure occurs between ages 1 and 5 years, with full maturation into the adult molecular composition occurring around age 13 [ 64 ].However, it is unknown exactly when these transitions occur and thus diffi cult to estimate at what age microsurgical techniques can be used without affecting phonation.OCT's ability to evaluate the subepithelial microstructure of the pediatric larynx can allow clinicians to assess stage of laryngeal development and develop individualized treatment plans.
Given the unique anatomical c onfi guration of the neonatal airway, patients requiring long-term endotracheal intubation and mechanical ventilation are at risk for subglottic mucosal injury.In managing neonates with suspected subglottic stenosis, pediatric otolaryngologists do not have means to comprehensively evaluate the airway without subjecting patients to general anesthesia and further airway instrumentation.Direct laryngoscopy and bronchoscopy is the gold standard for diagnosis of SGS [ 65 ].However, this procedure may lead to further mucosal abrasion (removal and reinsertion of endotracheal tube, endoscope contact with airway mucosa) and is limited to a view of surface anatomy.Furthermore, risks of general anesthesia in patients with preexisting cardiopulmonary insuffi ciencies must be considered.These factors, combined, often lead to a delay in diagnosis of SGS and increased morbidity.Thus, there exists a need for a less invasive imaging modality which can characterize the subepithelial tissue of the subglottis and identify precursors to subglottic stenosis.at various airway landmarks with a fl exible, handheld probe positioned manually or with endoscopic guidance.Layered microstructure of normal laryngeal and subglottic tissue and distinct pathologic changes (mature scar, granulation tissue, edema, ulceration) were identifi ed on OCT and correlated with endoscopic photographs [ 66 ].The following year, the same group conducted in vivo OCT of the larynx in intubated neonates to identify tissue microstructure of the larynx, subglottis, and proximal trachea.This data provided a framework for using OCT to identify subepithelial injury at the subglottis in neonates under long-term intubation [ 67 ].
Boudoux et al. conducted ex vivo imaging of both pediatric and porcine larynges using FD-OCT, spectrally encoded confocal microscopy, and full-fi eld optical coherence microscopy followed by comparison with histologic section.They noted that the combined application of OCT with confocal microscopy allows for comprehensive evaluation of subepithelial microstructure and cellular and subcellular components of the vocal folds.Combined, these modalities may allow for longitudinal analysis of vocal fold development and differentiation [ 68 , 69 ].

OCT Limitations
The primary limitation of OCT imaging of the larynx is the depth of signal penetration.Current OCT systems offer optical penetration depths of up to 1-2 mm.As benign and malignant processes are differentiated by depth of invasion and disruption of the basement membrane, achieving adequate signal penetration is critical to providing a defi nitive diagnosis of laryngeal pathology.Larger, exophytic lesions of greater than 2 mm depth cannot be identifi ed to their full extent, limiting the diagnostic sensitivity of OCT.Furthermore, certain tissue interfaces (e.g., cartilage, bone) with high optical scattering or absorption properties impede photon penetration and cast a distal shadow within A-lines, further limiting diagnostic sensitivity in such tissues.Additional factors which limit image quality include physician hand tremor and patient movement during offi ce-based OCT, or equipment vibrations which may all translate into image distortion of far greater magnitude.Lastly, the OCT resolution limit (approximately 10 μm) precludes evaluation of laryngeal microanatomy at the cellular level.This degree of microanatomical analysis is fundamental for differentiation between dysplastic, precancerous, and cancerous lesions, necessitating formal histopathologic analysis for a defi nitive diagnosis of malignancy.
At present, near-contact sampling probes offer the best OCT image resolution, clarity, and diagnostic sensitivity.This method of imaging requires general anesthesia for OCT under microlaryngoscopy or adequate topical anesthesia for awake, offi ce-based laryngeal OCT.However, offi ce-based near-contact imaging may Optical Coherence Tomography of the Larynx: Normative Anatomy... not be well tolerated by all patients, given the risk of paroxysmal airway reactions such as coughing, gagging, or choking.While these reactions may be self-limiting and the incidence of true laryngospasm is rare, noncontact imaging may cause discomfort or anxiety to some patients.Current efforts are focused on developing full range OCT systems using vertical-cavity surface-emitting lasers (VCSEL), which may ultimately allow for transoral imaging of the vocal cords from a sampling arm cantilevered in the oropharynx while maintaining current spatial resolution.

Malignancy
OCT has been shown to provide detailed microanatomical information about the laryngeal epithelial layer and integrity of the BM and LP [ 29 -31 ].By identifi cation of small foci of BM breakdown, OCT can identify premalignant or invasive cancerous lesions with high sensitivity [ 70 , 71 ].In 1997, Sergeev et al. were the fi rst to study normal and cancerous tissue of the larynx and reported a "loss of normal tissue stratifi cation in tumors" [ 19 ].Later, Shakhov et al. conducted TD-OCT in 26 patients with small laryngeal squamous cell carcinomas.Similarly, they described a stratifi cation of layered tissue within mucosa of the healthy larynx, the disappearance of which signifi es pathologic changes [ 26 ].Further OCT studies on premalignant and malignant lesions of the larynx are illustrated and reviewed elsewhere in this text [ 39 , 43 , 70 ].

In 2007 ,
Ridgway et al. were the fi rst to use OCT to image the pediatric upper aerodigestive tract.Using a fi ber-based OCT system, they imaged 15 pediatric patients during operative endoscopy Anatomy Future Applications OCT Studies Giriraj K. Sharma and Brian J.-F.Wong