Full anterior segment biometry with extended imaging range spectral domain optical coherence tomography at 1340 nm

Abstract. We demonstrate an extended-imaging-range anterior-segment optical coherence tomography (eAS-OCT) system for the biometric assessment of full AS in human eye. This newly developed eAS-OCT operating at 1340-nm wavelength band is simultaneously capable of an imaging speed of 120 kHz A-line scan rate, an axial resolution of 7.2 μm, and an extended imaging range of up to 16 mm in air. Imaging results from three healthy subjects and one subject with a narrow-angle demonstrate the instrument’s utility. With this system, it can provide anatomical dimensions of AS, including central corneal thickness, anterior chamber width, anterior chamber depth, crystalline lens vault, crystalline lens thickness, angle opening distance (AOD500/AOD750), and the area described by the trabecular-iris space (TISA500/TISA750) at 500/750  μm. We also use eAS-OCT to image and quantify dynamic functional changes of the AS in response to a light stimulus that induces physiological pupillary changes as well as accommodative efforts that induce lens changes. The results show that the described eAS-OCT is able to provide full anatomical biometry for AS and is useful for the studies where the dynamic response of AS compartment to certain stimulus is required.


Introduction
Optical coherence tomography (OCT) is now a popular highresolution optical imaging technology 1 capable of providing images of internal microstructures within biological tissue. The most successful application of OCT has been in ophthalmology, where the technology has become an indispensable tool to image both the anterior segment (e.g., cornea and anterior chamber angle) and the posterior segment (e.g., retina) of the human eye. The OCT technology provides critical information aiding diagnosis, treatment, and management of a wide range of eye disorders. Izatt et al. were the first to report the use of OCT to noninvasively visualize the cornea and anterior chamber in 1994. 2 Since then, extensive efforts have been made to improve the technology and to explore the clinical utility of anterior-segment OCT (AS-OCT). Clinical uses include evaluating iridocorneal angle and identifying risk factors in primary angle closure glaucoma (PACG), [3][4][5][6][7][8][9][10][11][12][13] visualizing cornea, anterior chamber, and crystalline lens in refractive surgery, such as in laser in situ keratomileusis (LASIK) and phakic or pseudophakic intraocular lens (IOL) implantation, [14][15][16][17][18][19][20] assessing the dynamic response of AS during physiological accommodation and assessing ametropia, particularly for eyes undergoing IOL surgery. [21][22][23][24][25][26][27][28][29][30] While different applications have their own requirements for system performance, an ideal AS-OCT system for imaging AS should have (1) sufficient penetration depth so that anatomical features deep within corneoscleral limbus can be imaged; (2) <10-μm axial resolution for visualizing sufficient anatomic details; (3) >10-mm image ranging distance; (4) sufficient imaging contrast for comprehensive evaluation of the whole anatomy from anterior cornea to posterior crystalline lens in one scan; and (5) >100-kHz line-scan rate for rapid acquisition of the complete three-dimensional (3-D) OCT dataset of the whole segment within an acceptable time frame.
Currently, several commercial AS-OCT systems are available for clinical use (Table 1), which can be classified into time-domain (TD), swept-source (SS), and spectral-domain (SD) based configurations. Among them, both the TD-based and the SS-based AS-OCTs operate as a stand-alone system dedicated to AS imaging. The preference of operating wavelength band for these systems is ∼1310 nm because longer wavelength offers deeper penetration depth at the iridocorneal angle, 31,32 permitting quantitative characterization of tissue spaces from the anterior cornea to the posterior crystalline lens. 12,14,17,21 Due to its inherent signal-to-noise ratio advantage, the Fourier-domain (FD) OCT (including SS-based and SD-based configurations) offers higher imaging speed (up to ∼20 to 40 kHz line-scan rate) than those based on TD configuration. Even at these imaging speeds it remains challenging to avoid the motion artifacts and to acquire a high-quality 3-D OCT dataset of the whole AS within a clinically practical time frame. However, such a 3-D OCT imaging is a necessary requirement for comprehensive screening of cornea/anterior segment disorders. Furthermore, the current ranging distance (imaging depth) of ∼6 to 7 mm precludes the imaging of full AS (from anterior cornea to posterior crystalline lens) in one scan, a capability highly beneficial for understanding anatomical/optical conditions in AS. Such an understanding is essential for the management of the cornea in refractive surgery, the management of the anterior chamber angle in PACG, [7][8][9][11][12][13]33 and the management of IOL decisions at cataract surgery as well as in evaluating the mechanisms of dynamic lens accommodation. [24][25][26][27][28] To address the imaging speed limitation, several custom-built AS-OCTs with enhanced system performance have been developed. However, the complex conjugate artifact (inherent in FD-OCT) and the spectral-sampling density in the detecting spectrometer (available number of pixel array detectors in line-scan camera) represent limitations that impose constraints on the ranging distance of the SD-based AS-OCT systems. In the earlier AS-OCT systems, the usable ranging distance was doubled by resolving the complex conjugate artifact. The doubling enhances the ability to visualize anterior crystalline lens, 34,35 delivering results comparable to those of the conventional TD-based AS-OCT systems.
Recently, a prototype ∼830-nm band AS-OCT system was demonstrated that is based on 2048 and 4096 pixel line-scan cameras (increasing the spectral-sampling density of detectors). 28 By combining the higher pixel density with the fullrange complex technique, an imaging depth of up to ∼14 to 16 mm was achieved, thereby successfully covering the full AS. 28,36 However, due to the higher light scattering at shorter wavelengths compared to the 1050∕1300-nm systems, the 830-nm systems suffer from shallower penetration depths at the corneoscleral limbus. 31,32,37 Benefiting from the rapid development of swept laser source technology, several SS-based AS-OCT systems have been demonstrated. 25,[38][39][40][41] Generally, due to the performance tradeoffs between imaging depth, imaging speed, and axial resolution, few of the SS-OCTs are able to provide concurrent features of high resolution (<10 μm), ultrafast speed (>100 kHz), and sufficient ranging distance (>12 mm) for 3-D imaging of the entire AS. Recently, a new swept laser based on an MEMS tunable vertical cavity surface-emitting laser (VCSEL) was demonstrated for SS-OCT imaging, 42 offering negligible sensitivity roll-off over 1.0 cm depth. The swept laser system provides the balanced system performances among penetration depth (working at 1060-nm wavelength), axial resolution (9 μm), imaging depth (13.6 mm), and imaging speed (100 kHz line-scan rate), which are much more attractive for imaging the full AS. 42 In parallel to the development of SS-OCT, an extendedimaging-range SD-OCT (eSD-OCT) system 43 was proposed for AS imaging, delivering performance comparable to that of a typical AS-OCT based on MEMS tunable VCSEL. 42 Although the VCSEL-based AS-OCT system provided superior sensitivity roll-off performance, 42 the eSD-OCT also successfully demonstrated the capability of full AS imaging in one scan with axial resolution and imaging speed similar to that of VCSEL-based ASOCT. 43 An advantage of the eSD-OCT system is its usage of the same types of technology as the 1050-nm band. Adoption of this standard technology then permits the use of almost all relatively inexpensive, readily available components in the current commercial systems. In particular, the eSD-OCT is based on the SD configuration. Compatibility with the current standard commercial SD-OCT systems provides an inherent advantage because it offers potential for a straightforward manufacturing and clinical translation.
In this study, we report a newly developed SD-based eAS-OCT operating at 1340 nm wavelength band that is capable of simultaneous 7.2 μm imaging resolution, 16 mm ranging distance, and 120 kHz A-scan rate. We then demonstrate static/anatomical full biometry of AS, investigating dynamic lens accommodation and studying the responses of AS compartments to light stimulation using the proposed eAS-OCT. Finally, we discuss the superiority, limitation, and potential applications of eAS-OCT for cornea, cataract, IOL, glaucoma, and refractive surgeries.

Material and Methods
The schematic of the prototype eAS-OCT used in this study is shown in Fig. 1, similar to the system reported in Ref. 43. A broadband superluminescence diode with a central wavelength of 1340 nm and a spectral bandwidth of 110 nm (LS2000B, Thorlabs Inc., USA) is used to illuminate the system, providing an axial resolution of ∼7.2 μm in air. This light source gives an enhanced penetration for OCT imaging, facilitating the visualization of the iridocorneal angle due to reduced light scattering in tissue compared to light sources with shorter wavelengths. In addition, an objective lens with a 100 mm focal length is installed in the sample arm offering a lateral resolution of ∼50 μm and a depth of focus of ∼3 mm. The focal plane of the objective lens is placed at the zero-optical-length-difference plane, which is situated close to the iris for balanced imaging quality of full AS. The detection system is a high-speed spectrometer equipped with a transmission diffraction grating (Wasatch Photonics, USA) with 1145 lines∕mm, an achromatic camera lens with a focal length of 100 mm, and a prototype InGaAs line-scan camera with a 2048 pixel-array detector capable of a line-scan rate of 120 kHz (Sensors Unlimited Inc., part of United Technologies Aerospace Systems, USA). The spectral resolution of the spectrometer is 0.056 nm, offering a measured imaging range of ∼8 mm and a sensitivity rolloff of 6 dB∕4 mm. After removing the complex conjugate artifact using a full-range complex technique, 44-46 the resultant imaging depth (ranging distance) is extended to ∼16 mm in air. With this design, the system sensitivity at 0.5 mm depth is measured at ∼100 dB.
In the sample arm, the scanner is designed to provide a telecentric scan in horizontal plane for minimal optical distortion. 47,48 For imaging, two raster scanning patterns are designed to drive the X-Y scanner: static 3-D mode and dynamic twodimensional (2-D) mode. In static 3-D mode, each B-frame is composed of 3000 A-lines with 2048 pixels per line, providing a high resolution and high definition cross-sectional image while fulfilling the requirement to achieve a full-range SD-OCT. 28,43,46 With a 120-kHz line-scan rate, it takes ∼2.5 s to complete a 3-D scan consisting of 2048 × 3000 × 100 voxels (16 × 16 × 12 mm 3 in air). For dynamic 2-D mode, the Y-scanner is disabled. The imaging is performed by repeated scanning at the same B-scan location. The duration of the scan is ∼8.25 s, allowing a total of 330 successive cross-sectional images to be acquired for assessing dynamic activities of the human eye.
We then demonstrate the system's capability to provide anatomical biometry for full AS, to study dynamic behavior resulting from lens accommodation and to study effects caused by light stimulation. During imaging, the sample is exposed to a light power of 2.0 mW, within the American National Standards Institute safety limit. The use of in-house OCT systems to image the AS of the human eyes was approved by the institutional review board at the University of Washington. Before imaging, informed consent was obtained from each subject.
For biometric analyses of the AS, optical distortion in OCT images was corrected based on Snell's principle, 48 in which the refractive indices of 1.38, 1.34, and 1.42 were used for cornea, aqueous humor, and crystalline lens, respectively. 25 The crosssectional slice including the corneal vertex location was extracted from the 3-D dataset at the horizontal meridian by using a custom semiautomatic algorithm. The distal limit of each angle recess was identified and marked. The distance from recess to recess was recorded as the anterior chamber width (ACW). A recess-to-recess line was drawn, and its perpendicular projection extending from the median point passing through the anterior cornea and posterior crystalline lens was generated (refer to Fig. 2). The perpendicular line can be considered as the anteroposterior (AP) axis. Several quantitative parameters, defined along the AP axis, were measured in this study. Table 2 summarizes the definition of these parameters: the central corneal thickness (CCT), AC depth (ACD), the crystalline lens vault (LVa), and the crystalline lens thickness (LT). These parameters have been defined and measured respectively in previous studies. 14,17,21,24,49 To quantify iridocorneal angle configuration (refer to Fig. 3), the scleral spur was first marked. Based on this landmark, the perpendicular distances from the trabecular meshwork to the iris at 500 and 750 μm anterior to scleral spur were automatically recorded as the angle opening distance (AOD500 and AOD750, respectively). 3,6,50 The trabecular-iris space area at 500∕750 μm (TISA500 and TISA750, respectively) was also calculated as the trapezoidal area bounded anteriorly by the AOD500 or AOD750, posteriorly by a line perpendicular to the inner scleral wall at the scleral spur, superiorly by the inner corneoscleral wall, and inferiorly by the iris surface. 3,5,6 Additionally, the radius of crystalline lens anterior surface (LR) was determined by measurements that permitted circular fitting to the anterior lens surface. In the same manner, the radius of cornea anterior and posterior surface (CRa and CRp) can be quantified. Pupil diameter (PD) was calculated as pupil marginto-margin distance.

System Performance Assessment
To assess how accurate the eAS-OCT system is to the biometric assessments of the key parameters, we elected to use a commercially available model eye (OEMI-7, Ocular Instruments Inc., Bellevue, WA) because its dimensions are known in prior. The model eye is composed of main ocular parts, including the cornea, pupil, crystalline lens, aqueous and vitreous media, as well as the retinal surface. The cornea and lens are composed of poly(methyl methacrylate) (n ¼ 1.49), with the deionized water (n ¼ 1.33) serving as the aqueous and vitreous media. 3-D OCT imaging was performed and repeated five times (the eye was repositioned into the system each time). The AS parameters of the model eye were measured. Because there is no angle recess in the model eye, the AP axis was determined based on the pupil for the system performance assessment.

Anatomical Biometry for Full Anterior Segment
Study enrollment to demonstrate the utility of eAS-OCT included four healthy subjects. Three of them (ages 23, 31, and 46) had no history of ocular disease and only modest refractive errors (0, −3 diopter, and −1 diopter, respectively). The fourth was a narrow-angle glaucoma suspect (age 24). Only the undilated right eye of each subject was imaged. The imaging was performed in a room with ∼500 lux interior illumination. Each imaging session was repeated five times for each eye (repositioning was required from the subject each time). During scanning, subjects were instructed to fixate on a distant target consisting of an "E" to ensure an absence of accommodation. The AS parameters and iridocorneal angle configuration were determined for each eye.

Lens Accommodation
To demonstrate the capability of imaging the lens response to accommodation, we used the system to image a 31-year-old subject who first was instructed to fixate on a distant target "E" [0 D (diopter), accommodation relaxed]. The subject was then instructed to fixate on a second target "E" set at a distance of 0.11 m, providing a 9 D accommodative stimulus. In addition to the parameters listed in Table 2, LR and PD were also calculated.

Light Stimulation
Dynamic 3-D imaging is challenging using the currently available AS-OCT imaging speeds. Therefore, to demonstrate the benefits of the eAS-OCT system, imaging was initially conducted under a bright (∼500 lux) and a dark (<1 lux) background, respectively. Room illumination was then dimmed to <1 lux. As illustrated in Fig. 4, the subject was asked to fixate on a laboratory-designed light-emitting diode target flashing at a frequency of ∼1 Hz; the scanning protocol was then switched to the dynamic 2-D mode to capture the real-time dynamic response to the flashing light stimulus.  Table 3, along with the values provided by the manufacturer. These data indicate that the eAS-OCT can deliver the biometric measurements with very good accuracy. In addition, the system also provides good consistency between each imaging session (as demonstrated by the small values of standard deviation).   The primary anatomic features in AS are clearly visible in the eAS-OCT images, including cornea, sclera, iris, and crystalline lens. More importantly, important landmarks necessary for the biometric assessment of AS can be easily identified, such as the scleral spur and angle recess (arrows in Fig. 6). Using these landmarks, the quantitative evaluation of the full anatomic architecture and iridocorneal angle configuration are readily achieved, as illustrated in Figs. 2(a) and 2(b), respectively. Based on the biometric assessments for healthy subjects, the mean values of the parameters are CCT ¼ 0.56 AE 0.06 mm, ACW ¼ 12 AE 0.57 mm, ACD ¼ 3.25 AE 0.28 mm, LVa ¼ 0.26AE 0.34 mm, and LT ¼ 3.77 AE 0.36 mm. These derived values are in good agreement with previously reported ones by other groups using different AS-OCTs, including custom-built and commercial TD-based AS-OCTs, 12,14,17,21,29,51 SS-based AS-OCT, 25,49 and SD-based AS-OCT. 26 Furthermore, our results for ACD, LVa, and LT are in agreement with age-related correlations reported in previous publications. 12,13,21,23     The parameters for iridocorneal angle configuration, e.g., AOD500 (mean ¼ 0.66 mm), AOD750 (mean ¼ 0.82 mm), TISA500 (mean¼0.23mm 2 ), and TISA750 (mean¼0.42mm 2 ), are within the same range as the measurements reported in Refs. 3, 5, and 6. The small variances seen in our study compared with previous reports are most likely due to individual iris and lens factors. 12,13 The youngest subject has the most concave iris, largest pupil size, and weakest OCT signal in the crystalline lens while having the widest iridocorneal angle, consistent with expected findings for young subjects.

Biometry of the Full Anterior Segment
By contrast, for the narrow angle suspect, we observe not only narrower angle configuration, but also smaller ACD, larger LVa and LT, obvious iris convexity, and increased thickness of the iris compared to those observed from subjects who were not  angle closure suspects. All of these findings are in excellent agreement with recent studies on PACG that reported a significant relationship of angle closure to anterior chamber, iris, and lens parameters, such as ACD, iris thickness, iris curvature, LVa, and LT. 7-9,11-13,33,52 The measured geometrical changes of the AS during accommodation indicate that several actions occur in response to an accommodative effort: (1) the lens thickness is increased; (2) the anterior lens pole moves toward the cornea, while the posterior pole changes only to a small extent; (3) the anterior lens surface becomes convex (most likely due to the decrease of the lens diameter); (4) synkinetic myosis occurs; and (5) the iris becomes concave due to the upward movement of anterior pole of the lens. All these observations are in accordance with the accommodation theory described in Ref. 53. However, the mechanism of accommodation is still not completely understood. The capability of visualizing and quantitatively assessing the in vivo 3-D architecture of the AS in a noncontact and noninvasive fashion provides a powerful, user-friendly approach to gain further understanding of the mechanism of accommodation and the changes associated with ametopia and aging. Figure 7 shows the results from the eAS-OCT imaging under bright (∼500 lux) and dark (<1 lux) room illumination, respectively. The iris thickness is increased and the pupil is dilated considerably under dark illumination (2.44 mm increase in pupil diameter) compared to that under bright room illumination. Figure 8 shows the real-time dynamic cross-sectional imaging of the same subject in response to a flashing light stimulus. The PD undergoes oscillatory change in size that is synchronous with the flashing light as plotted in Fig. 8(b).

Discussion and Conclusions
PACG is one of the leading causes of blindness, especially prevalent in Asian population, and is characterized by closure of the iridocorneal angle. Recent studies indicate that several important factors may be associated with the angle closure, such as ACD, LVa, LT, iris thickness, iris curvature, and iris dynamics. [7][8][9][10][11][12][13]33,52,54,55 We have demonstrated that eAS-OCT is capable of evaluating the angle configuration and providing quantitative assessments of all these parameters using one device. Therefore, we believe that eAS-OCT may become an efficient imaging tool for better understanding of the risk factors and pathogenesis in angle closure. By quantitatively assessing the parameters that increase PACG risk, patients can be better identified and decisions about monitoring or intervention can be made based on the combination of objectively obtained parameters.
In recent years, refractive surgeries, including corneal laser ablation surgery and phakic or pseudophakic IOL implantation, have been increasingly used for the refractive correction of patient eyes, but planning of such surgeries present several challenges. 15,16,[56][57][58] To provide optimal postoperative refractive results, precise preoperative measurements of anatomical dimensions (CCT, ACD, ACW, LVa, and LT) in AS are essential. The eAS-OCT is able to accurately quantify not only the anatomic dimensions of the full AS, but also the anatomic changes that characterize the functional parameters associated with accommodation. This extended performance has clear implication for clinical applications, including preoperative design/selection of appropriate surgery parameters, intraoperative surgery guidance, and postoperative evaluation, e.g., longterm assessment of procedure safety. In addition, this system provides a means for improving IOL designs through its ability to assess accommodation-induced IOL anterior-posterior position changes, while at the same time being able to characterize any associated design-dependent deformation of the IOL itself.
The proposed eAS-OCT is inherently compatible with the current standard commercial SD-OCT systems offering significant opportunities for consolidation of instrument functions. Integration of the SD-based eAS-OCT into commercial systems offers a combination of high resolution and fast speed of current SD-based AS-OCT, while at the same time providing the long ranging distance similar to that of the TD-based AS-OCT. With simultaneous high axial resolution, fast imaging speed, and extended imaging depth consolidated into one device, the eAS-OCT may, in addition, find valuable applications in areas such as in the assessment of the health of ocular surfaces to determine effects and complications from medications used in open-angle glaucoma, assessment of positioning and behavior of intrastromal corneal rings, as well as decision guidance before and monitoring after corneal transplantation.
Nevertheless, the eAS-OCT described here still has several limitations. For example, the sensitivity roll-off is limited to ∼6 dB∕4 mm. With this limitation, however, sufficient signal has been obtained at the center of cornea and crystalline lens, which would make the system clinically acceptable for imaging the full AS. In addition, with a lateral resolution of 50 μm, the current system offers a depth of focus of only ∼3 mm, giving rise to reduced OCT signal strength for the tissue structures outside of the depth-of-focus region, creating a potential problem for imaging whole cornea that is typically of curved structure with increased oblique angle at the peripheral edge regions. This was the primary reason why the OCT signals were much weak at the oblique parts of cornea and crystalline lens in this study. In future, more sophisticated scanning optics needs to be designed, through synergistic optimization between the lateral resolution and the depth of focus.
In summary, we have designed and demonstrated clinically useful applications for a 1340-nm eAS-OCT that is capable of imaging the full AS at 120 kHz line-scan rate, at the same time attaining 16 mm imaging range and ∼7.0 μm axial resolution. Such an extended system performance permits visualization and quantification of the static/anatomical dimensions of the AS (i.e., biometry of the full AS). At the same time, this same device can characterize and quantitate several dynamic/physiological ocular parameters, such as the lens or IOL response to accommodative efforts and the pupil response to a light stimulus. Potential clinical applications could be found in cornea, cataract, IOL, glaucoma, and refractive surgeries.