Identification of Schlemm's Canal and Its Surrounding Tissues by Anterior Segment Fourier Domain Optical Coherence Tomography

Abstract

Purpose.: To identify Schlemm's canal (SC) and trabecular meshwork (TM) by anterior segment Fourier-domain optical coherence tomography (AS-FD-OCT) with histologic confirmation in enucleated human eyes and to quantitatively evaluate SC and TM in living human eyes.

Methods.: In enucleated human eyes, the imaging of the anterior chamber angle by AS-FD-OCT was performed before and after surgical expansion of SC with an injection of a viscoelastic material, followed by histologic examination. In 60 living human eyes, the agreement of identification of SC between examiners was evaluated with the Cohen's κ values, and the lengths of SC and TM and the area of TM were measured on temporal and nasal sections of the AS-FD-OCT images.

Results.: In enucleated human eyes, SC was observed to be a thin, linear, lucent space in the AS-FD-OCT image obtained with the high-definition raster scan protocol, but not in those obtained with the bi-angle radial scan protocol. This space was enlarged after the SC expansion. In the histologic study, the SC was confirmed to be in the same position as in the AS-FD-OCT images. The κ values of observable SC in living human subjects were 0.92 or higher. The axial length of the SC averaged 347.2 ± 42.3 μm, TM length 466.9 ± 60.7 μm, and TM area 0.0671 ± 0.0058 mm². These measurements showed sufficient repeatability and reproducibility.

Conclusions.: Using the high-definition images of the AS-FD-OCT, SC and its surrounding tissues were successfully observed in most of the living eyes and were quantitatively evaluated in a noninvasive manner.

Noncontact and noninvasive imaging of the anterior ocular segment has become possible using newly developed optical coherence tomography (OCT) techniques. ¹ An anterior-segment time-domain OCT (AS-TD-OCT), which provides the resolution of approximately 20 μm of angle structures with 2000 A-scans per second, enables precise morphometric evaluation and exhibits good reproducibility. ^2,3 A subsequent generation of AS-OCT, AS-FD-OCT has been developed recently. The FD-OCT has wavelength-tunable lasers that use a swept-source technique, and the system exhibits improved imaging speed (25,000 A-scans per second) with a finer resolution (∼10 μm). ^4,5  

AS-OCT techniques are now used for many clinical applications including angle evaluation, ^6,7 corneal pachymetry, ^{8 –10} and bleb analysis. ^{11 –13} Evaluation of the peripheral anterior chamber configurations, including Schlemm's canal (SC) and the trabecular meshwork (TM), is important for the screening, diagnosis, and treatment of glaucoma, especially angle-closure glaucoma, which is one of the leading causes of visual impairment in Asian countries. ^{14 –17} Detailed evaluation of the angle structures, including SC and TM, is difficult with the AS-TD-OCT. ¹⁸ Asrani et al. ¹⁹ recently reported that AS-FD-OCT permits more detailed imaging of angle structures. OCT imaging of SC and TM, however, may easily produce several undesirable artifacts, probably due to the coexistence of several tissues such as the cornea, sclera, SC, and TM. This is because these tissues may have different light reflection properties and different polarization characteristics. Therefore, the anatomic orientations of the angle structures in OCT images should be confirmed at least one time by an alternative means such as comparison with histologic findings. At the time of our study, no such comparison has been done in human eyes. Moreover, the quantitative and statistical study of OCT imaging of SC and TM in living human eyes has not been performed to date. The goals of the present study were (1) to identify SC and TM in high-definition (HD) images generated via AS-FD-OCT and to compare their identification with the histologic examinations of enucleated human eyes and (2) to assess the potential of AS-FD-OCT for use in the quantitative evaluation of SC and TM in living human eyes. 

Materials and Methods

The present study involving human subjects was approved by the institutional review board of University of Tokyo School of Medicine (Tokyo, Japan), and written informed consent was obtained from all subjects in accordance with the Declaration of Helsinki. Before the AS-FD-OCT examinations, comprehensive ophthalmic examinations were conducted including autokeratorefractometry, uncorrected and corrected visual acuity measurements, slit lamp examination, applanation tonometry, indirect funduscopy, and measurements of axial length and anterior chamber depth using the IOL-master (Carl Zeiss Meditec, Dublin, CA). 

AS-FD-OCT Examination

Imaging of the anterior segment was performed via AS-FD-OCT (SS-1000 CASIA; Tomey Co., Nagoya, Japan) by an experienced ophthalmologist (KM) who was unaware of the gonioscopic findings. The AS-FD-OCT was based on the optical frequency-domain imaging technology. ^5,18 The light source had a center wavelength of 1300 nm, which enabled greater penetration of the ocular tissue than with the 830-nm light source used in conventional OCTs for the fundus examinations. In the present study, the AS-FD-OCT imaging was first performed according to two different protocols: the bi-angle radial scan protocol (bi-angle mode) and the HD raster scan protocol (HD mode). In the bi-angle mode, the anterior chamber was imaged with 128 radial B-scans, centered on the pupil center by a fixation light. Each scan included 512 A-scans on 16-mm line that ran from one angle to the opposite one. This scan protocol took 2.4 seconds. In the HD mode, a rectangular area of 8.0 × 4.0 mm centered on the limbus was imaged with 64 horizontal raster B-scans and 512 A-scans, with a duration of 1.2 seconds. Six nearby images were automatically averaged to reduce speckle noise. 

Location of SC and TM in Human Enucleated Eyes

To confirm agreement on the locations of SC and TM between the AS-OCT images and the histologic figures in human enucleated eyes (n = 2), the AS-OCT imaging was performed before and after the surgical expansion of the SC by injection of a viscoelastic material mimicking the procedure of viscocanalostomy, and the microscopic histology of SC and TM was then further studied. The human eyes were imported from the Northwest Lions Foundation Eye Bank (Seattle) and preserved (Optisol-GS; Chiron Intraoptics, Irvine, CA) until the experiments. At first, an enucleated eye was fixed on the chin rest of the AS-OCT, and intraocular pressure was measured by pneumatic tonometry. The intraocular pressure was then adjusted to approximately 20 mm Hg via the intravitreous injection of saline, and the AS-OCT images were then generated immediately thereafter, by using protocols for the bi-angle and the HD modes. The studied locations were marked with 10-0 nylon suture at the limbus. The AS-OCT images were obtained several times within 30 seconds, and the images with the best quality were used for the evaluation. Subsequently, one eye was fixed in the surgical procedure mimicking viscocanalostomy performed at the limbus approximately 90° away from the position of the AS-OCT images. The surgical procedure was performed as follows: A limbal-based 3 × 3 mm outer scleral flap and 2 × 2 mm inner scleral flap were made to expose the SC. Approximately 0.1 mL of the viscoelastic material (Healon; AMO Japan, Tokyo, Japan) was gently injected into both sides of the SC by using a 27-gauge blunt needle. After the manipulation, the eye was again fixed on the chin rest of the AS-OCT, the intraocular pressure was adjusted as described above, and the AS-OCT imaging was repeated at the same locations in the same manner described above. Subsequently, the eye was also subjected to histologic examination as described. 

Since the OCT images from below the cornea were deformed due to light refraction through the cornea, the images were mathematically corrected according to the curvature of the anterior corneal surface. On the B-scan images, the SC was defined as the length of the thin, black, lucent space mentioned above, and the meridional axial length of the SC (SC length) was measured with the software in the apparatus. Paraffin blocks were cut every 10 μm of the captured area, and histologic sections were stained with hematoxylin and eosin. Microscopic photographs were then captured and analyzed to measure SC length. Working magnification on the computer monitor was 600×. 

Identification and Quantification of SC and TM in Living Human Eyes with the AS-FD-OCT

The AS-FD-OCT measurements were performed in 60 eyes of 30 subjects, including 34 normal eyes of 17 subjects and 26 eyes of 13 subjects with a shallow peripheral anterior chamber with a van Herick's angle grade of 1 or 2. After written informed consent was received and the comprehensive ophthalmic examination was completed, imaging with the AS-FD-OCT was performed according protocols of the bi-mode and the HD modes as previously noted. With the bi-angle mode, figures of the temporal and nasal angles (3 and 9 o'clock positions) were obtained simultaneously in one horizontal image centered on the corneal center. With the HD mode, the temporal and nasal limbus was imaged separately after adjustment of the fixation to the nasal and temporal area. The superior and inferior limbi were not observed in the present study, because these portions were often covered with the upper or lower eyelids. During the examination, the subjects were encouraged to open their eyes as wide as possible, and, if necessary, the examiner gently helped them to keep their eyes open by using his fingers, taking care to avoid putting pressure on the eye. An eye speculum was not used. If the eye nictitation was noted during the examination, the examination was repeated up to three times. 

Two experienced examiners (TU, AT) independently judged either the presence or absence of the definitive figures of SC in the AS-FD-OCT images. In images of the bi-angle mode, SC was judged as observable when at least a part of the thin, black, lucent space, which had been confirmed to be the SC in the eye-bank eye study described above, was found in 2 or more consecutive horizontal B-scan images of 128 images. In images of the HD mode, the SC was judged to be observable when a thin, black, lucent space was found in two or more consecutive horizontal B-scan images and a thin, lucent line along the outer curve of the anterior chamber was confirmed in the reconstructed vertical images. For an evaluation of the agreement on the presence of SC on the OCT imaging between the examiners, Cohen's κ was calculated. The κ values were interpreted as follows: 0, no agreement; <0.4, poor agreement; 0.40 to 0.59, fair agreement; 0.60 to 0.74, good agreement; 0.75 to 0.99, excellent agreement; and 1, perfect agreement. 

In eyes, in which the SC was observable with the HD mode, the length of SC and TM, and area of TM were quantified using software included in the apparatus (Fig. 1). The images were also mathematically corrected according to the curvature of the anterior corneal surface. On the corrected B-scan images, SC length was defined as the meridional axial length of the thin, black, lucent space mentioned earlier. The scleral spur was defined as the point between the TM and the ciliary body (Fig. 1). The area of the TM was drawn freehand and depicted the area surrounded by the scleral spur, the posterior endpoint of the SC, and the anterior endpoint of the TM. TM length was defined as the meridional axial length between the scleral spur and the anterior endpoint of TM (Fig. 1). 

Figure 1.

 

SC in a corrected B-scan image in angle HD mode. The length of the thin, black, lucent space in the angle was defined as the length of SC (a). The area of TM (b) was drawn in freehand depicting the area surrounded by the scleral spur (c), the posterior end point of SC (d), and the anterior end point of TM (e), which is the point of intersection of the SC line and Descemet's membrane. TM length (f) was measured from the scleral spur (c) to the anterior end point of the TM (e).

The repeatability and reproducibility coefficients and intraclass correlation coefficients (ICCs) for the measurements of SC length, TM length, and TM area were also assessed. The coefficient of repeatability was defined as 2 SD of the differences between the measurements obtained for the same subjects obtained in a different session by the same observer (TU). The coefficient of reproducibility was defined as 2 SD of the differences between the measurements obtained for the same subject obtained at the same visit by different observers (TU and NH). 

Results

Location of SC and TM in Human Enucleated Eyes

Before the procedure that mimicked viscocanalostomy, a thin black space connecting to the scleral spur was found in the horizontal image of the HD mode (Fig. 2A) and then confirmed in the vertical section, which was reconstructed using the horizontal images (Fig. 2B). These findings were obtained in both eyes. After the viscocanalostomy, the black space in the angle became wider, indicating that the space was surely SC in both the horizontal and reconstructed vertical images (Figs. 2D, 2E). In the histologic sections, SC after the surgical procedure was observed to be dilated when compared with SC before surgery (Figs. 2C, 2F), which corresponded well with the OCT findings. SC length was determined as 354 μm on the OCT image and 384.6 μm on the histologic sections obtained before the viscocanalostomy and 430 and 472 μm after the procedure. 

Figure 2.

 

Angle structure in human enucleated eyes. Before the pseudo viscocanalostomy was performed, a thin lucent space (arrow) was also visualized, even in human enucleated eyes by the angle HD mode due to their tipped position (A, B). After surgical manipulation, this thin space (arrow) was apparently enlarged (D, E). Histologic sections supported our conclusion that this thin black space is SC (arrows, C, F).

Identification of the SC in Living Human Eyes in Images of the AS-FD-OCT

With regard to the 60 eyes of 30 subjects, the subjects' age averaged 56.7 ± 14.8 (mean ± SD) years (range, 29–81); refractive error (spherical equivalent), −1.3 ± 3.8 D (range, −6.25 to +3.13); axial length, 24.12 ± 1.76 mm (range, 21.18–27.99 mm); and anterior chamber depth, 3.05 ± 0.48 mm (range, 2.43–3.82 mm). The angle images generated using the bi-angle mode (average observable area, 274.3° ± 10.8°) are shown in Figure 3. In images of the bi-angle mode, SC was not observable in any eyes, even in high-magnification images (Fig. 3). In images generated using HD mode, a thin, lucent, linear space was located anterior to the TM and could be visualized in the horizontal and reconstructed vertical sections (Fig. 4). These observations all corresponded well with ex vivo studies in human enucleated eyes as mentioned above (Fig. 2). 

Figure 3.

 

Horizontal sectional images captured by bi-angle mode in a radial scan (A; low magnification, B; nasal high magnification, C; temporal high magnification). This case is the left eye of a 53-year-old male. The observation area was centered on the pupil, using the bi-angle mode.

Figure 4.

 

Sectional images captured by angle HD mode using a raster scan. Case A is same subject as in Figure 1. Top: a thin hyposignal lucent space (arrow) was visualized in HD mode, together with lateral eye position, which was centered on the limbus. Reconstructed vertical images from horizontal sections also show a thin lucent line (arrow) along the outer curve of the anterior chamber that appears to be SC. Case B (45-year-old man) is another example of detection of SC (arrow). C, cornea; AC, anterior chamber; I, iris.

The κ values for judgment of an observable SC in the images generated with the HD mode between the examiners were 0.96 and 0.96 in the temporal and nasal sections of right eyes and 0.92 and 1.0 in those of left eyes, indicating excellent agreement between examiners. The frequencies of eyes in which SC was completely or partially observable in HD mode images are summarized in Table 1. SC was determined to be observable when both examiners judged that it was. If the length of SC in the OCT image was deemed identical with the full length of SC expected according to the histologic study, the SC was further judged to be completely observable. The percentage of sections in which SC was observable at least partially ranged from 85.0% to 90.0%, showing no statistically significant differences between right and left eyes (P = 0.52, Fisher exact test) and between temporal and nasal sections (P = 0.44). As the results of the logistic regression analysis indicate, neither the age (P = 0.41), refractive error (P = 0.27), axial length (P = 0.33), or anterior chamber depth (P = 0.36) were significantly associated with SC's observability. 

Table 1.

 

Eyes with Observable SC in Cross-sectional Images

	Completely Observable	Partially Observable	Totally Observable
Right eyes (n = 30)
Temporal	19/30 (63.3)	7/30 (23.3)	26/30 (86.7)
Nasal	18/30 (60.0)	7/30 (23.3)	25/30 (83.3)
Left eyes (n = 30)
Temporal	20/30 (66.7)	6/30 (20.0)	26/30 (86.7)
Nasal	27/30 (90.0)	1/30 (3.3)	28/30 (93.3)
Total (n = 60)	84/120 (70.0)	21/120 (17.5)	105/120 (87.5)

 Data are the number with observable SC (% of total group) obtained with the high-definition angle mode scan of the AS-FD-OCT in 60 eyes of 30 subjects.

Measurement of SC Length, TM Length, and TM Area by AS-FD-OCT in Living Human Eyes

SC length, TM length, and TM area were measured using only images in which SC was completely observable. In total, SC length averaged 347.2 ± 42.3 μm, TM length averaged 466.9 ± 60.7 μm, and TM area averaged 0.0671 ± 0.0058 mm² (Table 2). Differences in parameter measurements were not statistically significant between temporal and nasal sections and between right and left eyes. Age, refractive error, axial length, or anterior chamber depth did not exhibit a statistically significant association with SC length, TM length, and TM area, as indicated by the results of multiple regression analysis. 

Table 2.

 

Measurements of SC and TM Parameters in Cross-sectional Images

	SC Length (μm)		TM Length (μm)		TM Area (mm2)
Right eyes
Temporal (n = 19)	337.3 ± 43.1	P = 0.55	470.5 ± 50.3	P = 0.75	0.0663 ± 0.0059	P = 0.33
Nasal (n = 18)	333.4 ± 38.9		470.0 ± 57.8		0.0682 ± 0.0058
Left eyes
Temporal (n = 20)	353.1 ± 37.7	P = 0.18	474.2 ± 58.5	P = 0.19	0.0672 ± 0.0058	P = 0.80
Nasal (n = 27)	358.9 ± 45.0		456.4 ± 62.5		0.0672 ± 0.0059
Total (n = 84)	347.2 ± 42.3		466.9 ± 60.7		0.00671 ± 0.0058

 Images were obtained with the high-definition angle mode scan with the AS-FD-OCT.

Table 3 shows the repeatability and reproducibility of the SC length, TM length, and TM area measurements by AS-FD-OCT. The repeatability and reproducibility against those parameters were all excellent with AS-FD-OCT. 

Table 3.

 

Repeatability and Reproducibility of Measurements of SC and TM Parameters in Cross-sectional Images

	SC Length	TM Length	TM Area
Repeatability
ICC	0.97	0.90	0.93
Coefficient of variability	0.11	0.12	0.08
Reproducibility
ICC	0.91	0.95	0.89
Coefficient of variability	0.11	0.12	0.08

 Images were obtained with the high-definition angle mode scan with the AS-FD-OCT.

Discussion

Our knowledge has been limited, largely due to the lack of a noninvasive imaging system that permits visualization of SC and TM in vivo, because SC and TM are hardly detectable using devices such as ultrasound biomicroscopy (UBM) and AS-TD-OCT. However, the AS-FD-OCT now enables us to visualize more details of angle structure due because of its higher resolution. The present study revealed that SC was observable by AS-FD-OCT at high detection rate without any discomfort to the subjects. Further, with the software used in conjunction with the AS-FD-OCT system, we were able to measure the length of SC and TM and the area of TM. To the best of our knowledge, this is the first study that discusses measurement of SC and TM by AS-FD-OCT in vivo. 

The SC is a circular channel in the eye that lies in the outer portion of the internal scleral sulcus and conducts aqueous humor from the trabecular region to the episcleral venous network via the collector channels. This canal is lined with endothelium. ²⁰ The lumen is elongated and oval and may sometimes appear triangular in a cross-section due to being divided by septa or exhibiting multiple channels. ¹⁹ Since water is not depicted in the images of OCT, it is reasonable that the inside space of SC appears as a thin lucent black space in the images. Asrani et al. ¹⁹ recently published a paper that suggested that another system of AS-FD-OCT could also detect SC and TM. They reported that the SC was an arched-shape black space that was located at a depth of two thirds of the corneal thickness from the corneal surface at the limbus, ¹⁹ which was not consistent with our observations. However, in our histologic study using the enucleated eyes, SC was not apparently arched shaped and was located more deeply than two thirds of the corneal thickness (Fig. 2). We also found that the seemingly arched-shaped black space sometimes coexisted with the thin black space identified as SC in the same sectional images (Fig. 4), suggesting that this arched black space is not SC and the white hypersignal portion surrounded this lucent space is not the TM. Although the origin of the arched-shaped black space in the OCT images could not be determined, the OCT images of the angle structure are adversely affected by artifacts that probably formed due to the coexistence of several tissues that have different light reflection properties and different polarization characteristics, such as the cornea, sclera, SC, and TM. Several reports have recently suggested that shorter wavelength OCT (∼800–900 nm) enables observers to visualize angle structures including SC ²¹ (Aung T, et al. IOVS 2010;51:ARVO E-Abstract 3855), although, with shorter wavelength OCT, it is difficult to view the angle recess. It is mandatory to compare the visualization of angle structures generated in this study using AS-FD-OCT with those found in images generated by the same device in future studies. 

In our study, SC was observable, at least partially, in approximately 90% of cases in which living human subjects were examined with the HD mode, whereas the SC could be observed in none of the images generated with the bi-angle mode. Several explanations for this apparent discrepancy can be proposed. The first possible explanation is that in the bi-angle mode, SC was located far from the zero point, the reference position in the depth direction on a tomographic image, while, in the HD mode SC was located very near the 0 point. Image sensitivity is generally higher, as the measured object is nearer the 0 point in the OCT system. These different imaging conditions may result in poorer resolution at the angle imaged by the bi-angle mode compared with the HD mode. Another explanation may be that the line or space of SC was positioned perpendicular to the light of the OCT system in the HD mode, whereas SC was positioned with an inclination of approximately 30° to the light. This may have resulted in further aggravation of the separation of SC from the surrounding tissues. Moreover, in the HD mode, three consecutive images were averaged to reduce speckle noise. This procedure may have improved the image resolution. 

In this study, we measured SC length, TM length, and TM area in vivo. Generally, SC is reported to be approximately 200 to 400 μm long in its meridional axis and 10 to 25 μm in its shorter axis. ²⁰ The present study suggested that average meridional SC length was approximately 350 μm, which is compatible with the previous description. ²⁰ Otherwise, to the best of our knowledge, this is the first report of the measurement of TM area and length in living human subjects. According to a previous histopathologic study performed in donor eyes, the dimensions of SC in glaucomatous eyes were significantly smaller than in normal eyes. ²² Furthermore, the location of SC and TM in relation to the iris is essential in angle-closure glaucoma and related conditions. Therefore, the identification and measurement of angle parameters obtained from AS-OCT may have a significant impact on angle assessment in vivo because angle structures may be altered according to different glaucoma phenotypes, pharmacologic agents, and surgery. 

In the present study, scleral spur was observed in all (100%) the eyes; however, Schwalbe's line was identifiable in only 38 (31.7%) of the eyes; both structures were detected by Cirrus OCT (Carl Zeiss Meditec, Dublin, CA) according to the paper by Wong et al. ²³ A possible explanation of this difference should be that the present AS-FD-OCT had limited spatial resolution due to relatively longer wavelength of light source compared to the Cirrus OCT thought the shorter wavelength was beneficial for the deeper penetration. The difference in the scanning protocol, radial versus raster scan, may be another explanation. 

In conclusion, the HD imaging made possible by AS-FD-OCT could allow noninvasive real-time imaging of angle structures with a high detection rate of SC and measurement of SC and TM. Because information on detailed angle structures can be considered essential in attempts to elucidate the mechanisms of increased intraocular pressure in various types of glaucoma, this new device may assist in these investigations. Further studies using the AS-FD-OCT are necessary to further understand the alterations of SC and TM that occur in response to pathologic conditions, pharmacologic agents, or surgery. 

Footnotes

 Supported by grants-in-aid for scientific research from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Footnotes

 Disclosure: T. Usui, None; A. Tomidokoro, None; K. Mishima, None; N. Mataki, None; C. Mayama, None; N. Honda, None; S. Amano, None; M. Araie, None