Schematic diagram of the handheld probe used for imaging of the anterior segment. The optical coherence tomographic system allows for real-time imaging (4-16 frames per second) with a pixel resolution of 8 µm (depth) × 15 µm (lateral) over a field of view of 4.25 mm of free-space depth (3.08 mm in tissue) by 5 mm of lateral range.
The use of real-time optical coherence tomography (OCT) for anterior segment imaging. The OCT probe was held approximately 1 cm from the subject's eye, and was positioned interactively while viewing the image on the computer screen. Computer "hot-keys" provided capabilities for image freezing, archiving, averaging, and color/gray-scale selection.
Real-time optical coherence tomographic image of the cornea. The corneal epithelial and stromal layers are delineated; their thicknesses were directly measured from the image data.
Real-time optical coherence tomographic image of the angle region. The root of the iris, the angle recess, the ciliary body (CB), scleral spur (SS), and the trabecular meshwork (TM) are observed. The strongest reflections in this image of a darkly pigmented iris are from the anterior limiting layer.
Imaging of the pupillary light reflex. Following a sudden light stimulus, constriction of the iris sphincter was observed in real time as a stretching of the iris profile as well as a thickening of iris tissue in the pupillary region. The strongest reflections in this image of a lightly pigmented iris are from the posteriorly placed iris pigment epithelium.
Real-time optical coherence tomographic image of the pupillary region. The anterior capsule of the lens is clearly seen. The pupillary ruff, which is formed by the iris pigment epithelium turning anteriorly at the pupillary margin, can be well discerned.
Real-time optical coherence tomographic image of the angle region of a darkly pigmented eye. The corneal epithelium and stroma, sclera, iris, and ciliary body are shown. The anterior chamber angle is clearly seen and readily quantified. The ciliary body is visible although its outline is not well defined because of the overlying highly scattering sclera.
Real-time optical coherence tomographic image of the region just posterior to the limbus. The ciliary muscle and the ciliary epithelium are defined. Structures superficial to the sclera, such as the episcleral blood vessels and the rectus muscle with its insertion into the sclera, can also be seen.
Radhakrishnan S, Rollins AM, Roth JE, Yazdanfar S, Westphal V, Bardenstein DS, Izatt JA. Real-Time Optical Coherence Tomography of the Anterior Segment at 1310 nm. Arch Ophthalmol. 2001;119(8):1179–1185. doi:10.1001/archopht.119.8.1179
Recent advances in high-speed scanning technology have enabled a new generation of optical coherence tomographic (OCT) systems to perform imaging at video rate. Here, a handheld OCT probe capable of imaging the anterior segment of the eye at high frame rates is demonstrated for the first time.
To demonstrate real-time OCT imaging of anterior segment structures.
Survey of anterior segment structures in normal human subjects.
Main Outcome Measures
Achieving real-time imaging of the anterior segment, satisfactory image quality, and convenience of a handheld probe.
Optical coherence tomographic imaging of the anterior segment of the eyes of human subjects was performed using 1310-nm wavelength light with an image rate of 8 frames per second. Imaging trials demonstrated clear resolution of corneal epithelium and stroma, sclerocorneal junction, sclera, iris pigment epithelium and stroma, and anterior lens capsule. The anterior chamber angle was clearly visualized. Limited imaging of the ciliary body was performed. Real-time imaging of pupillary constriction in response to light stimulus was also performed.
High-speed OCT at 1310-nm wavelength is a potentially useful technique for noninvasive assessment of anterior segment structures.
Our results suggest that real-time OCT has potential applications in glaucoma evaluation and refractive surgery.
THERE ARE many situations in clinical ophthalmology that require precise understanding of the spatial relationships and dimensions of various structures in the anterior segment of the eye. A technique that is capable of anterior segment imaging with micrometer scale resolution would be valuable in obtaining such information, enabling superior delineation of anterior ocular morphology and highly precise biometry. In clinical situations such as refractive corneal surgery and glaucoma evaluation it would be desirable if high-resolution images of dynamic events could be obtained and displayed as they occur in real time.
Several techniques have been reported for high-resolution noninvasive or marginally invasive examination of the anterior eye. Ultrasound biomicroscopy1,2 provides resolution ranging from 20 to 60 µm with a depth of penetration of about 4 mm and can obtain images of structures concealed by opaque media. Imaging time is 8 frames per second, allowing in vivo observation of movements of ocular structures in real time. However, this technique requires immersion of the eye in a water bath solution and it is difficult to determine the exact location of the examined area. Confocal microscopy3,4 has been used for high-resolution imaging in transparent tissues of the anterior segment and provides en face images of the corneal layers. Video-rate imaging has been achieved with scanning slit confocal microscopy,5 which has a lateral resolution of 0.8 µm and scans optical sections that are 10 µm in thickness (with a ×50 objective lens). The technique requires the use of an index matching gel to optically couple the tip of the microscope objective to the cornea. Also, micrometer scale resolution measurements using confocal microscopy are limited to the cornea. Optical interferometric techniques6- 8 have been described for ocular biometry, and continuous corneal pachymetry9 has been demonstrated with optical low-coherence reflectometry. Image information obtained from these techniques is, however, limited to one dimension. Magnetic resonance imaging using special surface receiver coils has achieved an in vivo intraocular imaging resolution of 230 µm in human eyes,10 with an acquisition time of 6.5 minutes per sequence. Recently, cryogenically cooled surface coils have been used in anesthetized rabbits to obtain an intraocular resolution of 117 µm.11 To obtain magnetic resonance images of good quality, patients have to maintain eye and head positions for the duration of imaging, and chemical-shift artifacts have to be considered before interpretation of the images. This technique may be useful for elucidation of physiologic or pathologic disease mechanisms; however, it is impractical in many clinical situations.
Optical coherence tomography12 (OCT) is a high-resolution imaging modality that can overcome many of the limitations of current techniques used to image the anterior eye. It is a completely noninvasive technique that uses low-coherence interferometry to provide in vivo cross-sectional images of tissue structure with a spatial resolution of 10 to 20 µm. Optical coherence tomography has been predominantly used for imaging of the posterior segment of the eye.13- 15 Anterior segment imaging using OCT was first demonstrated in 199416 using light with a wavelength of 830 nm. In recent studies OCT has been used to evaluate anatomic outcomes of refractive surgery17 and has also been implemented as a slitlamp-adapted system for routine clinical examination of the anterior segment.18 Other studies of the anterior segment include comparison of in vivo OCT imaging of cataracts with histopathologic findings19 and the use of OCT for in situ monitoring of laser interactions in biological tissue.20 Recently, transcleral OCT with 1310-nm wavelength light21 has been described. However, all of these OCT systems have been relatively slow, with acquisition times of 1 to 5 seconds per image. A system capable of imaging in real time would reduce misalignment and patient motion artifacts and enable imaging of dynamic ocular events. The requirements for real-time imaging with OCT are that scans must be performed, processed, acquired, and displayed quickly and the optical source power must be increased with the frame rate so that the signal-to-noise ratio is maintained. High-speed OCT at 4 to 32 frames per second has already been achieved.22,23 Near real-time OCT systems for endoscopic imaging of the gastrointestinal tract are currently in clinical trials.24,25 We have developed a high-speed (4-16 frames per second) OCT system coupled to a handheld probe, suitable for ophthalmic examination. To image the anterior segment at a high frame rate we used high-speed Fourier domain optical depth scanning technology and an efficient interferometer design described previously.22,26 To enable higher optical power incident on the eye, we used a semiconductor optical amplifier light source operating at 1310-nm wavelength. In this wavelength region, the transmittance of ocular media is reduced27 such that higher-power illumination may be used without danger to the eye.28 The added advantage of using illumination at this longer wavelength is that the amount of scattering in ocular tissue is less than at 830 nm. The reason for this is that absorption and scattering in most tissue constituents is a decreasing function of wavelength in the near infrared spectrum29 whereas absorption in water (the primary constituent of vitreous humor) increases sharply, being approximately an order of magnitude higher at 1300 nm than at 830 nm. Thus, using 1310-nm rather than 830-nm illumination for anterior segment OCT allows for increased penetration in scattering tissues, such as the sclera and iris, while simultaneously permitting sufficient illumination power to be used to enable high-speed imaging. In this article we demonstrate the application of high-speed OCT to perform real-time cross-sectional imaging of the anterior eye in vivo for the first time to our knowledge. Real-time visualization of the anterior eye was achieved, with better morphological detail than previously possible.
Our study was approved by the institutional review board of the University Hospitals of Cleveland. All subjects underwent scanning after giving informed consent. Five healthy volunteers with clear ocular media and uncorrected or best-corrected visual acuity of 20/20 were examined. One subject wore soft contact lenses.
The real-time OCT system used in this study was similar to a system we have previously reported for high-speed endoscopic imaging in the gastrointestinal tract.22,25,26 The system employed a semiconductor optical amplifier light source capable of emitting 22 mW of low-coherence light with a central wavelength of 1310 nm and a spectral bandwidth of 68 nm full width at half maximum. The source had a free-space coherence length of 11.1 µm; thus the imaging depth resolution was 8.1 µm (following division by an assumed tissue refractive index of 1.38).30 High-speed scanning was achieved using a rapid scan Fourier-domain delay line, which generated longitudinal depth reflectivity scans (A scans) at a rate of 4 kHz. These A scans were subdivided into B scans according to the desired image rate. The system was further optimized for high-speed operation by employing a novel power-conserving interferometer design using an optical circulator26 and dual-balanced detection for excess noise reduction.25 Image data were displayed in real time on a computer monitor and simultaneously archived to high-quality (S-VHS) videotape for later review. Selected image frames could be frozen by the operator and saved in digital format. Computer"hot-keys" allowed for selection of gray or false-color display scale, frame averaging, image freezing, and saving of frozen images.
A novel aspect of the OCT system developed for anterior segment imaging is the use of a handheld OCT probe (Figure 1). The probe employs a miniature computer-controlled galvanometer scanner to provide a single axis of lateral scanning of the 1310-nm light on the subject. The probe optics transfers a magnified image of the sample arm fiber tip into the anterior segment while simultaneously imaging the galvanometer mirror plane into the plane of the objective lens to avoid vignetting during galvanometer mirror motion. The probe provides lateral scanning of up to 5 mm at imaging rates adjustable between 4 and 16 frames per second, and has a focal spot size of 15 µm full width at half maximum.
Figure 2 illustrates the use of the handheld probe for anterior segment imaging. The examination was done with the subject in the sitting position. The OCT probe was held with both hands as shown in the figure, with the examiner's fingers resting on the subject's face for support. Alternatively, in some cases, the subjects held the probe themselves. The distance between the probe and the eye being examined (the working distance) was 10 mm. Scanning was done by placing the probe close to the area of interest and observing the resulting image on the screen. Different structures in the anterior segment were viewed by either moving the probe itself or by asking the subject to move his or her eyes. Because of the fast acquisition rate of real-time OCT, involuntary changes in eye position did not cause deterioration of image quality. Thus, image processing for removal of A-scan artifacts (as performed in a previous study of low-speed anterior segment OCT16) was not required. Subjects were allowed to blink during the examination.
The optical power used in our 1310-nm OCT system was in accordance with the American National Standards Institute maximum permissible exposure of 15.4 mW for intrabeam viewing through a 7-mm pupil for exposure times of up to 8 hours.28 This is substantially higher than the maximum permissible exposure of 700 µW at 830 nm, the wavelength used in previous OCT systems. The maximum permissible exposure is much higher at 1310 nm because transmittance of the ocular media is less by approximately an order of magnitude at this wavelength than it is at 830 nm.27
Healthy subjects were examined with the high-speed OCT system using a handheld probe. Each examination took approximately 10 minutes. All images presented are single digitized frames acquired from continuous real-time output at 8 frames per second, each originally 520 (vertical) × 312 (horizontal) image pixels. The digitized images were smoothed using a 3 × 3 blur filter prior to final printing. The use of 1310-nm illumination provided deeper penetration in highly scattering tissue such as the sclera, yielding images with better morphological detail than those obtained with the conventional 830-nm OCT system.
Figure 3 shows an in vivo OCT image of the cornea. Two different zones can be visualized. The posterior zone measures 433 µm in thickness and includes the corneal stroma, the Descemet membrane, and the endothelium, which could not be individually resolved. The anterior zone, consisting of a surface interface reflection and a hyporeflective(dark) region measures 55 µm in thickness and represents the corneal epithelial structures. Because of the corneal curvature, the OCT beam is not perpendicular to the corneal surface over the whole scanning range, leading to refraction of the scanning beam at the air-cornea interface. Corneal measurements are therefore accurate only at points where the incident beam is perpendicular to the corneal surface. At all other points, accurate measurements can be made only after correcting for refraction. Development of an algorithm for refractive correction in anterior segment OCT images is currently underway. In Figure 3, the corneal thickness values were obtained at the perpendicular point assuming a corneal refractive index of 1.38.30Figure 4 shows an OCT image of the angle region and part of the cornea of a subject wearing a soft contact lens. The anterior surface of the contact lens is visualized as a highly reflective line. The posterior surface of the contact lens and the tear film under the soft contact lens are not separately resolved.
With high-speed OCT, the different layers of the iris that can be delineated include the iris pigment epithelium, the iris stroma, and the anterior limiting layer. The reflectivity of these layers differs according to the amount of pigmentation in the eye. Figure 5shows OCT images of a lightly pigmented eye. The largest reflected signals are from the posteriorly placed pigment epithelium; the anterior limiting layer is not clearly visualized. In a darkly pigmented eye (Figure 4), however, the largest reflected signals are from the anterior limiting layer of the iris. This is consistent with the larger amount of scattering pigment present in the anterior limiting layer and stroma of darker irides. Additional morphological details visible by OCT imaging include the iris crypt(Figure 4), the iris sphincter (Figure 5), and the pupillary ruff (Figure 6), which is formed by the iris pigment epithelium turning anteriorly at the pupillary margin.
Real-time OCT enables imaging of dynamic events such as pupillary constriction. Figure 5 demonstrates the contractile response of the iris to light (the pupillary light reflex). Following a sudden light stimulus (shining a flashlight into the eye), the iris sphincter constricted and this was observed in real time as a stretching of the iris profile. High-speed OCT could also be useful in real-time evaluation of the change in peripheral iris configuration in response to darkness (the dark-room provocative test for assessing angle occludability).31
The ciliary body is visualized with better detail than in previous anterior segment OCT imaging studies. This is because of the deeper penetration of 1310-nm wavelength relative to 830-nm wavelength used in previous ophthalmic OCT systems. In Figure 7 the pars plicata of the ciliary body is seen as a hyporeflective structure lying just beneath the sclera. The pars plicata, which contains the ciliary processes, is the region of maximum radial thickness of the ciliary body. Since attenuation of light by the sclera leads to decreased reflectivity with increasing depth, the outline of this portion of the ciliary body is poorly delineated. However, the part of the ciliary body that forms the angle of the anterior chamber is clearly visible (Figure 4).
Figure 8 shows the posterior portion of the ciliary body. The triangular ciliary muscle is seen on the right-hand side of the image and tapers posteriorly into the pars plana. In contrast to the pars plicata, the pars plana is thin, which allows the inner boundary formed by the ciliary epithelium to be clearly discerned. The ciliary epithelium is continuous anteriorly with the iris pigment epithelium and posteriorly with the retinal pigment epithelium.
The sclera in OCT images appears as a highly reflective structure. The sclerocorneal junction or the limbus is clearly outlined by the difference in reflectivity between the highly scattering sclera and the weakly scattering cornea (Figure 4 and Figure 7). The scleral spur, which is an important landmark in determining angle configuration, is identified as the pointed inward projection from the inner surface of the sclera. The scleral spur also gives attachment to the ciliary body. Figure 8 is an image of the region just posterior to the corneoscleral limbus. Superficial structures such as the episcleral vessels and the rectus muscle insertion into the sclera are clearly visualized. However, the episclera and sclera are not differentiated and appear as a single highly reflective complex.
Full-thickness visualization of the angle structures was achieved with real-time OCT. Figure 7 is an OCT image of the angle region of a darkly pigmented eye. The angle of the anterior chamber is clearly defined throughout its radial extent. Figure 4 is an image of another darkly pigmented eye and illustrates further details of the angle region. Structures that can be discerned are the root of the iris, the angle recess, the anterior surface of the ciliary body, the scleral spur, and the trabecular meshwork. The angle region is better delineated in darkly pigmented eyes probably because of the higher amount of scattering pigment. Although the canal of Schlemm is not visible in this image, it was visualized during real-time imaging. Measurement of the angle of the anterior chamber was made directly from the image in Figure 7.
Figure 6 shows an OCT image of the pupillary region. The anterior lens capsule is imaged within the pupil as a highly reflective line. The retroiridial portion of the lens capsule was not seen because of the highly backscattering iris pigment epithelium. Although the lens cortex is not visible in this image, it could be seen sometimes in the pupillary area. The retroiridial position of the lens cortex as well as the zonular fibers were, however, not resolved.
The OCT systems used in research and commercial ophthalmic applications thus far have used 830-nm wavelength, with image acquisition times of 1 to 5 seconds. Very recently, ophthalmic OCT at 1310-nm wavelength has been described, with an acquisition time of 3.3 seconds. All of these systems require image-processing techniques to remove artifacts caused by patient motion during data acquisition.32 A system capable of faster data acquisition would not be affected by involuntary eye movement and would allow real-time display. To achieve real-time imaging with OCT, the requirements are that images must be acquired rapidly and any increase in the rate of image acquisition must be accompanied by a proportional increase in source optical power so that the signal-to-noise ratio is maintained. We have achieved an image acquisition rate of 8 frames per second by using high-speed Fourier domain optical depth scanning technology. To obtain higher optical power, we used a semiconductor optical amplifier light source operating at a wavelength of 1310 nm. In addition to enabling the use of higher power without damage to the eye resulting, this wavelength also has the advantage of increased penetration in scattering tissue such as the sclera and iris.
The high-speed system was coupled to a handheld probe suitable for examination of the anterior segment of the eye. For steadier handling, the probe was held with both hands with the fingers resting on the subject's face for support. A disadvantage to this method is that the examiner's hands are not free for other maneuvers—a problem that could be overcome by incorporating the real-time scanning system into a conventional slitlamp. However, the handheld probe allows convenient scanning in any position and would be advantageous in situations in which positioning at the slitlamp is difficult, such as in the pediatric age group and in patients whose general condition is poor.
In contrast to other high-speed micrometer resolution imaging techniques such as ultrasound biomicroscopy and confocal microscopy, real-time OCT is completely noninvasive and does not require contact with the eye or immersion of the eye in a water bath. The depth resolution of 8.1 µm provided by real-time OCT approaches that obtained by confocal microscopy, with the additional benefit given to axial resolution by coherence properties of the source, and does not depend on the available numerical aperture or the quality of the beam focus. The cross-sectional imaging capability of real-time OCT is similar to ultrasound biomicroscopy, with the added advantage of better spatial resolution. However, OCT cannot obtain images through opaque media and it provides limited penetration of the ciliary body. In addition, delineation of angle structures with OCT is poorer in lightly pigmented eyes.
The imaging capability of the high-speed OCT system coupled with the real-time display allows several potential applications in the anterior segment, especially in evaluation of the anterior chamber angle and the cornea. Clinical examination of the anterior chamber angle is routinely performed by direct visualization by gonioscopy, a technique that provides limited information about structures behind the iris. High-speed OCT provides in vivo cross-sectional images of the anterior eye, similar to histological sections, and could be a potential tool for noninvasive evaluation of the anterior chamber angle. The images shown in Figure 4 and Figure 7 are comparable to those obtained by ultrasound biomicroscopy, a technique that has been extensively used for elucidation of the etiopathogenesis of various types of glaucoma.33- 35 The micrometer resolution imaging capability of OCT would be useful in evaluating the structural causes of angle-closure glaucoma syndromes such as plateau iris syndrome, malignant glaucoma, and pupillary block glaucoma. An important part of the glaucoma workup is to study the alterations in anatomical configuration of angle structures in response to light and accommodation. These structural alterations effect a rise in intraocular pressure in conditions such as plateau iris syndrome, pigmentary glaucoma, and primary angle-closure glaucoma. Real-time OCT imaging of the angle could be especially useful for assessment of these conditions, including performing provocative tests for assessing angle occludability, such as the dark-room and the prone provocative tests.36 Another application of OCT in the imaging of the angle would be in cases of anterior segment trauma. Since it is noninvasive, OCT can be used safely in situations in which ocular tissue has been lacerated or punctured—conditions under which gonioscopy cannot be performed because of the danger of aqueous leakage. Optical coherence tomography would also be useful in angle assessment in the pediatric age group.
Real-time imaging of the corneal layers with high-speed OCT could be applied in the field of keratorefractive surgery, especially in laser in situ keratomileusis (LASIK).37 Corneal flap thickness is an important parameter in LASIK because it determines the amount of residual stroma available for ablation. Currently there is no technique that directly measures corneal flap thickness and it has been demonstrated that there are noteworthy discrepancies between intended and actual flap thickness values.17,38 A technique that can directly measure corneal flap thickness intraoperatively could potentially improve the predictability of LASIK. Real-time OCT is ideally suited for this purpose. For intraoperative use of our high-speed OCT system it would be necessary to increase the lateral field of view by increasing the limiting aperture of the probe. Another application of OCT in keratorefractive surgery would be to determine the corneal ablation rate. Real-time cross-sectional imaging of the corneal microstructures can allow continuous monitoring of the ablation process. Variations in corneal ablation rates caused by factors such as stromal hydration39 may be responsible for the differences observed between planned and actual ablation depths in LASIK17,38 and ablation rates could potentially be studied with high-speed OCT. With its superior resolution of morphological detail, this technique can also be used for postoperative assessment of the anatomical correlates of the refractive outcome.
High-speed OCT at 1310-nm wavelength demonstrates better morphological detail of nontransparent ocular structures than OCT at 830-nm wavelength. This makes it a useful modality for the assessment of tumors and cysts of the iris and the ciliary body. Optical coherence tomography would be helpful in accurate localization of tumors of the iris and ciliary body, measurement of tumor size, and evaluation of factors such as depth of penetration and extrascleral extension. It would also be useful in assessing peripheral iris pigment epithelial cysts (iridociliary cysts), which are difficult to visualize by conventional techniques such as slitlamp biomicroscopy and gonioscopy.
In conclusion, we have demonstrated real-time imaging of the anterior eye using high-speed OCT. Fast data acquisition allowed real-time display of high-quality images in which delineation of the corneal layers and full-thickness visualization of angle structures was possible. Measurements of corneal epithelial and stromal thickness and of the anterior chamber angle were made. These preliminary results suggest the potential use of real-time OCT in the clinical assessment of the anterior segment, especially as an adjunct in glaucoma evaluation and for intraoperative monitoring of corneal changes during keratorefractive surgery.
Accepted for publication February 1, 2001.
This research was supported by grant R24 EY 13015-01 from the National Institutes of Health, Bethesda, Md (Drs Izatt and Radhakrishnan) and by Research to Prevent Blindness Inc, New York, NY (Dr Bardenstein).
We acknowledge the technical help of Brian A. Wolf.
Corresponding author and reprints: Joseph A. Izatt, University Hospitals of Cleveland, Division of Gastroenterology, 11100 Euclid Ave, Cleveland, OH 44106-5066 (e-mail: firstname.lastname@example.org).