Histologic Correlation of Pig Retina Radial Stratification with Ultrahigh-Resolution Optical Coherence Tomography

Abstract

purpose. To compare ultrahigh-resolution optical coherence tomography (OCT) cross-sectional images of the pig retina with histology, to evaluate the potential of ultrahigh-resolution OCT for enhanced visualization of intra- and subretinal structures.

methods. Ultrahigh-resolution OCT images were acquired with 1.4-μm axial × 3-μm transverse resolution from in vitro posterior eyecup preparations of the domestic pig. Frozen sections were obtained in precise alignment with OCT tomograms, by using major blood vessels as orientation markers and were counterstained with cresyl violet or unstained and examined by differential interference contrast microscopy. Micrographs from histologic sections were linearly scaled to correct for tissue shrinkage and compared with OCT tomograms.

results. In the proximal retina, ultrahigh-resolution OCT signal bands directly corresponded to the main retinal layers. For the wavelength region used (∼800 nm), axodendritic layers (nerve fiber layer, inner and outer plexiform layers) were more reflective than cell body layers (ganglion cell layer, inner nuclear layer, outer nuclear layer). In the distal retina, substructures of the photoreceptor layer such as the interface between inner and outer segments were visualized, and the retinal pigment epithelium, the choriocapillaris, and superficial choroid layers were resolved. In addition, the time sequence of a retinal detachment event was monitored by ultrahigh-resolution OCT.

conclusions. In vitro ophthalmic ultrahigh-resolution OCT imaging reveals retinal morphology with unprecedented detail. The specific assignment of OCT signal patterns to retinal substructures provides a basis for improved interpretation of in vivo ophthalmic OCT tomograms of high clinical relevance.

Optical coherence tomography (OCT) is the optical analog of ultrasonic pulse-echo imaging and provides a noninvasive technique that enables high-resolution, cross-sectional in vivo imaging of both transparent and nontransparent biological tissue. ^{1 2 3} Because the eye is essentially transparent and transmits light with only minimal attenuation and scatter, OCT is particularly attractive for ophthalmic imaging, providing easy access to both the anterior segment and the retina. OCT was first evaluated in ophthalmology, ⁴ and consequently ophthalmic diagnosis represents one of the most developed clinical OCT applications. Several studies using either animal (Huang LN, Schuman JS, Pedut-Kloizman T, et al., ARVO Abstract 3913, 1997). ^{1 5 6 7 8 9 10 11} or human cadaveric ⁷ retinas have been conducted to compare OCT tomograms with histology to gain a better understanding and interpretation of OCT images. However, the exact correlation of OCT tomograms with retinal morphology has been a long-standing controversy. ⁷ One of the limiting factors is insufficient axial resolution of approximately 10 to 15 μm, as given by the bandwidth of the used light source, usually a superluminescent diode, and therefore reduced visualization of intraretinal layers. Recently, the emergence of ultrabroad-bandwidth femtosecond laser technology has aided in development of a new generation of OCT technology. Ultrahigh-resolution OCT has been demonstrated to achieve unprecedented axial resolution of approximately 1 μm in nontransparent media ¹² and 3 μm in imaging of the human retina in vivo, ^{13 14} which is two orders of magnitude higher than can be achieved by conventional ultrasound imaging. 

The goal of the present study was to compare cross-sectional images of the retina obtained by ultrahigh-resolution OCT with those obtained by light microscopy, to identify the morphology visualized in OCT tomograms. We demonstrated in the current study that ultrahigh-resolution OCT can be a powerful tool for ophthalmic diagnosis by enabling optical biopsy of the retina—that is, the visualization of retinal microarchitecture, which previously has been possible only with histopathology. 

Materials and Methods

Because human retinal tissue quickly loses histological and optical quality after death, pig tissue was chosen for the present study. The pig eye is increasingly used as a model in vision studies and is readily available for experimentation. Besides its size, several features are similar to those of the human eye. The domestic pig has two areas lateral to the main medial ascending vessels, as evidenced by density peaks of ganglion cells, ^{15 16} cones, ¹⁷ and spectral cone types. ^{18 19} Cone densities remain high in peripheral retina, indicating good overall adaptation to diurnal activities. Pig cone inner segments reveal paraboloid morphology similar to human perifoveal and peripheral cone inner segments, thus allowing inference of similar optical properties. Finally, the pig retina is holangiotic. The prominent vascular architecture provides suitable landmarks for precise alignment of ultrahigh-resolution OCT images with histologic sections. 

Ocular Tissue

All animals were handled in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Eyes were enucleated from domestic pigs that were used for unrelated experiments, either shortly before or immediately after death. After immersion in cooled oxygenated 0.01 M phosphate-buffered saline (PBS, pH 7.4), eyes were transferred to the laboratory within 30 minutes, and the cornea, lens, and vitreous were removed. The reason for removing the anterior eye segment and the vitreous was to obtain high transverse resolution that would be limited by ocular aberration in case of using the intact eye. From the eyecup, a horizontal strip including the optic disc was prepared, the remaining vitreous removed, and the fundus digitally photographed before transferring the sample to the ultrahigh-resolution OCT setup. 

Ultrahigh-Resolution OCT

A new generation of a compact ultrahigh-resolution OCT system was developed and used in the present study. The system consists of a high-speed scanning unit (up to 250 Hz, 400 mm/s) integrated in a fiber optic-based Michelson interferometer using a compact, user-friendly, state-of-the-art sub-10-femtosecond titanium:sapphire laser (800-nm center wavelength, up to 170-nm [full width at half maximum; FWHM], optical bandwidth, 400-mW output power; Compact Pro; Femtosource, Vienna, Austria). The interferometer was interfaced to a microscope delivery system. Both the fiber-optic interferometer and the optical components of the microscope were designed to support the propagation of very broad bandwidth light throughout the OCT system and to compensate for any polarization and dispersion mismatch between the sample and reference arms of the interferometer. ¹² To achieve high transverse resolution, a specially designed achromatic objective with 10 mm focal length and a numerical aperture of 0.30 was used, achieving 3-μm free-space transverse resolution, resulting in a confocal parameter of approximately 40 to 60 μm in air, degrading the OCT image outside the focused zone. To overcome the depth-of-field limitation and to maintain high transverse resolution at various depths through the image, a zone focus and image fusion technique was used. ¹² Separate images with different focal depths of the optics were recorded, while maintaining the same interferometer delay depth (2 mm). These tomograms were then fused together. This technique is similar to C-mode scanning used in ultrasound imaging. ²⁰ Up to 80- to 100-μm imaging depth was obtained without significant image degradation. This image fusion technique would not be necessary in case of in vivo ultrahigh-resolution OCT imaging. Due to ocular aberration the best transverse resolution possible in the living human retina is limited to 10 to 15 μm, resulting in a more than 500-μm depth of focus to cover the whole retinal thickness. Special single-mode fibers (570-nm cutoff wavelength) and special broad band, wavelength-flattened, 3-dB fiber couplers were used to maintain ultrabroad bandwidth and single-mode propagation. Applying laser light centered at 800 nm with up to a 170-nm bandwidth (FWHM), axial resolution of 2.0 μm in air, corresponding to 1.4 μm in biological tissue, was achieved with this system. A signal-to-noise ratio of 105 dB was achieved at 1 MHz carrier frequency by using an incident power of 5-mW, using dual-balanced detection. Although applied in ex vivo tissue, retinal exposure must be taken into account in studies using the ultrabroad-bandwidth light generated by a titanium:sapphire laser. The American National Standards Institute (ANSI) standards for retinal exposure account for wavelength, exposure duration, and multiple exposures of the same spot of the retina. Because the laser source generates femtosecond pulses, the laser output was coupled into a 100-m-long optic fiber that was used to provide dispersive stretching of the pulse duration to hundreds of picoseconds. This reduces the peak pulse intensities by several orders of magnitude and, because the laser operates at an 80 MHz repetition rate, the output can be treated as a continuous wave. Persistent illumination of the retina with laser light centered around 800 nm with 500 μW is allowed for only 20 seconds. Therefore the microscope OCT system has been designed to avoid direct illumination of the focused beam into the eye. Full interference fringe signal OCT data were digitized with a high-speed (10 megasamples/s) and high-resolution (16-bit) analog-to-digital (A/D) converter followed by software demodulation. 

During OCT imaging, real-time imaging display enabled simultaneous, immediate cross-sectional visualization of the imaged area. Using a scanning frequency of up to 130 Hz resulted in a measurement time of approximately 16 seconds for an OCT tomogram consisting of 2000 A-scans. Except for image fusion, no other technique was used to generate OCT tomograms. Position, orientation, and length of OCT scanned cross sections were recorded on the digital fundus micrographs and used to achieve matching orientation of specimens in subsequent histologic sectioning. 

Histologic Preparation

OCT-imaged tissue was fixed in 4% paraformaldehyde and 0.1 M PBS (pH 7.4) for 12 to 72 hours, cryoprotected in an ascending series of sucrose and PBS, infiltrated in 20% sucrose and PBS (two parts) and optimal cutting temperature medium (one part; Miles, Elkhart, IN) for 18 hours, flash frozen, and mounted to receive vertical sections matching the plane of the OCT scans. Series of 12-μm frozen sections were collected onto chrome-alum gelatin-coated slides, air dried, and stained with cresyl violet or left unstained for differential interference contrast (DIC) microscopy. 

Data Analysis

Sections matching the position of OCT scans were identified by using blood vessels as landmarks and photographed with a camera (NU-200 CCD; Photometrics, Tuscon, AZ) attached to a microscope (Eclipse 600; Nikon, Tokyo, Japan). Image-analysis software (Photoshop 5.5; Adobe Systems, San Jose, CA) was used for linear scaling of micrographs to correct for volume changes of the tissue that occurred in the course of histologic processing, and the micrographs were overlaid on OCT images to evaluate their correlation. OCT image enhancement was applied when appropriate and included gradient subtraction and noise or median filtering (Image Processing Tool Kit, Reindeer Graphics Inc., Asheville, NC). Shadow extending from blood vessels and nerve fiber bundles was reduced by orientational fast Fourier transform (FFT) masking (Digital Micrograph; Gatan Inc., Warrendale, PA). To obtain averaged density profiles, pixel-line projections were performed along the radial axis of OCT images in Image SXM (http://reg.ssci.liv.ac.uk/; developed by Steve Barrett, University of Liverpool, Liverpool, UK, available in the public domain). 

Results

Correlation of OCT Images with Histology

Profiles of blood vessels and their relative positions were reliable landmarks for identifying matching histologic sections to an almost 1-μm precision. Ultrahigh-resolution OCT clearly delineated the inner retinal border with the attached major vascularization and resolved the different cross-sectional status of both arteries and veins (Figs. 1 2 4) . This precise positional and cytoarchitectural information of OCT images was used to fit micrographs obtained from matching histologic sections to their respective tomograms, by simply performing a linear scaling manipulation. As a result, we identified a close correspondence of at least eight bands of OCT signal with specific retinal layers (Fig. 1) . We found no significant evidence that volume changes occurring in the course of tissue processing differently affected the different retinal layers and thus interfered with our approach. Mismatches encountered when superimposing histologic and OCT images were found to be due to artifacts arising from tissue handling, such as different grades of retinal detachment and collapse of blood vessels (Figs. 1B 4E) , or to slight deviations of the sectioning plane from the plane of the scan. Further, due to the high resolution of structural detail in the deeper retina, positional noise did not complicate the straightforward correlation of OCT signal patterns with retinal structure. This was particularly evident in the retinal layers between the internal (ILM) and external (ELM) limiting membranes and in the retinal pigment epithelium (RPE) and choroid. 

OCT of Proximal Retinal Layers

Ultrahigh-resolution OCT identified the nerve fiber layer (NFL) by the strongest signal (Fig. 1) . In scans made perpendicular to the papillary radiation, individual axon bundles were easily distinguished between Müller cell columns (Figs. 2B 2C ; arrows). Distally, a band of lower signal precisely corresponded to the ganglion cell layer (Figs. 1 2 , GCL). Both the shape and thickness of the band changed with retinal eccentricity and were found to correspond strictly to the number of rows of ganglion cell somata present (Figs. 2B 2C) . The adjacent highly reflective band, at both its proximal and distal borders, precisely correlated with the location and extension of the inner plexiform layer (IPL), as revealed by either DIC or bright-field microscopy (Figs. 1 2) . We noted numerous patches of higher signal (dark spots) within this inhomogeneous band (Fig. 1A) but were unable to confirm their correspondence to profiles of intraretinal capillaries. Two further distinct bands of low and high signal, respectively, corresponded to the inner nuclear (INL) and outer plexiform (OPL) layers (Figs. 1 3) . 

OCT of Distal Retinal Layers

The adjacent band of lower signal largely corresponded to the outer nuclear layer (ONL). However, its outermost aspect was not unequivocally attributable to the position of the ELM (Figs. 1 3) . Densitometric profiling confirmed sublayering and discriminated a wider proximal portion of higher signal from a slender distal portion of lower signal (Fig. 3B , m). A further high signal band of speckled appearance was found in correlation with the location of the cone ellipsoids (Figs. 1 3) . Beyond the photoreceptor layer, a distinct band of high signal corresponded to the RPE (Figs. 1A 3 4) . Densitometric profiling (Fig. 3B) confirmed that the signal intensity was the most prominent in the outer retina. This was particularly evident in OCT images focused on the distal retina (Figs. 3A 3C) . Distal to the PE layer, a thin, light band corresponded to the position of the choriocapillary layer, and diffuse high signal interspersed with patches of low signal closely correlated with the pigmented choroidal stroma and the lumina of larger blood vessels (Figs. 3A 3C , arrows). 

Recording of Local Retinal Detachment

In a series of OCT scans, the initiation of a retinal detachment event was monitored within a time frame of 30 minutes (Fig. 4) . Ultrahigh-resolution OCT visualized a focal elevation of the neural retina concomitant with an increase in subretinal space (Fig. 4A , arrow). Fifteen minutes later, alterations were observed within the monolayered band of the PE signal. Although still continuous, the PE signal appeared triple layered at the initial locus of detachment, with a stripe of bright signal framed by two darker bands (Fig. 4B , arrowheads). After an additional 15 minutes, all retinal layers were observed to be bent inward, and measurements of their relative thicknesses (not shown) indicated increased thickness of the proximal retinal layers. In the PE signal, the bright inclusion had increased, whereas the innermost aspect of the signal appeared eroded (Figs. 4C 4D , arrow). Histologic examination of the matching retinal position demonstrated a significantly extended region of detachment (Fig. 4E) . The pigment epithelium was lesioned at the initial locus of detachment, and fragments of tissue were dislocated in the subretinal lumen (Fig. 4E , asterisk). 

Discussion

Ultrahigh-Resolution OCT and Optical Properties of Retinal Sublayers

The present study demonstrates that in vitro ultrahigh-resolution OCT imaging reveals retinal microarchitecture with unprecedented detail. By using histologic sections precisely matching the plane and retinal position of cross-sectional OCT scans for comparison, we chose an approach that enabled us to correlate the various components of the OCT image directly to histology. Despite the enhanced axial resolution, however, caution is imperative when assigning specific bands of OCT signal to specific retinal layers. OCT images are determined by the optical properties of the tissue. As a consequence, tissue components that give strong contrast in histologic staining may appear fairly inconspicuous in the tomogram. Sample characteristics that strongly modulate the signal include thickness, absorption, and refractive index. OCT signal arising from the different retinal sublayers results from the averaged refractive indices of both intracellular and extracellular components. Differences in histologic architecture lead to differences in scatter, reflection, or guidance of light. Absorption and reflection profiles similar to that obtained by light microscopy are evident in the hemoglobin of blood vessels (Figs. 1 2 3) and the melanin that is present in the RPE and choroid (Figs. 1 3 4) . The tight packing of lipid membranes in bundles of the ganglion cell axons, axodendritic layers, and subcellular components, such as ellipsoidal mitochondria, outer segment discs, microvilli, and calycal processes is associated with loci of higher refractive indices. Therefore, scatter phenomena differ between cell body layers and axodendritic layers and contribute to the different signal from these regions. 

Several factors further complicated the correlation of structural detail with OCT images. Inherent in optical imaging techniques is the degradation of signal contrast with distance from the focus level. We compensated for this partly by combining multiple scans acquired at different focal levels. Mismatches between histology and OCT images resulted from the limits of precision in selecting sections closest to the OCT plane from an entire series, possible deviations from perpendicular orientations of both OCT images and histologic sections, and volume changes in the tissue after scanning and during histologic processing. A particular problem is the frequent detachment along the PE/photoreceptor border with subsequent buckling of the retina. With the enhanced resolution of ultrahigh-resolution OCT, however, positional mismatches between histology and OCT signal provided a source of information to evaluate artifactual modification of the tissue in the course of histologic processing. In histologic preparations, epiretinal blood vessels were observed to protrude into the vitreous (Fig. 4E , V). They appeared smoothly embedded in the inner retinal surface in the OCT tomograms (Fig. 4D) . Choroidal lacunae were clearly visible in OCT scans (Figs. 3 4) , whereas in the course of tissue processing they frequently collapsed. This further helps to discriminate procedural artifacts from pathologic alterations. 

Assignment of Signal to the Inner Retinal Layers

Taking into account these limitations, our approach allows reliable correlation of all main retinal layers with specific bands of the OCT signal. Ultrahigh-resolution OCT imaging precisely distinguishes the NFL from the GCL. OCT scans performed perpendicular to the papillary radiation allow evaluation of the full cross-sectional status of nerve fiber bundles (Figs. 2B 2C) . We did not resolve individual ganglion cell somata, although the size of the large alpha ganglion cells is well within the range of resolution of the technique. Unequivocal confirmation of the representation of individual somata in an ultrahigh-resolution OCT tomogram requires correlation at the cytological level and was beyond the scope of this study. The inhomogeneous appearance of OCT signal corresponding to the IPL and the INL implies that with further refinements of image acquisition and processing, substructures within both layers (IPL sublayers, capillaries) will become identifiable. 

Assignment of Signal to the Photoreceptor-Pigment Epithelium-Choroid Complex

The detailed evaluation of the photoreceptor layer, the pigment epithelium, and the choroid is of prime interest for clinical diagnosis. Ultrahigh-resolution OCT distinguishes these layers. The most intense signal deriving from the outer retina can be safely correlated with the position of the pigment epithelium, thus providing a landmark for the delineation of the retina versus the choroidal layers. This is confirmed by our record of a progressive retinal detachment. As the neural retina increasingly detached, additional alterations affected the pigment epithelium signal, which split at the site of detachment with a brighter cleft of approximately 5 to 8 μm framed by two darker bands. Although the present data do not suffice to resolve the cytological equivalent of the cleft (i.e., whether the pigment epithelium is lesioned between villous processes and cell bodies or at Bruch’s membrane), the observation demonstrates the resolving power of the technique. 

To assign substructures within the photoreceptor layer at a micrometer scale is challenging. Artifactual alterations of fragile photoreceptor outer segments in the course of tissue processing interferes with direct alignment, and it is likely that disproportionate vertical representation due to refractive index variation further complicates the assignment of signal. Apparently, the complex structural subtiering of the photoreceptor layer is not reflected by an equivalent number of bands in the OCT tomogram. For example, in the myoid portion of the cone inner segments (IS) cresyl violet contrasts with a rich Nissl substance, the light microscopic equivalent of the endoplasmic reticulum at the electron microscope level (Fig. 3C) . ²¹ This zone appears to have no OCT correlate. Therefore, in an attempt to assign observed bands of OCT signal to specific components of the photoreceptors, we considered their optical properties. The ONL comprises two to four rows of rod somata and a monolayer of cone cell bodies, ²² located proximal to the ELM. The regular arrays of cone somata and the tight association of photoreceptor myoids and microvillous processes of Müller cells adjacent to the proximal and distal border of the ELM, respectively, may provide for enhanced transparency and similar refractive properties of the tissue and therefore be represented as the delicate light band discernible at the outermost aspect of the ONL (Figs. 1 3) . This interpretation is supported by DIC microscopy. Proximal and distal to the ELM, corresponding to the location of cone cell somata and the photoreceptor inner segment myoid portion, respectively, DIC microscopy identified two bands of high transparency but low relief (Fig. 1B) , indicating a region of similar transmissive and refractive properties. Evidently, the light signal band would then include the ELM, which seems to give no separate signal (Fig. 3B , elm). An alternative interpretation would correlate the borderline between high and low signal to the ELM. However, this would leave a very small radial representation of the ONL. We consider this unlikely for both optical and morphologic reasons. 

Using DIC optics, the ellipsoid portions of the cone inner segments were clearly distinguished by their high relief (Fig. 1B) . OCT imaging identified this zone as a high-signal band. In pig, cone inner segment ellipsoids comprise approximately 70% of the cross-sectional light-capture area. ²³ Their paraboloid morphology and their high refractive index may result in periodical alterations of reflectivity within this zone, which is consistent with the speckled appearance of the OCT signal. 

Photoreceptors attain specific shapes beyond their myoid portion (rod inner and outer segments [IS and OS]: cylindrical; cone OS: conical, cone IS ellipsoids: paraboloid). ²⁴ They are supposed to constitute individual optical elements, with refractive indices (n) of 1.4 or more separating them from the surrounding interphotoreceptor matrix (n = 1.34). Under physiological conditions, particularly the cone ellipsoids guide light toward the photopigments located in the OS. ²⁵ Further, photoreceptor OS consisting of densely stacked disc membranes have high refractive indices. ^{26 27} The transition zone to the OS appears to produce a reflective signal that is more prominent in the all-cone foveal photoreceptor layer. ¹⁴ Of interest, in macaque fovea, approximately 20 tapering calycal processes arise from the ellipsoid surrounding the cone outer segment base as a dense collar. ^{28 29} The processes are supposed to provide structural support or semioccluded periciliary compartmentation. These processes are less numerous around the base of the longer rod outer segments. It is possible that cone calycal processes contribute to the optical properties of the IS-OS transition zone. In pig and human retina, the photoreceptor layer is tiered with fat, short cone IS collecting major portions of the incoming light at the basal aperture. ³⁰ The cone OS are tapered and possibly distribute nonabsorbed light from their shorter tips to the surrounding rods. ^{23 24} Cone IS are shorter than rod IS, and their short, tapered OS lead to the positioning of cone OS tips just above the rod OS-IS transition. 

Together, these observations suggest that interactions of the OCT beam with the outer retina are less homogeneous than in the proximal retina. The interpretation proposed herein attempts a coarse correlation of photoreceptoral subtiers with the OCT image. Further studies are needed to clarify how the various subcellular components of the photoreceptors contribute to the OCT signal. We may expect that, beyond the ELM, OCT signal patterns are different in cone-prominent retinas (as in the current study), rod dominant retinas, and the fovea, with its homogeneous array of elongated cones. Evaluation of these differences may allow the interpretation of pathologic conditions, such as congenital or progressive cone dystrophies. 

Although the pig retina may well approximate the human peripheral and extrafoveal retina, we advise caution in directly transferring our layering assignment to the human fovea. There are no rods in the fovea. Both cone IS and OS have rodlike morphology. ²⁹ The refractive index of the elongated foveal outer segments has been estimated (n = 1.419) to be higher than in peripheral retina. ²⁴ Finally, foveal cones have been reported to exert a spatial-frequency filter effect ³¹ for coherent light with wavelengths from 410 to 654 nm (where two waveguide modes are carried) and a flat response for wavelengths greater than 654 nm (the single-mode region). Together, the specific morphologic and optical properties of the foveal receptors may significantly alter the banding pattern of the OCT signal, particularly at the level of their IS and OS, and our layering assignment may need specific adaptation for reliable interpretation of ultrahigh-resolution OCT images of the normal and pathologic fovea. 

In summary, this study demonstrates that ultrahigh-resolution OCT enables unprecedented visualization of retinal microarchitecture. A time-lapse sequence OCT recording of a progressive retinal detachment demonstrates the potential of the technique to monitor dynamic processes in the retina at sufficient resolution to track the fate of specific retinal layers. Preliminary in vivo results using a laboratory prototype laser system demonstrated recently that the quality and performance of in vivo retinal ultrahigh-resolution OCT images in normal subjects is comparable to the present results, as far as axial resolution and sensitivity are concerned ¹⁴ : axial resolution of 3 μm and a sensitivity of only 5 to 7 dB less than was achieved in the present ex vivo study. This axial resolution was a factor of two less, mainly due to the chromatic aberrations of ocular media, but was sufficient to visualize all main retinal layers similar to what can be achieved in histopathology and to what is presented in the present ex vivo study. Transverse OCT imaging resolution is much worse in studies in vivo, because of corneal aberrations that limit the best possible transverse resolution on the retina to 10 to 15 μm. Interfacing adaptive optics to the ultrahigh-resolution ophthalmic OCT system may improve transverse resolution. The lower sensitivity in the in vivo measurements is mainly due to the lower power (500–800 μW compared with 5 mW) necessary for safety reasons, but is sufficient to achieve comparable sensitivity and intraretinal layer visualization in vivo. 

In addition, a clinically viable ultrahigh-resolution ophthalmic OCT system has been developed recently and used in clinical imaging in 56 eyes of 40 selected patients with different macular diseases and has enabled unprecedented visualization of intraretinal morphology comparable to the results presented in the present ex vivo study. ³² The ultimate clinical availability of this ultrahigh-resolution OCT technology will depend on the availability of ultrabroad bandwidth light sources that are suitable for OCT applications. The principal disadvantage of current femtosecond laser technology is its extremely high cost. With continuing research, more compact and less expensive light sources for ultrahigh-resolution OCT imaging can be expected in the near future and therefore ultrahigh-resolution OCT may provide a powerful tool for the diagnosis of retinal disorders at both the photoreceptor and the ganglion cell levels. 

 Supported by Austrian Science Fund Grants FWF-P14218PSY (WD), FWF-Y159PAT (WD), and CRAF-199970549 (WD); the Christian Doppler Society (WD); European Community Grant QLK6CT200100279 (PKA); and the Vienna University Jubilee Fund (PKA).

 Submitted for publication July 2, 2002; revised October 1, 2002; accepted November 2, 2002.

 Disclosure: M. Gloesmann, None; B. Hermann, None; C. Schubert, None; H. Sattmann, None; P.K. Ahnelt, None; W. Drexler, Carl Zeiss Meditec (F)

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Comparison of in vitro ultrahigh-resolution OCT imaging with histology in the pig retina. (A) Cross-sectional OCT image with 1.4-μm axial × 3-μm radial resolution of an area of 330 × 410 μm consisting of 6600 × 410 pixels. (B) DIC micrograph of a frozen section obtained from the matching retinal position. From the proximal (top) to the distal (bottom) retina, alternate dark-light bands of signal in the OCT image directly correlate with the retinal layers. Distal to a band attributable to the ONL, a more delicate bright ribbon is likely to represent both the ELM and the myoid portion of the cone IS. The adjacent dark and stippled signal is in alignment with the cone ellipsoids prominent with DIC optics. A distal dark band is possibly associated with the cone OS tips and, finally, a dark-light banding is attributable to the RPE-choriocapillaris complex. (B) Due to histologic processing, the RPE was detached and the photoreceptor layer slightly bent. Scale bar, 100 μm.

Figure 2.

 

Peripapillary GCL. (A) Photograph of the in vitro fundus showing the papilla and major blood vessels. Bar: the orientation of the OCT scan. (B) Ultrahigh-resolution OCT image as obtained by focusing on the proximal retinal layers with 1.4-μm axial × 3-μm transverse resolution of an area of 220 × 390 μm consisting of 4400 × 390 pixels. (C) Micrograph of the matching radial frozen section stained with cresyl violet. The dark NFL is well distinguishable from the bright band of the underlying GCL. Toward the inner retinal border, bundles of ganglion cell axons (B, C, arrows) are separated by Müller cell end feet. Scale bar: (B, C) 100 μm.

Figure 3.

 

Distal retina and choroid. (A) Ultrahigh-resolution OCT image with 1.4-μm axial × 3-μm transverse resolution of an area of 240 × 360 μm consisting of 4800 × 360 pixels. (B) Left: noise reduction profile obtained by performing median-density projection on the OCT image along the x-axis; right: density plot. Bm, Bruch’s membrane; bv, blood vessels. (C) Micrograph of the matching frozen section stained with cresyl violet. Distal to the ONL (onl), a light band corresponds to the position of the ELM (elm) and the proximal parts (myoids, m) of the photoreceptor IS (is). An adjacent band of dark signal presumably results from the specific light-guiding properties of the paraboloid cone ellipsoids (e) and is followed by a light band corresponding to the (predominately rod) photoreceptor OS (os). Adjacent to the distinct signal of the RPE (pe/Bm), a band of light signal corresponds to the choriocapillary lumen (chc). Diffuse dark signal within the choroid (ch) is locally lightened by the profiles of the major blood vessels (bv; arrows). Tissue processing caused detachment of the retina (C, arrowheads). Scale bar: (A, C) 100 μm.

Figure 4.

 

Time lapse sequence of retinal detachment during in vitro ultrahigh-resolution OCT imaging with 1.4-μm axial × 3-μm transverse resolution. (A–C) Enlarged image portions of the detachment zone covering an area of 0.5 × 0.5 mm consisting of 10,000 × 500 pixels, fast Fourier transform (FFT) filtered for reduction of radial shadowing effects. Time intervals between these images was 15 minutes. In the aligned micrographs, progressive bending and swelling of tissue is evident. (A) All photoreceptor sublayers have begun to bend proximally, resulting in a subretinal lumen. (B) The subretinal blebs have significantly increased. Underneath the detachment zone, the dark band of signal corresponding to the RPE/Bruch’s membrane complex is still continuous (arrowheads) but locally triple layered, as indicated by a bright inclusion (arrow). (C) The innermost aspect of the triple-banded signal appears interrupted (arrow). (D) Full OCT image of an area of 0.7 × 2 mm consisting of 14,000 × 2000 pixels. (E) Corresponding histologic section. Tissue processing led to a more extended retinal detachment. Debris from the damaged RPE (arrow) is visible in the subretinal space (✶). Scale bars, 100 μm.

The authors thank Angelika Unterhuber, Kostadinka Bizheva, Boris Povazay, and Leopold Schachinger, the Department of Medical Physics; Andreas Stingl, Andreas Poppe, Gerald Jung, Tuan Le, and Christian Warmuth, Femptolasers Inc., and Carl Zeiss Ophthalmics Systems, Inc. for technical support; and the Institute of Biomedical Research for providing the tissue.