A novel technique, Multiply Scattered Light Tomography (MSLT), and confocal Infrared Imaging are used to provide diagnostic information using a comfortable, rapid, and noninvasive method. We investigated these techniques in detecting neovascularization in age-related macular degeneration. The MSLT used a Vertical Cavity Surface Emitting Laser (VCSEL) at 850 nm, while the confocal imaging technique used either the VCSEL or a 790 nm laser diode. Both were implemented into the topographical scanning system (TopSS, Laser Diagnostic Technologies, Inc.) Confocal imaging with both lasers provided different information about neovascularization as a function of focal plane, and different also from MSLT.
©2000 Optical Society of America
Age-related macular degeneration (AMD), which attacks the photoreceptors and the retinal pigment epithelial cells beneath them, is the leading cause of blindness in adults in industrialized countries (1). The detection and localization of pathology is frequently based upon outmoded technology, never optimized for AMD or older patients. Examples include ophthalmoscopy with visible wavelength light that fails to penetrate cataractous lenses, color photography with film, and fluorescein angiography through poor ocular media. The latter is particularly degraded with aging, since excitation is with short wavelength light. We tested whether confocal imaging with infrared light can detect and localize neovascularization, the most severe complication of AMD, which often leads to extensive vision loss.
Before the advent of modern photonics, imaging of human retinal and subretinal tissues was limited by light safety, visible wavelength sources, and low sensitivity detectors. The problems specific to the eye include small pupils, reflections from the surfaces of the cornea and lens, poor ocular media anterior to the retina, poor return of light from the retinal and subretinal layers, long-range scatter over large retinal areas, and uncontrollable eye movements. Technical advances in the past two decades have improved light efficiency, the sampling of light from target tissues, and the speed of image formation.
One major advance is laser scanning, in which a narrowly focused laser beam illuminates retinal loci sequentially (2–4). An image is formed digitally, so that optical cross-talk between retinal locations is eliminated. There is no long-range scattered light to reduce the contrast of an image, a main problem with conventional fundus photography. The reduction of light needed for a good quality image is great. This increases safety and improves comfort.
A second major advance is the confocal design of imaging instruments, with an accessible focal plane conjugate to the target tissue, in which apertures are inserted (5–6). The light returning from the target, in this case the retina, is controlled by use of an aperture. A pinhole aperture aligned with the optical axis of the instrument is used to emphasize light that is directly backscattered from retina, and therefore can pass through the narrow aperture. These are called confocal images. However, confocal apertures can include a variety of shapes, not only the circular pinhole with a central free zone, but also a slit, a pinhole that is offset from the optical axis, or an annular aperture with a central stop. The annular apertures, offset pinhole, or offset slit all block light that is directly backscattered from the plane of focus (7–11). The collected light is often of smaller magnitude per unit free zone of the aperture. Scattered light can be either an insignificant or a dominant portion of an image acquired without benefit of confocal apertures. Annular apertures collect mainly light multiply scattered from the plane of focus, or scattered from out of focus planes. Conventional fundus photography lacks apertures; light returning from a large illumination area is collected. Research fundus camera optics with confocal apertures have been used to great advantage, without laser scanning, to measure spectra from well specified retinal locations (11–12).
novel illumination sources
Laser scanning allows the use of a broad range of laser sources (7–11, 14). Of interest is near infrared light, with data from human eyes dispelling the misconception that the point spread function is uselessly broad. Once used to provide a retinal reflection of gross features to indicate that the eye was open, infrared images now visualize clinically meaningful, deeper structures unattainable with other methods (7–11, 12–16). Several new types of solid state lasers had their first imaging or biomedical application in laser scanning ophthalmoscopes, e.g. Ti:SaF, Cr:Li:SaF, and Vertical Cavity Surface Emitting Lasers (VCSELs) (4,11).
types of scanning laser instruments
Scanning laser instrumentation for laboratory use has had a variety of realizations (2–9, 17), but the two main types of scanning laser ophthalmoscopes with commercial markets have significant differences in optical design and intended use. One was developed by Webb and colleagues, initially marketed by Rodenstock Instrumente (Ottobrunn-Riemerling, Germany), to image the eye with the lowest possible amount of light (2–3, 7–11). The image also provides monitoring during functional measurements (18). The light efficiency is accomplished by the use of a small illumination beam in the center of the entrance/exit pupil of the instrument. Light returning from ocular structures is sampled in the remaining annular region, which covers nearly all the dilated human pupil. The pinhole apertures in this instrument are typically a minimum of 100 µm in the plane of the retina. This fairly large pinhole, combined with the exit pathway not sampling the light returning at the highest angle from the retina, provides an image containing information from a fairly thick section of tissue.
Laser scanning ophthalmoscopes derived from the Heidelberg University group have much smaller confocal pinholes, to sample directly backscattered light. The apertures are similar in size (27–40 µm) to the beam focused on the retina in all three types of instruments (10–30 µm). These instruments are available from Laser Diagnostic Technologies (San Diego, CA) and Heidelberg Engineering (Heidelberg, Germany). Topographical information is derived from a series of images acquired sequentially from varying focal planes. These tomographic instruments have a beam separator/combiner and symmetric entrance/exit pupils of 3–3.5 mm diameter. With the beamsplitter, smaller exit pupil, and small pinhole, more light is needed to form an image. Present instruments use near infrared or red sources to avoid excessive absorption by retinal and choroidal tissues.
The optics of the eye impose limitations on the axial transfer function of light returning from the retinal and subretinal layers, the resulting function having more than 10 times the half-width of confocal microscopy. Correction of ocular aberrations is often suggested, and the initial approach was with deformable mirrors in a laser scanning tomographic instrument (19). Post processing of grayscale information is another current approach.
interaction of light and tissue
Modeling light-tissue interactions in the living retina is as important as instrumentation. Geometric structures including fluid-filled lesions such as retinal cysts or subretinal new vessels have been shown to alter the light-tissue interactions, providing clinical information (8,10,13,15–16, 20–27). Elevated lesions include cysts, edema around macular holes, pigment epithelial detachments, choroidal neovascularization, and macular edema. The vitreo-retinal interface, or nerve fiber layer, typically returns far more light than deeper or more superficial layers. The relative axial location of this brightest layer has been used to perform calculations of relative heights across the retina. A single-peaked transfer function is a typical assumption, although several authors have used the additional peaks, inflections, or width of the function (23–26). The retina often is elevated over such lesions, and the relative height of the brightest structures across the surface indicates the three-dimensional extent of pathology. An exception is when a cyst in the edematous retina contains sufficient fluid, in this case the highest peak in the axial transfer function can be beneath the retinal surface.
All imaging modalities, whether confocal imaging with reflected light, interferometry, or polarimetry, depend upon the distribution of index of refraction changes. Tissues that lack a strong index of refraction change within a narrow enough region, situated in a configuration perpendicular to the illumination beam, produce weak direct backscatter, interference fringes, or polarization signals. A decrease in signal can be interpreted as absorption, variations in the geometry of the tissues with respect to the illumination beam, or lack of the necessary index of refraction change to result in backscatter. Multiply scattered light can result from light-tissue interactions following forward scatter, as well as the generally accepted backscatter from out-of focus planes and lateral scatter off tissues that are poor absorbers. Multiply scattered light, like dark field microscopy, can elucidate structures not visualized or amenable to measurements with directly backscattered light.
We have previously used confocal imaging to study exudation in age-related macular degeneration with exudative lesions (13,22,23). These are characterized by the formation of new blood vessels that leak. Typically these originate from beneath the retina, from the choroidal circulation, and are poorly visualized by many techniques employing visible wavelength light. This is particularly true for pathology beneath the macular pigment. Recently, we showed that exudative lesions that are high, particularly with respect to their diameters, are strongly correlated with severe vision loss. We interpret these data as indicating that there are complex or multiple sources of neovascularization and a build-up of fluid, with these cases typically involving not only the choroidal vasculature, but the retinal vasculature as well. This can directly damage the neural elements of the retina.
We use confocal imaging and a new technique, multiply scattered light tomography (MSLT), which utilizes light that is scattered multiple times without being swamped by a strong signal from directly backscattered light in the plane of focus. Our goal is to detect and localize neovascularization in AMD in a rapid, non-invasive method. The height maps from confocal tomography are objective. MSLT provides new opportunities to use differing sources of light for computations. We show that confocal images from different focal planes provide different information, despite the broad axial transfer function of the human eye.
Confocal image series were acquired and height measurements were made with a Topographic Scanning System TopSS™ (Laser Diagnostic Technologies, Inc., San Diego, CA). The TopSS™ uses near infrared illumination, penetrating moderate cataract or cloudy media. In a large segment of the aging population, cloudy media or small pupils are present. The illumination is readily tolerated, and the instrument is optimized for a 3 mm pupil. Data acquisition consists of 32 sections acquired from anterior to posterior, sequentially in 0.9 sec. Each 256×256 pixel image may be acquired as a 10×10, 15×15, or 20×20 deg field, or the instrument may be adjusted to 10×10, 20×20 or 30×30 deg, as it typically has been for recent MSLT work or instruments that also can perform angiography. When the intensity for a given fundus location, as determined by its lateral co-ordinates, is compared for each image in the series, the resulting axial transfer function may be used to characterize the light-tissue interactions at that location.
Two illumination sources were used for the CT data given here. A 790 nm laser diode is used in the commercially available instrument, as shown in the first case in the Results. For the second case, a prototype instrument, the center element of an 850 nm VCSEL (VIXEL Corp., Broomfield, CO), was aligned on axis with the confocal pinhole (11). Additional optics were introduced so that the highly diverging VCSEL beam was collimated, and the laser spacing on the retina was minified to 47 µm. The pinhole size was 40 µm in diameter with respect to the retina, while the usual TopSS™ pinhole is 24 µm. If wavelength were the only factor, the lateral transfer functions theoretically would be 1.08 times broader with the 850 nm source and given the larger pinhole, the axial transfer also worse. However, the VCSEL beam shape is radially symmetric, i.e. lacks the asymmetric divergence of an edge emitting diode. Thus the VCSEL provides the potential for better initial focus on the tissues of interest and narrowed point spread functions. In addition, we have previously shown that the somewhat longer wavelengths can produce a higher contrast for the deeper structures in patients with AMD(10).
For the MSLT data shown here, the image series were acquired simultaneously with the confocal image series when the VCSEL was used. There was line-by-line alternation of the center laser with the surrounding 8 lasers, which were off-axis with respect to the pinhole. The MSLT images are from the light that does not return from single scattering off a plane fairly orthogonal to the direction of the illumination, but instead result from multiply scattered light. The alternation allows the virtually simultaneous acquisition of pairs of images matched in both lateral and axial position, removing the effects of unwanted patient eye movements between pairs of images. In pilot work it was shown to be necessary to increase the axial range of the MSLT image series to a nominal 4.3 mm to explore both of the typical two peaks of the axial transfer function. Thus, the data using the VCSEL have a slightly larger axial range than with the commercial instrument, a nominal 3 mm.
Our TopSS data set includes 60 patients with exudative retinal disease, including 45 patients with exudative AMD, as well as numerous other types of patients and control subjects without retinal disease, all tested in Boston. Our MSLT data set from three clinical sites and Dr. Elsner’s laboratory has 56 patients with retinal disease, including exudative AMD, non-exudative AMD, diabetic macular edema, epiretinal membrane, retinitis pigmentosa, and myopia, plus a broad age range of subjects with normal or sub-clinical fundus findings. All patients and normal subjects signed a consent form prior to testing that was reviewed, along with a research protocol, by either the Institutional Review Board of the Schepens Eye Research Institute or Laser Diagnostic Technologies, depending on the location of testing.
Individual image sections or averages of 2–4 images were produced using a Matlab program (MathWorks, Natick, MA) as well as the calculations from a series of 32 images. To obtain full information for the three-dimensionality of structures, a cross-sectional approach was taken. Intensity values from the axial intensity functions, or the intensity difference values, were plotted to form a height map or a cross-sectional image from the 32 sections in the x,z plane. The Matlab program allows any x,z or y,z plane to be selected by the operator, according to the feature of interest.
Confocal tomography (CT) of a 77 year old female patient with AMD showed exudation in a similar location as the traditional clinical techniques, but without the injection of dye or bright lights (Figs. 1–3). Visual acuity was severely reduced: 20/300 in the right eye and 20/200 in the left. A dilated fundus examination of the right eye showed a serous retinal detachment of the macula. The fluorescein angiogram of the right eye revealed classic choroidal neovascularization with a retinal pigment epithelial tear, seen in the confocal images. The tear, a sign of poor prognosis, is readily visualized in the CT images. The tear is somewhat seen in the color image, not at all in the red free image, and to varying degrees in the fluorescein angiograms.
Data from the MSLT device shows the exudation in an 82 yr old male is shown in Figs. 4–5. Both modalities, CT and MSLT, visualized features of exudation in AMD. The fluid leakage may be thought in the same manner as a protein, such as milk, in low concentration as a test liquid: there is a great deal of scatter, far more than would be expected from the absorption. The result is little return of light from these locations in the fundus (10,15).
Individual sections provided distinctly different views (Figs. 1–5, link 29). The CT images provided the basis for calculations showing retinal elevation associated with the exudation (Fig. 3). Both confocal imaging and MSLT gave information concerning the lateral extent of the exudative lesions, with the confocal images indicating the elevation and distortion of the retina. Subretinal fluid appears dark with respect to the surrounding retina for both imaging modes. The MSLT images showed the borders of exudative lesions beneath the retina, while minimizing the fairly uniform layers of fluid located anterior to the new vessel membranes. In contrast, CT emphasized the retinal features, including the elevation and deformation of the retinal nerve fiber layer by subretinal exudation. CT emphasizes small exudates beneath the retina. Live MSLT images provide the added cue of focusing and motion parallax, and the lesion is localized in seconds. The movies emphasize the differences in sections, from anterior to posterior.
Confocal tomography and MSLT provided a rapid, noninvasive method to detect and localize macular degeneration and pathological structures found in eyes of older patients. There were clear-cut differences among images from different focal planes, despite the view that the axial transfer of the aberrated eye is too broad to selectively image different layers. Features of the pathological fundus were visualized differently in different depth planes, permitting calculations concerning their relative depths. Adequate depth resolution is demonstrated for the evaluation of potential fluid detachment of the neurosensory retina, an important clinical finding. Small elevations with respect to adjacent or more normal portions of the retina may not be measurable with these techniques, but macular cysts associated or topographical changes common in exudative AMD are readily measured (22,23). Future research will build upon the notion of an axial transfer function with multiple inflections in a population with pathological structures associated with AMD.
One example of the clinical utility of these data is illustrated by the first case, shown in Figs. 1–3. The elevation is less than the highest pigment epithelial detachments that we previously reported with the same instrument. These high detachments were shown by angiography to have retinal vascular changes as well as choroidal. This is important clinically because the patients with retinal and choroidal changes, including anastomoses, typically have a poorer prognosis and are the most difficult to treat. However, this technology showed with excellent contrast that this patient had a retinal pigment epithelial tear, disrupting the overlying photoreceptors that provide the first step of vision and depend upon the retinal pigment epithelium for metabolic support.
A second example of clinical utility is illustrated by the second case, with both CT and MSLT shown. The treatment selected for this case was a method used to treat widespread exudation, transpupillary thermal therapy with a low power infrared laser.
Acknowledgements: Supported by EY07624 and Wold Foundation to A.E.E., EY12178 to A.W.D., and DE-FG 02-91ER61229 to Dr. Robert H. Webb.
References, notes, and links:
1. H. Leibowitz, D. E. Kruger, L. R. Maunder, R. C. Milton, M. M. Kini, H, A. Kahn, R. J. Mickerson, J. Pool, T. L. Colton, J. P. Ganley, and J. Loewenstein, “The Framingham Eye Study Monograph: an ophthalmological and epidemilogical study of cataract, glaucoma, diabetic retinopathy, macular degeneration, and visual acuity in a general population of 2631 adults, 1973–1975,” Surv. Ophthalmol. 24, (Suppl) 335–610 (1980). [PubMed]
2. R. H. Webb, GW Hughes, and O. Pomerantzeff, “Flying spot TV ophthalmoscope,” Appl. Optics 192991–2997 (1980). [CrossRef]
4. A. W. Dreher and R. N. Weinreb, “Accuracy of topographic measurements in a model eye with the laser tomographic scanner,” Invest. Ophthalmol. Vis. Sci. 32, 2992–2996 (1991). [PubMed]
6. A.W. Dreher, P.C. Tso, and R. N. Weinreb, “Reproducibility of topographic measurements of the normal and glaucomatous optic nerve head with the laser tomographic scanner,” Am. J. Ophthalmol. 111, 221–229 (1991). [PubMed]
7. For a review of the early infrared scanning laser techniques, see A. E. Elsner, A. H. Jalkh, A.H, and J. J. Weiter“New devices in retinal imaging and functional evaluation,” in Practical Atlas of Retinal Disease and Therapy, W. Freeman, ed. (Raven, New York, 1993) pp. 19–35. The work of several groups is found inA. E. Elsner, D-U. Bartsch, J. J. Weiter, and M. E. Hartnett, “New devices in retinal imaging and functional evaluation,” in Practical Atlas of Retinal Disease and Therapy, W. Freeman, ed. (Lippincott-Raven, New York, 1998) 2nd edition, pp. 19–55.
8. A. E. Elsner, S.A. Burns, S.A. Kreitz, M.R., and J. J. Weiter, “New views of the retina/RPE complex: quantifying sub-retinal pathology,” in Noninvasive Assessment of the Visual System, Vol. 1 of OSA Technical Digest (Optical Society of America, Washington, D.C., 1991, pp. 150–153).
11. A. E. Elsner, A. W. Dreher, Q. Zhou, E. Beausencourt, S. A. Burns, and R. H. Webb, “Multiply scattered light tomography: vertical cavity surface emitting laser array used for imaging subretinal subretinal structures,” Lasers and Light in Ophthalmology 8, 193–202 (1998).
13. L. M. Kelley, J. P. Walker, G. L. Wing, P. A. Raskauskas, and A. E. Elsner, “Scanning laser ophthalmoscope imaging of age related macular degeneration and neoplasms,” J. Ophthalmic Photography 3, 89–94 (1997).
14. A.E. Elsner, S. A. Burns, J. J. Weiter, and M. E. Hartnett, “Diagnostic applications of near infrared solid-state lasers in the eye,” LEOS ’94, IEEE Catalog number 94CH3371-2 , Library of Congress number 93–61268, 1, 14–15 (1994).
15. M E. Hartnett and A. E. Elsner, “Characteristics of exudative age-related macular degeneration determined in vivo with confocal direct and indirect infrared imaging,” Ophthalmol. 103, 58–71 (1996).
16. M. E. Hartnett, J. J. Weiter, G. Staurenghi, and A E. Elsner, “Deep retinal vascular anomalous complexes in advanced age- related macular degeneration,” Ophthalmol. 103, 2042–2053 (1996).
17. J. F. Le Gargasson, F. Rigaudiere, J. E. Guez, A. Gaudric, and Y. Grall, “Contribution of scanning laser ophthalmoscopy to the functional investigation of subjects with macular holes,” Doc. Ophthalmol. 86, 227–238 (1994). [CrossRef] [PubMed]
18. J.-F. Chen, A. E. Elsner, S. A. Burns, R. M. Hansen, P. L. Lou, K. K. Kwong, and A. B. Fulton, “The effect of eye shape on retinal responses,” Clinical Vision Sciences 7, 521–530 (1992).
19. W. Dreher, J. F. Bille, and R. N. Weinreb RN, “Active-optical depth resolution improvement of the laser tomographic scanner,” Appl. Opt. 28, 804–808 (1988). [CrossRef]
20. Remky, O. Arend, A. E. Elsner, F. Toonen, M. Reim, and S. Wolf, “Digital imaging of central serous retinopathy using infrared illumination,” German J. Ophthalmology 4, 203–206 (1995).
21. E. Beausencourt, A. E. Elsner, M. E. Hartnett, and C. L. Trempe, “Quantitative analysis of macular holes with scanning laser tomography,” Ophthalmology 104, 2018–2029 (1997). [PubMed]
22. C. Kunze, A. E. Elsner, E. Beausencourt, L. Moraes, M. E. Hartnett, and C. L. Trempe, “Spatial extent of pigment epithelial detachments in age-related macular degeneration,” Ophthalmology 9, 1830–1840 (1999). [CrossRef]
23. E. Beausencourt, A. Remky, A. E. Elsner, M. E. Hartnett, and C. L. Trempe, “Infrared scanning laser tomography of macular cysts,” Ophthalmology 107, 376–385 (2000).
24. D.-U. Bartsch, M. Intaglietta, J. F. Bille, A. W. Dreher, M. Gharib, and W. R. Freeman, “Confocal laser tomographic analysis of the retina in eyes with macular hole formation and other focal macular diseases,” Am. J. Ophthalmol. 108, 277–87 (1989). [PubMed]
26. C. Hudson, F. G. Flanagan, G. S. Turner, and D. McLeod, “Scanning laser tomography Z profile signal width as an objective index of macular retinal thickening,” Br. J. Ophthalmol. 82, 121–30 (1998). [CrossRef] [PubMed]
27. E. Jaakkola, I. Vesti, and I. Immonen, “The use of confocal scanning laser tomography in the evaluation of retinal elevation in age-related macular degeneration,” Ophthalmology 106, 274–9, (1999). [CrossRef] [PubMed]
28. The link to additional figures is http://color.eri.harvard.edu/annhom.htm.
29. Updates on infrared scanning laser topographic instrumentation are at http://www.laserdiagnostic.com.