Stereoscopic Photography With a Scanning Laser Ophthalmoscope Donald A. Frambach, M.D., Mark P. Dacey, M.D., and Alfredo Sadun, M.D. Digital fundus angiography has advantages over conventional photography. However, digital angiography as it is normally performed, provides no stereoscopic information. Because stereoscopic data can be crucial in the evaluation of conditions such as age-related macular degeneration, we developed methods to record stereoscopic information during video angiography with a scanning laser ophthalmoscope and to produce high-quality static stereoscopic images. Stereoscopic information was collected by moving the scanning laser ophthalmoscope side-to-side and with a modified Allen separator. Stereoscopic images were displayed as stereoscopic pairs on 35-mm film, stereoscopic pairs from a digital printer, and as stereoscopic images directly on a conventional video monitor. This system to capture and display stereoscopic information is convenient, uses readily available technology, and can be adapted to any digital angiography system. DIGITAL VIDEO FUNDUS ANGIOGRAPHY has several advantages over conventional angiography, including instant access to retinal images, the ability to do computer-image processing, and increased light sensitivity that makes indocyanine green angiography possible.':" However, digital video fundus angiography as normally performed provides no stereoscopic information. We use a scanning laser ophthalmoscope for digital video fundus angiography because it may improve resolution on the retina that exceeds that of conventional photography'< and provides much more temporal information than
Accepted for publication July 28, 1993. From the Doheny Eye Institute and the Department of Ophthalmology, University of Southern California School of Medicine, Los Angeles, California. Reprint requests to Donald A. Frambach, M.D., Doheny Eye Institute, 1450 San Pablo St., Los Angeles, CA 90033.
484
conventional angrography.r" We used a simple and convenient system to record and display stereoscopic data collected with the scanning laser ophthalmoscope. This system for stereoangiography can be adapted to any other system used for digital fundus angiography.
Material and Methods A scanning laser ophthalmoscope (Rodenstock Model 102, Danbury, Connecticut) was used to produce continuous video fluorescein angiograms. The video angiograms were stored on a super-VHS video cassette recorder (JVC HR-SS8000, Elmwood Park, New Jersey). To gather the stereoscopic data, the scanning laser ophthalmoscope was manually moved side-toside in a manner similar to that commonly used to obtain stereoscopic photographs with a conventional fundus camera." On replay of the videotape of the scanning laser ophthalmoscope angiogram, individual frames were chosen from the right and left perspectives. To maximize stereopsis, we selected images just before they became vignetted by the right and left side of the pupil, respectively. The frames were digitized by using a video digitizing card (Data Translation DT 2861, Marlboro, Massachusetts) in a microcomputer (80386 33-MHz ISA bus). These image files were then written to a 3S-mm film recorder (Matrix PCR, Matrix Instruments, Orangeburg, New York) by using custom software to produce standard 3S-mm slides that were then grouped and viewed as stereoscopic pairs with a regular stereoscopic viewer (Fig. 1). These images were also written to a digital printer (Kodak XL-7700, Rochester, New York). Custom software was used to produce either 4.0 x 6.S-cm transparencies or positive prints. The images were viewed with a stereoscopic viewer consisting of a trial frame fitted with +4.00diopter lenses and 30.00-diopter base-out prisms.
©AMERlCAN JOURNAL OF OPHTHALMOLOGY
116:484-488,
OCTOBER,
1993
Vol. 116, No.4
Stereoscopic Photography and Scanning Laser Ophthalmoscopy
485
Fig. 1 (Frambach, Dacey,'ana Sadun). Stereoscopic pair from the scanning laser ophthalmoscope showing an area of subretinal neovascularization sized to be viewed with +S.OO-diopter lenses. Note the elevated retinal vessels over the serous retinal detachment, the deep hemorrhage, and the irregular surface of the hyperfluorescent neovascularization.
The stereoscopic images were also viewed directly on the video screen (Sony Model PVM1342Q, Tokyo, Japan) by using the interlaced video output from the video digitizing card. To do this, the two corresponding stereoscopic images were digitally superimposed using custom software that used a first-order warping algorithm with three tie points per image. The two superimposed images were then broken down into corresponding video fields and recombined into a single image, such that one video field corresponded to the left stereoscopic frame and the other video field corresponded to the right video frame. (Each field was alternately presented to the monitor at 30 Hz.) This combined image was viewed through wireless liquid-crystal diode stereoscopic viewing goggles (Stereo Media, Stereoscopic Medical Imaging Systems, Burbank, California) that were synchronized with the video screen so that the observer's left eye saw the left-perspective image and the right eye saw the right-perspective image. (The goggles alternately occluded each eye at 30 Hz.) We also prepared a reverse stereoscopic image, whereby the left eye saw the right perspective and the right eye saw the left perspective. Custom software permits instant toggling between the stereoscopic and reverse stereoscopic image. We also modified an Allen separator (Carl Zeiss, Inc., Princeton, New Jersey) to collect stereoscopic information with the scanning laser ophthalmoscope. A piece of flat optical glass, 75 x 55 x 3-mm thick, was mounted on
an extending rod fastened to the Allen separator. The Allen separator itself was connected to an X,Y,Z-axis positioning device (Edmund Scientific, Barrington, New Jersey) so that the plano optical glass could be optimally positioned between the scanning laser ophthalmoscope and the patient's eye (Fig. 2). This modified Allen separator worked exactly like an Allen separator on a conventional fundus camera and provided the right and left perspectives required to create the stereoscopic images.
Results
We initially displayed the stereoscopic information as 35-mm slide pairs of quality comparable to that of conventional stereoscopic pairs on slide film. However, the images required time to develop and mount. Accordingly, we made digital prints of the stereoscopic pairs. The resulting 4.0 x 6.5-cm transparencies- and positive prints provided larger images for viewing and were available almost instantly. Unfortunately, the quality of these images was not sufficient for optimal interpretation of the angiograms. The best means of stereoscopic viewing was to display the image on a video screen. These stereoscopic images can be prepared within three minutes and normal and reverse stereoscopic images can be instantly interchanged for enhancement of depth perception. The stereo-
486
October, 1993
AMERICAN JOURNAL OF OPHTHALMOLOGY
Discussion
Fig. 2 (Frambach. Dacey, and Sadun). The modified Allen separator. The optical glass is positioned on an extending rod positioned between the laser aperture and the patient's pupil. A solenoid rapidly rotates the glass so the scanning laser ophthalmoscope records the right- and left-perspective images.
scopic image displayed on the monitor can be viewed from an angle as wide as 60 degrees from normal by as many people as are wearing the viewing goggles, in a small room or large auditorium. In practice, we found it difficult to select the optimal left- and right-perspective images for the stereoscopic pairs because, as the scanning laser ophthalmoscope moved from side-toside, the greatest stereoscopic information was captured just before the image was degraded by interference from the edge of the pupil. It was difficult to select images from real-time video just before they became degraded. The modified Allen separator solved this problem because it could be adjusted to collect maximum stereoscopic data without vignetting from the iris.
A major problem with video angiography as it is typically performed is that important stereoscopic data is not collected or displayed. This stereoscopic information is often essential to diagnose macular diseases. For example, the Macular Photocoagulation Study Group has emphasized the importance of stereoscopic information to detect occult subretinal neovascularization." Consequently, we developed a means to collect and display stereoscopic information conveniently with the scanning laser oph thalmoscope. Like the scanning laser ophthalmoscope, standard fundus cameras have no inherent provision for capturing stereoscopic data, so ophthalmic photographers often simply move the fundus camera from side-to-side to record right- and left-perspective images." An Allen separator can also be used for this purpose." We used both these techniques with the scanning laser ophthalmoscope to capture stereoscopic data and have found it more convenient to use the Allen separator because the Allen separator can be adjusted to obtain maximum stereopsis without vignetting, whereas selecting images from real-time video as the scanning laser ophthalmoscope is moved side-to-side is difficult. There is a unique aspect of the scanning laser ophthalmoscope that increases the amount of stereoscopic information collected, as compared to that obtained with conventional photography. Stereopsis is a function of the horizontal parallax between the right- and leftperspective images. The amount of stereopsis can be calculated from the distance between the centers of the optical axes from the right- and left-perspective images. For any given eye pupil size, smaller optical pupils (of the right- and left-perspective images) permit greater stereopsis because they allow for a larger center-tocenter distance and hence a greater subtended angle (Fig. 3) The optical pupil of the scanning laser ophthalmoscope is 1 mm, whereas the optical pupil of a conventional fundus camera is about 4 mm." For a 4-mm eye pupil, the 4-mm optical pupil of a conventional camera cannot generate stereopsis without vignetting. In contradistinction, two I-mm optical pupils from the scanning laser ophthalmoscope can pass through a 4-mm eye pupil with their centers 3 mm apart. For such a patient, assuming a
Vol. 116, No.4
Stereoscopic Photography and Scanning Laser Ophthalmoscopy
487
= sino' 4/19 = 1'r
conventional photography ,,
________ a'
4mm
4mm
maximum stereopsis obtained.through constraints of pupil
pupil
8 mm
:..3mm ----.
.
pupil
7mm --- ------- ....---
. SLO photography
= sin-1 3/19 = 9°
Fig. 3 (Frambach, Dacey, and Sadun). Comparison of stereopsis between conventional photography and the scanning laser ophthalmoscope (SLO). Left, Stereoscopic separation is limited by the pupil size. The pupil is about 19 mm from the retina. Middle, Stereopsis cannot be obtained with conventional photography without vignetting from the edge of a 4-mm eye pupil because the optical pupil of a fundus camera is the same size. However, with a 4-mm eye pupil, the center-to-center distance of the scanning laser ophthalmoscope's 1-mm optical pupil can measure up to 3 mm without vignetting and 9 degrees of stereopsis can be obtained. Right, More stereopsis (22 degrees vs 12 degrees) can be obtained through an 8-mm eye pupil with the scanning laser ophthalmoscope. Gullstrand schematic eye, l2 an angle of sin " 3/19 or 9 degrees is subtended. At the other extreme, with an 8-mm pupil, the angles subtended by the 4-mm and I-mm optical pupils (without vignetting) are 12 degrees and 22 degrees, respectively. Therefore, much more stereopsis can be obtained with scanning laser ophthalmoscopy than with conventional photography. Moreover, stereoscopic information can be obtained with the scanning laser ophthalmoscope through pupils that are too small to obtain stereoscopic photographs with conventional photography. Additionally, the software and display hardware, which can display stereoscopic images from any source, can rapidly alternate between stereoscopic and reverse stereoscopic images, thereby effectively doubling the amount of perceived stereoscopic information in the images.
As the cost of computer components decreases and more highly refined software is developed, more ophthalmologists have begun to use digital video systems for fluorescein angiography. These video systems are required for indocyanine green angiography because indocyanine green fluoresces so weakly that conventional photography requires high illumination levels that are toxic to the retina.' The system we used to capture and display stereoscopic information is easily implemented by using readily available components and can be adapted to any digital video system used for fluorescein or indocyanine green angiography. ACKNOWLEDGMENTS
The software we used was developed by Donald A. Frambach, M.D., and is the property
488
AMERICAN JOURNAL OF OPHTHALMOLOGY
of the Doheny Eye Institute, Los Angeles, California, and is available at no charge to users of digital angiography systems. Robert Ogawa modified the Allen separator to the authors' specifications.
References 1. Bischoff, P. M., and Flower, R. W.: Ten years experience with choroidal angiography using indocyanine green dye. A new routine examination or an epilogue? Doc. Ophthalmologica 60:235, 1985. 2. Yannuzzi, L. A., Slakter, J. S., Sorenson, J. A, Guyer, D. R., and Orlock, D. A.: Digital indocyanine green videoangiography and choroidal neovascularization. Retina 12:191, 1992. 3. Gabel, V. P., Birngruber, R., and Nasemann, J.: Fluorescein angiography with the scanning laser ophthalmoscope (SLO). Lasers Light Ophthalmol. 2:35,1988. 4. Plesch, A., Klingbeil, D., Rappl, W., and Schrodel, c.: Scanning ophthalmic imaging. In Nasemann, R. E., and Burk, R. O. W. (eds.): Scanning Laser Ophthalmoscopy and Tomography. Munich, Germany, Quintessenz, 1990, pp. 23-32. 5. Frambach, D. A: Video fluorescein angiography with a scanning laser ophthalmoscope. Invest. Ophthalmol. Vis. Sci. 33:723, 1992.
October, 1993
6. Tanaka, T., Muraoka, K., and Shimizu, K.: Fluorescein fundus angiography with scanning laser ophthalmoscope. Visibility of leukocytes and platelets in perifoveal capillaries. Ophthalmology 98: 1824, 1991. 7. Wolf, S., Arend, 0., Toonen, H., Eng, D., Bertram, B., lung. F., and Reirn, M.: Retinal capillary blood flow measurement with a scanning laser ophthalmoscope. Preliminary results. Ophthalmology 98:996,1991. 8. Von Winning, C. H. O. M.: Stereophotography in ophthalmology. In Henkes, H. E. (ed.): Photography, Electro-Ophthalmology and Echo-Ophthalmology in Ophthalmic Practice. The Hague, Dr. W. Junk, 1973, pp. 43-64. 9. Macular Photocoagulation Study Group: Subfoveal neovascular lesions in age-related macular degeneration. Guidelines for evaluation and treatment in the Macular Photocoagulation Study. Arch. Ophthalmol. 109:1242, 1991. 10. Allen, L., Kirkendall, W. M., Snyder, W. B., and Frazier, O. Instant positive photographs and stereograms of ocular fundus fluorescence. Arch. Ophthalmol. 75: 192, 1966. 11. Webb, R. H.: Scanning laser ophthalmoscope. In Masters, B. R. (ed.): Noninvasive Diagnostic Techniques in Ophthalmology. New York, Springer-Verlag, 1990, pp. 438-450. 12. Sadun, A. A., and Brandt, J. D.: Optics for ophthalmologists. A board review manual. New York, Springer-Verlag, 1987, p. 23.
OPHTHALMIC MINIATURE
Enrico left with a gleam of hope in his eyes. All night long Carlos imagined it slowly fading away to pain, disillusionment, and finally confusion over why he had told such a terrible lie. Lawrence Thornton, Imagining Argentina New York, Bantam, 1991, p. 29