ACTA OPHTH ALMOLOGICA VOL. 58 1 980
`
`Department of Ophthalmology1 (Head: Henrik Forsius), University of Oulu, Oulu, Finland,
`The Applied Physics Laboratory2 (Head: Alexander Kossiakoff)
`and the Wilmer Ophthalmological Institute (Head: Amall Patz),
`of the Johns Hopkins University and Hospital, Baltimore, Maryland, USA.
`
`INDOCYANINE GREEN FLUORESCENCE ANGIOGRAPHY
`
`BY
`
`LEA HYVARINEN1 and ROBERT W. FLOWER2
`
`Indocyanine green (ICG) fluorescence angiography has been further refined
`for use in both laboratory and clinical investigations. In the present modifica­
`tion of the Zeiss fundus camera all lenses except the aspherical objective lens
`have been specially antireflection coated to increase light transmission in the
`spectral region around 800 nm. A 300 watt indium iodide lamp continuous
`light source has replaced the conventional xenon flash lamp. This light source
`produces a retinal irradiance of 265 mw, and therefore restricts retinal
`exposure time to 11.9 seconds, but that time is more than adequate to record
`passage of dye through the choroid. Spatial resolution of the fundus on the
`film has been increased from 11.7 microns to 7.4 microns.
`With these technical refinements the choroidal circulation can be studied at 20
`frames per second, which is adequate to document the very rapid movement of
`blood through the vasculature. ICG angiography may change our interpreta­
`tions of choroidal circulatory phenomena which are now based on fluorescein
`angiography, and it clearly is an effective tool in laboratory (experimental)
`investigations.
`
`Key words: indocyanine green fluorescence — angiographic studies of the eye -
`choroidal circulation.
`
`Sodium fluorescein angiography has been an important part of ophthalmology's
`armamemtarium for well over a decade now, and its clinical diagnostic value in
`retinal vascular diseases cannot be impugned. However, its usefulness in studying
`bloodflow dynamics is limited, especially in the choroid where transmission of
`visible light wavelengths is poor and extravasation of fluorescein dye readily occurs.
`
`Received December 7, 1979.
`
`528
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 1 of 11
`
`

`

`ICG angiography
`
`There are even instances in which misinterpretation of fluorescein angiograms may
`have led to erroneous conclusions about choroidal bloodflow (Flower 1972, 1980).
`Development of indocyanine green (ICG) fluorescence angiography was initially
`persued solely as a way to obviate the limitations imposed by fluorescein dye during
`experimental studies of choroidal bloodflow dynamics. It was serendipitous that
`early on in development of the technique it became evident that ICG angiography
`could be used safely and routinely on human subjects as well. Clinical ICG
`angiography has been performed since about late 1973, and the data obtained have
`been invaluable to refinement of the technique. The accumulated angiographic
`material is still insufficient to warrant interpretation of choroidal circulatory
`phenomena in pathological conditions. However, some observations can be repor­
`ted.
`Both ICG absorption angiography and the early method of fluorescence choroi­
`dal angiography have been described in detail before (Flower 1972a,b; Flower &
`Hochheimer 1973, 1976; Flower 1976), therefore it is the refinements and resulting
`improvements in spatial and temporal resolution achieved since these which are
`described below.
`
`Technique
`
`During the development of the new camera for ICG angiography approximately
`150 ICG human angiograms were made at the University of Oulu, Department of
`Ophthalmology and 480 in the Wilmer Ophthalmological Institute. ICG and
`fluorescein angiograms were done a few minutes apart. Injection of ICG dye was
`made via a 3-way cannula followed immediately by a saline flush (Flower 1973). No
`allergic or vasovagal reactions occured, and none of the patients were discomforted
`by the illumination used for ICG near-infrared photography.
`220 ICG angiograms of rhesus monkeys were made at the Wilmer Ophthal­
`mological Institute. The injection technique was identical with that in human
`studies except that smaller volumes were injected, and all animals were anesthetized
`with halothane.
`Fluorescence ICG angiography is performed in much the same way as fluorescein
`angiography, the major difference being that ICG angiography utilizes near-
`infrared light wavelengths while fluorescein utilizes light wavelengths in the visible
`region of the spectrum. Descriptions of the absorption and emission ICG dye
`spectra as well as characteristics of the photographic film and excitation and barrier
`filters used in choroidal angiography have been previously published (Flower &
`Hochheimer 1973, 1976). Briefly, the advantages of ICG dye in choroidal angio­
`graphy are that there is no extravasation from the choriocapillaris, that it absorbs
`
`Acta ophthal. 58,4
`
`34
`
`529
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 2 of 11
`
`

`

` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 3 of 11
`
`

`

`ICG angiography
`
`Graph indicating maximum level of safe retinal irradiance as a function of exposure time.
`
`light and fluoresces in a spectral region where retinal and choroidal pigments are
`fairly transparent, and that its long wavelengths of emitted light are more than six
`times less scattered by the ocular media than the shorter visible light wavelengths
`emitted by fluorescein. Its principal disadvantage is that it does not fluoresce
`efficiently. Whereas the quantam efficiency of fluorescein dye in blood is nearly I,
`that of ICG dye is only 0.13, and if the relative fluorescence intensity of fluorescein
`is arbitrarily set at 1, that of ICG is 25 times less. Therefore, developing the
`instrumentation for choroidal angiography essentially has been an exercise in
`optimizing ability to record the fluorescent light energy emitted by ICG dye in the
`ocular blood vessels.
`
`Fig. 2.
`Ten consecutive frames of a 20 frame per second ICG fluorescence angiogram made of an
`adult rhesus monkey. Arrows indicate the location of the same individual choroicapillaris
`lobule throughout the sequence. In frames A-D, the arterial feeder of the lobule can be
`identified. In frames D-F, to the left of the arrow, the drainage venules into which the lobule
`empties can be seen. Note that in frame A-F the lobule appears white as it fills with dye and
`then appears black in frames G-J where it has essentially emptied. Maximum filling of the
`lobule occurred by frame D, indicating a filling time of only 0.2 seconds.
`
`551
`
`34*
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 4 of 11
`
`

`

`Lea Hyvarinen and Robert W. Flower
`
`The lenses of the fundus camera are normally antireflection coated to efficiently
`transmit light throughout the visible spectrum with maximum transmission oc­
`curring at about 530 nm wavelength. However, near 800 nm wavelength where
`ICG fluorescence occurs, the measured loss of light energy at each lens surface is
`about 4% (the loss nominally occuring at most wavelengths when uncoated lens
`elements are used). By antireflection coating lens elements specifically for 800 nm
`wavelength, this loss can be reduced to as little as 0.3% per lens surface. This was
`done to all lenses in the Zeiss fundus camera imaging optics except the aspherical
`objective lens which is normally uncoated. Providing one of the auxiliary accomoda­
`tion lenses is not used, there are 14 lens surfaces in the fundus camera imaging
`optics. At 800 nm wavelength, with the original lens coatings, total light transmission
`was 0.96014=0.560 or 56%; with proper antireflection coatings, total transmission
`became 0.99714=0.959 or 96% which amounts to 40% more light available to make
`film exposures.
`The effective aperture stop in the Zeiss camera is the hole drilled through the
`diagonal mirror located between the aspherical objective lens and the auxiliary lens
`wheel. Normally this hole is 5 mm in diameter, but by enlarging it to 1 cm, 4 times as
`much light can pass through the imaging optics to the photographic film.
`Reduction of this diagonal mirror surface which reflects light into the eye only
`results in an 11 % decrease of 800 nm wavelength excitation light energy, and that is
`regained by replacing the aluminum coatings on both diagonal mirrors in the
`illumination optics with quartz-clad silver. Added to the gain in transmission
`achived by using better lens coatings, a total gain of 5.7 times in light energy
`transmission at 800 nm wavelength makes it possible to photograph the choroid at
`the 2.5 times magnification of the standard Zeiss fundus camera.
`Of equal importance is improving temporal resolution of the rapid choroidal
`dye-filling sequence of events. Since no flash lamp light source is available which can
`be re-cycled at a sufficiently high rate and yet deliver enough light energy on each
`flash to perform ICG angiography, a continuous light source was installed. Initially,
`a 150 watt quartz halogen lamp was used, and in order to increase the available light
`intensity, an adjustable mirror was also installed behind the lamp. More recently we
`used a 300 watt indium iodide lamp which has a much larger in light energy
`component at 800 nm wavelength. Caution must be exercised in using such a
`continuous light source to insure that retinal irradiance does not exceed a safe level.
`For the approximate 35 degree fundus area illuminated by the Zeiss fundus camera
`and for the light wavelengths required to excite ICG dye to fluorescence, the
`maximum permissible exposure of the retina is 3160 mw-seconds (American
`National Standard for the safe use of lasers 1973). Retinal irradiance is shown as a
`function of continuous exposure time in Fig. 1. The indium iodide light source
`produces a retinal irradiance of 265 mw, consequently, an exposure time of no
`
`532
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 5 of 11
`
`

`

`Fig. 3.
`ICG-angiograms with unusually good visibility of both arteries (A, B) and veins of choroid (C, D).
`
`533
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 6 of 11
`
`

`

`Lea Hyvarinen and Robert W. Flouier
`
`more than 11.9 seconds is permitted, but this is more than adequate to record the
`passage of dye through the choroidal circulation.
`The brighter light source coupled with improved light transmission efficiency of
`the camera optics makes it easy to now record angiograms at rates up to 20 per
`second; this compares favorably with the 4 or 5 frames per second previously
`achieved. At the same time, spatial resolution of the fundus on film was increased
`from 11.7 microns to 7.4 microns which is nearly equal to that theoretically
`achieved in fluorescein angiography.
`During the development of the new camera several hundred angiograms, were
`made to test the different improvements and the clinical application of the method.
`In these angiograms a number of interesting phenomena were recorded, some of
`which are reported in this paper.
`
`Results
`
`Animal experiments
`We have recently succeeded in photographing the passage of ICG dye through the
`choroid at 20 frames per second. The ten consecutive frames in Fig. 2 permit
`visualization of a sharp dye wavefront moving through the entire choroidal
`vasculature of an adult rhesus monkey eye following retrograde injection of less
`than 0.05 ml of dye into the contralateral common carotid artery. They indicate
`that the average time required for dye transit across individual choriocapillaris
`lobules is on the average 1/5 second. The dye wavefront may be easily followed in
`the particular lobule indicated by the arrow throughout the sequence. This series of
`angiograms and others made the same way confirm and even refine choriocapillaris
`bloodflow measurements made earlier. But more significantly, in the context of the
`present paper, they demonstrate the necessity for using high speed angiography to
`study the choroidal circulation. Obviously, subtle changes in choroidal bloodflow
`could be completely missed in angiograms of less temporal or spatial resolution.
`
`Human Studies
`Although the choroid is nearly transparent to both those wavelengths of light used
`to excite ICG dye to fluorescence and those emitted by the dye, there was a wide
`variation in visibility of choroidal vessels among different individuals. The reasons
`for this variation are not apparent; choroidal vessels were quite often very poorly
`visualized in young healthy persons with no known choroidal pathology, yet poor
`visibility was not due simply to poor photographic technique/resolution of visible
`details was good. In less than one-third of the angiograms were choroidal vessels
`seen as clearly as those in Fig. 3 where both the short ciliary arteries and their
`
`534
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 7 of 11
`
`

`

`ICG angiography
`
`Fig. 4.
`ICG is more reliable than fluorescein for measurement of dye appearance time in choroid,
`but ICG fluorescence is also attenuated by the pigment. Note that the fluorescence intensity
`of dye in the cilioretinal artery (era) is higher than in nearby choroidal vessels although dye
`concentrations in these vessels most probably are identical.
`
`branches and later the choroidal veins are well defined; smaller wein branches and
`choriocapillaris are not visualized.
`For measurement of dye appearance time in the choroidal vasculature, ICG is
`more reliable than fluorescein because it is more easily seen through the retinal and
`choroidal pigment, and it does not become extravasated. ICG fluorescence,
`however, can be attenuated by pigment as demonstrated in Fig. 4. First appearance
`of dye traces in the cilioretinal artery is seen in the frame preceeding Fig. 4, whereas
`the retinal vessels are barely visible. Note that the fluorescence itensity of dye in the
`cilioretinal artery is higher than that in the nearby choroidal vessels although dye
`concentration in both these vessels would be expected to be nearly identical.
`Blockage of ICG fluorescence is similar to that observed with fluorescein; blood
`or pigment clumps in retina can produce distinct areas of reduced or non-
`fluorescence in angiograms; similarly, the egg-yolk substance in Best’s disease
`blocks choroidal ICG fluorescence. But on the other hand, ICG angiography does
`
`535
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 8 of 11
`
`

`

`In ICG-angiograms (A, B) there is an area of decreased dye fluorescence (arrow) which is not
`detectable in corresponding fluorescein angiograms (C, D). Times measured From the first
`appearance of the dye: 1,6 seconds (A, C), 3.2 seconds (B, D,).
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 9 of 11
`
`

`

`ICG angiography
`
`Fig, 6
`A: Fluorescein angiogram of a large submacular neovascularization with a large feeding
`artery (arrow). The draining vein wraps around the artery, and its proximal part cannot be
`seen because of diffuse fluorescence near the disc.
`B: In the ICG-angiogram the vein is seen not to follow the artery to the disc margin and thus
`photocoagulation was placed close to the edge of the disc.
`
`sometimes reveal circulatory changes in choroid which are not detectable in
`corresponding fluorescein angiograms because of the masking effect of fluorescein
`which diffuses from the choriocapillaris into the intervascular space (Flower et al.
`1977; Hyvarinen & Maumenee 1971). Figs. 5A and 5B show an area of delayed
`filling in the choroid, but in corresponding fluorescein angiograms (Figs. 5C and
`5D) no such area of delayed filling is detectable.
`So far ICG angiography has not been used for clinical diagnosis, but the findings
`have occasionally affected clinical decisions. In Fig. 6A a large submacular neovas­
`cularization with a large feeding artery is shown. The draining vein could not be
`seen in the fluorescein angiograms clearly enough to warrant photocoagulation of
`the feeder at the disc margin. In the ICG angiogram (Fig. 6B) the vein can be seen
`to follow the artery only half way to the disc. Based upon this observation,
`photocoagulation was made at the disc margin and the neovascularization collapsed
`without danger of venous bleeding.
`
`537
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 10 of 11
`
`

`

`Lea IIyvdrinen and Robert W. Flower
`
`Discussion
`
`Development of ICG angiography has been ongoing for several years. The earlier
`method which used adaptors and an otherwise unmodified Zeiss fundus camera has
`given some insight into the choroidal circulation (Craandijk 8c Van Beek 1976;
`Forsius et al. 1977). The technical refinements now make possible high-speed
`choroidal angiography in clinical investigations. Before it can be used as a
`diagnostic tool, however, a fairly large number of studies are needed to develop the
`basic interpretations of different circulatory patterns. At the present time ICG
`angiography has already demonstrated that fluorescein angiography can give a
`false impression of circulatory status in some clinical conditions.
`
`References
`
`American National Standard for the Safe Use of Lasers, publication ANSI Z136.1-1973
`(1973) by the American National Standards Institutes, Inc., New York.
`Craandijk A. & Van Beek C. A. (1976) Indocyanine green fluorescence angiography of the
`choroid. Brit.J. OphthaL 60, 377—385.
`Flower R. W, (1972a) Infrared absorption angiography of the choroid and some observations
`on the effects of high intraocular pressures. Amer.J. Ophthal. 74, 600—614.
`Flower R. W. (1972b) Simple adaptors for fast conversion of a fundus camera for
`rapidsequence ICG fluorescence choroidal angiography./. Brit. Photog. Assoc. 45, 43-47.
`Flower R. W. 8c Hochheimer B. F. (1973) A clinical technique and apparatus for simultaneous
`angiography of the separate retinal and choroidal circulations. Invest. Ophthal. 12,248—261.
`Flower R. W. (1973) Injection technique for indocyanine green and sodium fluorescein dye
`angiography of the eye. Invest. Ophthal. 12, 881—895.
`Flower R. W. & Hochheimer B. F. (1976) Indocyanine green dye fluorescence and infrared
`absorption choroidal angiography performed simultaneously with fluorescein angiography.
`Johns Hopkins Med. J. 138, 33-42.
`Flower R. W. (1976) High Speed human choroidal angiography using indocyanine green dye
`and a continuous light source. In: DeLaey J. J. (Ed.). International Symposium on
`Fluorescein Angiography. Documenta Ophthalmologica Proceedings Series, Vol. 9, pp.
`59—64. Dr. W. Junk bv., the Hague,
`Flower R. W., Speros P. 8c Kenyon K. R. (1977) Electroretinographic changes and choroidal
`defects in a case of central retinal artery occlusion. Amer.J. Ophthal. 83, 451—459.
`Flower R. W. (1980) Choroidal fluorescent dye filling patterns: a comparison of high speed
`ICG and fluorescein angiograms (To be published).
`Forsius H., Hyvarinen L., Nieminen H. 8c Flower R. W. (1977) Fluorescein and indocyanine
`green fluorescence angiography in study of affected males and in female carriers with
`choroideremia. Acta ophthal. (Kbh.) 55, 459—470.
`Hyvarinen L. & Maumenee A. E. (1971) Interpretation of choroidal fluorescence. Amalric,
`P., (Ed.). Proc. int. Symp. Fluorescein Angiography, Albi (1969) pp. 183—188. S. Krager.
`Basel, New York.
`
`Author’s address:
`R. Flower, Applied Physics Laboratory, Johns Hopkins Road, Laurel, Maryland 20810, USA.
`
` VISIONSENSE - 1014
` VISIONSENSE v.
` NOVADAQ TECHNOLOGIES
` Page 11 of 11
`
`