- Email: [email protected]
Retinal oxygen saturation levels in patients with central retinal vein occlusion3
Retinal Oxygen Saturation Levels in Patients with Central Retinal Vein Occlusion Shin Yoneya, MD, PhD,1 Tamiya Saito, MD,1 Yoko Nishiyama, MD,1 Tatsuya Deguchi, OP,1 Masayuki Takasu, OP,1 Tamir Gil, PhD,2 Eli Horn, PhD2 Purpose: To validate the efficacy of Fourier transform-based spectral retinal imaging (SRI) in quantifying retinal ischemia. Design: Prospective, observational case series. Patients and Methods: Eleven eyes of 10 patients with central retinal vein occlusion (CRVO) and 4 fellow unaffected eyes of selected patients were examined by both fluorescein angiography and SRI. The fluorescein angiograms were correlated with oxygen saturation maps that were calculated from the SRI. Results: Oxygen saturation levels in the fundus were shown as color grading in 35° fundus images processed by SRI. This grading correlated well with the severity of CRVO, as estimated by fluorescein angiography. Conclusions: The severity of circulatory disturbances in CRVO may be graded by applying SRI. Ophthalmology 2002;109:1521–1526 © 2002 by the American Academy of Ophthalmology. Central retinal vein occlusion (CRVO) is a relatively common and potentially vision-threatening disease. Central retinal vein occlusion can be divided into ischemic and nonischemic types based on fluorescein angiographic findings. The visual outcome may be especially poor when neovascular glaucoma develops after ischemic CRVO. According to the Central Vein Occlusion Study Group, 34% of eyes with nonischemic CRVO progressed to ischemic CRVO within 3 years.1 To prevent the development of iris neovascularization and associated neovascular glaucoma in patients with ischemic CRVO, panretinal photocoagulation has to be initiated in a timely manner. At present, fluorescein angiography is an approved clinical tool that can help to distinguish ischemic from nonischemic CRVO. Recent technological developments have enabled us to measure oxygen saturation levels in the fundus with a 35° view. This technique can be used to study oxygen levels in retinal vessels in a wide variety of diseases and may allow us to investigate clinically the pathophysiologic features of Originally received: May 11, 2001. Accepted: January 28, 2002. Manuscript no. 210316. 1 Department of Ophthalmology, Saitama Medical School, Moroyama, Saitama, Japan. 2 Applied Spectral Imaging Co., Migdal Haemak, Israel. Presented in part as a poster at the annual meeting of the American Academy of Ophthalmology, Dallas, Texas, October 2000. Supported by the Ministry of Education, Science, Sports and Culture, Tokyo, Japan (Grant-in-Aid of Scientific Research (C) no.: 12671721). The authors have no commercial interests in the products or devices mentioned herein. Reprint requests to Shin Yoneya, MD, PhD, Department of Ophthalmology, Saitama Medical School, 38 Morohongo, Moroyama, Iruma, Saitama 350-0451, Japan. © 2002 by the American Academy of Ophthalmology Published by Elsevier Science Inc.
the eye from a new perspective. This technique, therefore, may serve as a crucial means to determine, noninvasively, the point at which CRVO changes from the nonischemic to the ischemic type. For this purpose, we evaluated and ranked CRVO according to the severity of ischemia using spectral retinal imaging (SRI) based on Fourier transformation.
Patients and Methods Between July 1999 and April 2000, 10 patients with recent-onset CRVO were recruited for this study after undergoing a complete ophthalmic examination that included fluorescein angiography. Informed consent was obtained from all patients enrolled. Bilateral CRVO developed in one patient. A retinal cube SRI system was used to measure the oxygen levels in the fundus image. Measurement was performed for 6 seconds in 11 diseased eyes and in 4 unaffected fellow eyes as a control. During the measurement, the fundus was illuminated with the white incandescent light of the fundus camera at the regular power setting. The fundus images obtained by interferometry were processed further for computer analysis. The oxygen map of the fundus was displayed with corresponding colors ranging from dark purple, representing oxygen levels of less than 40%, to red, representing a 100% oxygen saturation level. The theoretical background of SRI is that, to obtain images of the fundus, a retinal cube spectral retinal imager (ASI Co., Migdal Hemak, Israel) is used. The retinal cube system consists of an SD-200 optical head, which uses a Sagnac interferometer (ASI Co., Migdal Hemak, Israel) that has been mounted on top of a fundus camera (TRC-50IA; Topcon Co., Tokyo, Japan) and a software module consisting of an acquisition and an analysis module. This system allows the measurement of a spectrum in each pixel of the fundus image. The Sagnac interferometer divides the light beam that is reflected from the fundus into two coherent ISSN 0161-6420/02/$–see front matter PII S0161-6420(02)01109-0
1521
Ophthalmology Volume 109, Number 8, August 2002
Figure 1. Graph (A) shows standard hemoglobin absorption spectra of oxygenated hemoglobin (HbO2) and deoxygenated hemoglobin (Hb) by solid and dashed lines, respectively. Graph (B) depicts optical density that was calculated in tone pixel of a retinal spectral image. Estimating the oxygen saturation (OS) involves substituting standard hemoglobin absorption spectra (A) into a mathematical model that attempts to reproduce the measured optical density (B). Within the model the standard spectra are linearly combined to yield the best fit to the measured optical density. The ratio of the HbO2 absorption in this linear combination determines the OS value.
beams and creates a variable optical path difference between the two. The two beams are then recombined to interfere with each other. The resulting interference intensity is measured by the detector as a function of the optical path difference. This intensity versus optical path difference function, called an interferogram, is Fourier transformed to recover the spectrum, which is the intensity as a function of the wavelength. For every pixel of an image from the spectrum thus obtained, we calculated the oxygen saturation maps. An oxygen saturation map shows each point or pixel graphically as an oxygen saturation value graded between 0 and 100. This grading accounts for the percentage of hemoglobin that is oxygenated. The oxygen saturation value is calculated for each pixel from the spectrum attached to each pixel in the spectral image. Applying different colors to different oxygen saturation values allows the values in the fundus to be represented graphically. To estimate an oxygen saturation value from the spectrum in each pixel, we assumed the applicability of Beer-Lambert law when building a mathematical model to describe the measured spectrum as an attenuation of the illuminating light signal from the fundus layers. Specifically, we substituted standard extinction coefficients of oxygenated and deoxygenated hemoglobin (Fig 1A) and direct response in our system by a model based on BeerLambert, in which the oxygen saturation and blood column thickness are free parameters. These parameters were then estimated to obtain the best fit between the model and the actual measured spectrum (Fig 1B), yielding the oxygen saturation estimate for each pixel of the map. The oxygen saturation maps were then superimposed on maps of anatomic borderlines by a separate analysis to serve as a guide to the eye and as a spatial coordinating system. In this way, the oxygen saturation maps provided geographic pictures with which physicians are familiar. With this technique, the oxygen saturation was calculated as 90% to 98% at the retinal arteries and as 60% to 70% at the retinal
1522
veins, respectively [Yoneya S, Hayashi N, Mori K, et al. Invest Ophthalmol Vis Sci 1999;40(Suppl):S124].
Results In all 15 eyes examined, SRI displayed the oxygen levels in each pixel of the fundus images as a topographical color map within a 35° field. The severity of circulatory disturbance was well shown by SRI. The findings were in accordance with those observed by fluorescein angiography. Areas of capillary nonperfusion and dye leakage on fluorescein angiography appeared on SRI as dark blue, which signifies oxygen levels below 40%. Spectral retinal imaging also could delineate the severity of the circulatory disturbance of the fundus. In a case of CRVO, SRI showed the presence of circulatory disturbance not only in the affected hemisphere but also in the other hemisphere that appeared intact by ophthalmoscopy and fluorescein angiography. In eyes with advanced CRVO, oxygen saturation was decreased along the retinal artery and at the optic disc. Throughout the study, the presence of retinal hemorrhages did not affect the measurement of oxygen levels in retinal arteries and veins. The findings in four representative cases are described below.
Patient 1 Color fundus photograph of a right eye with CRVO showed extensive retinal hemorrhages, tortuous retinal veins, and diffuse retinal edema (Fig 2A). The fluorescein angiogram of the same eye (Fig 2B) demonstrated extensive capillary nonperfusion, fluorescein leakage, and dye staining of the vessel wall. Topographical SRI mappings illustrated that the entire fundus except the disc area was filled by dark purple dots, indicating an oxygen level below 40%. The retinal artery was displayed with greenish dots, indicating 75% oxygen saturation (Fig 2C).
Yoneya et al 䡠 Retinal Oxygen Saturation Levels in CRVO
Figure 2. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging topographical map (C) of central retinal vein occlusion in the right eye of patient 1. Figure 3. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging topographical map (C) of central retinal vein occlusion in the left eye, and fundus photograph (D), fluorescein angiogram (E), and spectral retinal imaging topographical map (F) of the right eye of patient 2.
Patient 2 Color fundus photograph of a left eye with CRVO showed engorged retinal veins and moderate retinal hemorrhages (Fig 3A). The latephase fluorescein angiogram (Fig 3B) demonstrated extensive fluorescein leakage along retinal veins. Capillary nonperfusion was milder than in patient 1. The oxygen saturation in the fundus was very low in some areas but not in the entire posterior fundus (Fig 3C). Yellow and orange dots were arranged linearly along retinal arteries, suggesting a near-normal oxygen level in the arteries. The fellow eye showed no
particular changes on ophthalmoscopy. Spectral retinal imaging studies were in agreement with findings, because the fundus showed predominantly red and orange dots (Fig 3D,E,F).
Patient 3 This patient had incipient CRVO in both eyes at presentation. Fundus photographs demonstrated scatter blot hemorrhages bilaterally (Figs 4A and 5A). The fluorescein angiogram showed scattered hypofluorescent dots resulting from blot hemorrhages and
1523
Ophthalmology Volume 109, Number 8, August 2002
1524
Yoneya et al 䡠 Retinal Oxygen Saturation Levels in CRVO Figure 4. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging topographical map (C) in right eye of central retinal vein occlusion in patient 3. Figure 5. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging topographical map (C) of central retinal vein occlusion in the left eye of patient 3. Figure 6. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging (SRI) topographical map (C) of hemi-central retinal vein occlusion in the right eye of patient 4 and fundus photograph (D) and SRI topographical map (E) of the fellow eye. 4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™
mild staining of the vessel wall (Fig 5B). The right eye had patchy areas of capillary nonperfusion (Fig 4B). Spectral retinal imaging showed a higher oxygen level in the left eye (Fig 5C) than in the right eye, in which blue dots dominated in the peripheral fundus (Fig 4C). The different features between the two eyes in the SRI topographical map could be confirmed by fluorescein angiography.
Patient 4 The color fundus photograph showed inferior hemi-CRVO in the right eye. Blot retinal hemorrhages were scattered in the inferior hemisphere (Fig 6A). Fluorescein angiography demonstrated mild dye leakage along the retinal veins in the involved hemisphere only (Fig 6B). Spectral retinal imaging studies showed moderately decreased oxygen levels over the entire fundus that were more severe in the posterior fundus area (Fig 6C). The color fundus photograph of the fellow eye showed normal findings and no hemorrhages (Fig 6D). Spectral retinal imaging topographical map also illustrated normal oxygen saturation levels, because the entire fundus was filled predominantly with reddish dots (Fig 6E).
Discussion Noninvasive measurement of oxygen levels in the human fundus has been attempted by a variety of techniques using spectrally resolved reflectance.2– 4 Spectral retinal imaging has many advantages over previous approaches.5,6 Some of the most important advantages are its high throughout, adjustable spectral resolution, broad spectral bandwidth, and relative insensitivity to polarization by incoming radiation. Spectral retinal imaging has enabled us to study oxygen levels in the fundus with a wide-angle view. By overlaying the oxygen saturation measurements on the fundus image, the status of retinal circulation can be expressed as oxygen saturation levels, which cannot be accomplished with fluorescein angiography. In this study of eyes with CRVO, varying severity of circulatory disturbance could be shown by SRI. Although due precaution is necessary in interpreting the findings because of the small number of patients, we compared the findings by SRI and fluorescein angiography. Areas of capillary nonperfusion and adjacent regions in fluorescein angiograms were always shown in blue, suggesting low oxygen saturation. This finding implies that the circulation was slow and poorly perfused, even in the normal-looking capillary bed adjacent to the areas of nonperfusion. In severe CRVO, fluorescein angiography showed fluorescein staining of vessel wall, leakage from the disc and retinal vessels, and variable areas of capillary nonperfusion. In such eyes, SRI showed the entire posterior fundus as being extremely low in oxygen saturation. Oxygen satura-
tion was poor even along retinal arteries. Conversely, CRVO was less severe in eyes with higher oxygen saturation levels along retinal arteries. The power of resolution in the present SRI system is somewhat limited, because the coupled charged device built into our system consists of 284 ⫻ 224 pixels. This resolution means that we are measuring an average of oxygen saturation in each pixel that is approximately 40 m square in size when the angle of view is set at 35°. This feature may pose difficulties when studying retinal arteries and veins of small caliber. As a result, retinal arteries and veins could not be illustrated linearly. Other spatial sensitivity concerning the depth in the measurement has to be cleared because our oxygen saturation estimation model is based on a summation of oxygen saturation from hemoglobin absorption along the optical path. Because the light used for the measurement is between 480 and 600 nm and because retinal pigment epithelium is present, major reflection light for the analysis could be from the retina. Only sufficient infrared light that reaches the choroid through the retinal pigment epithelium could be observed with a monochromatic fundus photograph. Also, the presence of diffuse drusen may interfere with our measurement by blocking the irradiated light. Even so, we still cannot neglect the potential influence of the choroid, which has abundant vessels and fast blood flow. Therefore, in this SRI analysis, we measured the oxygen saturation levels in the fundus instead of in the retina, even though the main source of the information might have come from the retinal circulation. Intraretinal circulation is one of the parameters with which to assess the severity of CRVO using fluorescein angiography. Our SRI system indicates that the degree of ischemia, or decreased oxygen saturation level, may serve as another parameter to evaluate the degree of disturbed circulation in CRVO. Moreover, SRI is noninvasive and repeatable and may prove to be a useful clinical tool in monitoring CRVO and in making decisions to perform timely panretinal photocoagulation in patients with this disease.
References 1. Natural history and clinical management of central retinal vein occlusion. The Central Vein Occlusion Study Group [published erratum appears in Arch Ophthalmol 1997;115:1275]. Arch Ophthalmol 1997;115:486 –91. 2. Brindley GS, Willmer EN. The reflexion of light from the macular and peripheral fundus oculi in man. J Physiol 1952;116:350–6. 3. van Norren D, Tiemeijer LF. Spectral reflectance of the human eye. Vision Res 1989;26:313–20. 4. Delori FC, Pflibsen KP. Spectral reflectance of the human ocular fundus. Appl Opt 1989;28:1061–77.
1525
Ophthalmology Volume 109, Number 8, August 2002 5. Cabib D, Buckwald RA, Garini Y, et al. Spatially resolved Fourier transform spectroscopy (spectral imaging): a powerful tool for quantitative analytical microscopy. In: Farkas DL, Leif RC, Priezzhev AV, et al, eds. Optical Diagnostics of Living Cells and Biofluids. Bellingham, WA: SPIE, 1996; 278 –91. (Proc SPIE; 2678).
6. Lavi M, Milman U, Cabib D, et al. A new compact design interferometer based spectral imaging system for bio-medical applications. In: Cogswell CJ, Conchello J, Lerner JM, et al, eds. Three-dimensional and Multidimensional Microscopy: Image Acquisition and Processing V. Bellingham, WA: SPIE, 1998;313–21. (Proc SPIE; 3261).
Historical Image
Lorgnette, c 1900. In the early 20th century, lorgnettes with very long handles were called “dowagerduchess” types. This English example, made from carved tortoise shell, is enhanced by a gold pin decoration. Published through the courtesy of the Alcon Laboratories Museum of Ophthalmology, The Rosenthal Collection, Fort Worth, Texas.
1526
the eye from a new perspective. This technique, therefore, may serve as a crucial means to determine, noninvasively, the point at which CRVO changes from the nonischemic to the ischemic type. For this purpose, we evaluated and ranked CRVO according to the severity of ischemia using spectral retinal imaging (SRI) based on Fourier transformation.
Patients and Methods Between July 1999 and April 2000, 10 patients with recent-onset CRVO were recruited for this study after undergoing a complete ophthalmic examination that included fluorescein angiography. Informed consent was obtained from all patients enrolled. Bilateral CRVO developed in one patient. A retinal cube SRI system was used to measure the oxygen levels in the fundus image. Measurement was performed for 6 seconds in 11 diseased eyes and in 4 unaffected fellow eyes as a control. During the measurement, the fundus was illuminated with the white incandescent light of the fundus camera at the regular power setting. The fundus images obtained by interferometry were processed further for computer analysis. The oxygen map of the fundus was displayed with corresponding colors ranging from dark purple, representing oxygen levels of less than 40%, to red, representing a 100% oxygen saturation level. The theoretical background of SRI is that, to obtain images of the fundus, a retinal cube spectral retinal imager (ASI Co., Migdal Hemak, Israel) is used. The retinal cube system consists of an SD-200 optical head, which uses a Sagnac interferometer (ASI Co., Migdal Hemak, Israel) that has been mounted on top of a fundus camera (TRC-50IA; Topcon Co., Tokyo, Japan) and a software module consisting of an acquisition and an analysis module. This system allows the measurement of a spectrum in each pixel of the fundus image. The Sagnac interferometer divides the light beam that is reflected from the fundus into two coherent ISSN 0161-6420/02/$–see front matter PII S0161-6420(02)01109-0
1521
Ophthalmology Volume 109, Number 8, August 2002
Figure 1. Graph (A) shows standard hemoglobin absorption spectra of oxygenated hemoglobin (HbO2) and deoxygenated hemoglobin (Hb) by solid and dashed lines, respectively. Graph (B) depicts optical density that was calculated in tone pixel of a retinal spectral image. Estimating the oxygen saturation (OS) involves substituting standard hemoglobin absorption spectra (A) into a mathematical model that attempts to reproduce the measured optical density (B). Within the model the standard spectra are linearly combined to yield the best fit to the measured optical density. The ratio of the HbO2 absorption in this linear combination determines the OS value.
beams and creates a variable optical path difference between the two. The two beams are then recombined to interfere with each other. The resulting interference intensity is measured by the detector as a function of the optical path difference. This intensity versus optical path difference function, called an interferogram, is Fourier transformed to recover the spectrum, which is the intensity as a function of the wavelength. For every pixel of an image from the spectrum thus obtained, we calculated the oxygen saturation maps. An oxygen saturation map shows each point or pixel graphically as an oxygen saturation value graded between 0 and 100. This grading accounts for the percentage of hemoglobin that is oxygenated. The oxygen saturation value is calculated for each pixel from the spectrum attached to each pixel in the spectral image. Applying different colors to different oxygen saturation values allows the values in the fundus to be represented graphically. To estimate an oxygen saturation value from the spectrum in each pixel, we assumed the applicability of Beer-Lambert law when building a mathematical model to describe the measured spectrum as an attenuation of the illuminating light signal from the fundus layers. Specifically, we substituted standard extinction coefficients of oxygenated and deoxygenated hemoglobin (Fig 1A) and direct response in our system by a model based on BeerLambert, in which the oxygen saturation and blood column thickness are free parameters. These parameters were then estimated to obtain the best fit between the model and the actual measured spectrum (Fig 1B), yielding the oxygen saturation estimate for each pixel of the map. The oxygen saturation maps were then superimposed on maps of anatomic borderlines by a separate analysis to serve as a guide to the eye and as a spatial coordinating system. In this way, the oxygen saturation maps provided geographic pictures with which physicians are familiar. With this technique, the oxygen saturation was calculated as 90% to 98% at the retinal arteries and as 60% to 70% at the retinal
1522
veins, respectively [Yoneya S, Hayashi N, Mori K, et al. Invest Ophthalmol Vis Sci 1999;40(Suppl):S124].
Results In all 15 eyes examined, SRI displayed the oxygen levels in each pixel of the fundus images as a topographical color map within a 35° field. The severity of circulatory disturbance was well shown by SRI. The findings were in accordance with those observed by fluorescein angiography. Areas of capillary nonperfusion and dye leakage on fluorescein angiography appeared on SRI as dark blue, which signifies oxygen levels below 40%. Spectral retinal imaging also could delineate the severity of the circulatory disturbance of the fundus. In a case of CRVO, SRI showed the presence of circulatory disturbance not only in the affected hemisphere but also in the other hemisphere that appeared intact by ophthalmoscopy and fluorescein angiography. In eyes with advanced CRVO, oxygen saturation was decreased along the retinal artery and at the optic disc. Throughout the study, the presence of retinal hemorrhages did not affect the measurement of oxygen levels in retinal arteries and veins. The findings in four representative cases are described below.
Patient 1 Color fundus photograph of a right eye with CRVO showed extensive retinal hemorrhages, tortuous retinal veins, and diffuse retinal edema (Fig 2A). The fluorescein angiogram of the same eye (Fig 2B) demonstrated extensive capillary nonperfusion, fluorescein leakage, and dye staining of the vessel wall. Topographical SRI mappings illustrated that the entire fundus except the disc area was filled by dark purple dots, indicating an oxygen level below 40%. The retinal artery was displayed with greenish dots, indicating 75% oxygen saturation (Fig 2C).
Yoneya et al 䡠 Retinal Oxygen Saturation Levels in CRVO
Figure 2. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging topographical map (C) of central retinal vein occlusion in the right eye of patient 1. Figure 3. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging topographical map (C) of central retinal vein occlusion in the left eye, and fundus photograph (D), fluorescein angiogram (E), and spectral retinal imaging topographical map (F) of the right eye of patient 2.
Patient 2 Color fundus photograph of a left eye with CRVO showed engorged retinal veins and moderate retinal hemorrhages (Fig 3A). The latephase fluorescein angiogram (Fig 3B) demonstrated extensive fluorescein leakage along retinal veins. Capillary nonperfusion was milder than in patient 1. The oxygen saturation in the fundus was very low in some areas but not in the entire posterior fundus (Fig 3C). Yellow and orange dots were arranged linearly along retinal arteries, suggesting a near-normal oxygen level in the arteries. The fellow eye showed no
particular changes on ophthalmoscopy. Spectral retinal imaging studies were in agreement with findings, because the fundus showed predominantly red and orange dots (Fig 3D,E,F).
Patient 3 This patient had incipient CRVO in both eyes at presentation. Fundus photographs demonstrated scatter blot hemorrhages bilaterally (Figs 4A and 5A). The fluorescein angiogram showed scattered hypofluorescent dots resulting from blot hemorrhages and
1523
Ophthalmology Volume 109, Number 8, August 2002
1524
Yoneya et al 䡠 Retinal Oxygen Saturation Levels in CRVO Figure 4. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging topographical map (C) in right eye of central retinal vein occlusion in patient 3. Figure 5. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging topographical map (C) of central retinal vein occlusion in the left eye of patient 3. Figure 6. Fundus photograph (A), fluorescein angiogram (B), and spectral retinal imaging (SRI) topographical map (C) of hemi-central retinal vein occlusion in the right eye of patient 4 and fundus photograph (D) and SRI topographical map (E) of the fellow eye. 4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™
mild staining of the vessel wall (Fig 5B). The right eye had patchy areas of capillary nonperfusion (Fig 4B). Spectral retinal imaging showed a higher oxygen level in the left eye (Fig 5C) than in the right eye, in which blue dots dominated in the peripheral fundus (Fig 4C). The different features between the two eyes in the SRI topographical map could be confirmed by fluorescein angiography.
Patient 4 The color fundus photograph showed inferior hemi-CRVO in the right eye. Blot retinal hemorrhages were scattered in the inferior hemisphere (Fig 6A). Fluorescein angiography demonstrated mild dye leakage along the retinal veins in the involved hemisphere only (Fig 6B). Spectral retinal imaging studies showed moderately decreased oxygen levels over the entire fundus that were more severe in the posterior fundus area (Fig 6C). The color fundus photograph of the fellow eye showed normal findings and no hemorrhages (Fig 6D). Spectral retinal imaging topographical map also illustrated normal oxygen saturation levels, because the entire fundus was filled predominantly with reddish dots (Fig 6E).
Discussion Noninvasive measurement of oxygen levels in the human fundus has been attempted by a variety of techniques using spectrally resolved reflectance.2– 4 Spectral retinal imaging has many advantages over previous approaches.5,6 Some of the most important advantages are its high throughout, adjustable spectral resolution, broad spectral bandwidth, and relative insensitivity to polarization by incoming radiation. Spectral retinal imaging has enabled us to study oxygen levels in the fundus with a wide-angle view. By overlaying the oxygen saturation measurements on the fundus image, the status of retinal circulation can be expressed as oxygen saturation levels, which cannot be accomplished with fluorescein angiography. In this study of eyes with CRVO, varying severity of circulatory disturbance could be shown by SRI. Although due precaution is necessary in interpreting the findings because of the small number of patients, we compared the findings by SRI and fluorescein angiography. Areas of capillary nonperfusion and adjacent regions in fluorescein angiograms were always shown in blue, suggesting low oxygen saturation. This finding implies that the circulation was slow and poorly perfused, even in the normal-looking capillary bed adjacent to the areas of nonperfusion. In severe CRVO, fluorescein angiography showed fluorescein staining of vessel wall, leakage from the disc and retinal vessels, and variable areas of capillary nonperfusion. In such eyes, SRI showed the entire posterior fundus as being extremely low in oxygen saturation. Oxygen satura-
tion was poor even along retinal arteries. Conversely, CRVO was less severe in eyes with higher oxygen saturation levels along retinal arteries. The power of resolution in the present SRI system is somewhat limited, because the coupled charged device built into our system consists of 284 ⫻ 224 pixels. This resolution means that we are measuring an average of oxygen saturation in each pixel that is approximately 40 m square in size when the angle of view is set at 35°. This feature may pose difficulties when studying retinal arteries and veins of small caliber. As a result, retinal arteries and veins could not be illustrated linearly. Other spatial sensitivity concerning the depth in the measurement has to be cleared because our oxygen saturation estimation model is based on a summation of oxygen saturation from hemoglobin absorption along the optical path. Because the light used for the measurement is between 480 and 600 nm and because retinal pigment epithelium is present, major reflection light for the analysis could be from the retina. Only sufficient infrared light that reaches the choroid through the retinal pigment epithelium could be observed with a monochromatic fundus photograph. Also, the presence of diffuse drusen may interfere with our measurement by blocking the irradiated light. Even so, we still cannot neglect the potential influence of the choroid, which has abundant vessels and fast blood flow. Therefore, in this SRI analysis, we measured the oxygen saturation levels in the fundus instead of in the retina, even though the main source of the information might have come from the retinal circulation. Intraretinal circulation is one of the parameters with which to assess the severity of CRVO using fluorescein angiography. Our SRI system indicates that the degree of ischemia, or decreased oxygen saturation level, may serve as another parameter to evaluate the degree of disturbed circulation in CRVO. Moreover, SRI is noninvasive and repeatable and may prove to be a useful clinical tool in monitoring CRVO and in making decisions to perform timely panretinal photocoagulation in patients with this disease.
References 1. Natural history and clinical management of central retinal vein occlusion. The Central Vein Occlusion Study Group [published erratum appears in Arch Ophthalmol 1997;115:1275]. Arch Ophthalmol 1997;115:486 –91. 2. Brindley GS, Willmer EN. The reflexion of light from the macular and peripheral fundus oculi in man. J Physiol 1952;116:350–6. 3. van Norren D, Tiemeijer LF. Spectral reflectance of the human eye. Vision Res 1989;26:313–20. 4. Delori FC, Pflibsen KP. Spectral reflectance of the human ocular fundus. Appl Opt 1989;28:1061–77.
1525
Ophthalmology Volume 109, Number 8, August 2002 5. Cabib D, Buckwald RA, Garini Y, et al. Spatially resolved Fourier transform spectroscopy (spectral imaging): a powerful tool for quantitative analytical microscopy. In: Farkas DL, Leif RC, Priezzhev AV, et al, eds. Optical Diagnostics of Living Cells and Biofluids. Bellingham, WA: SPIE, 1996; 278 –91. (Proc SPIE; 2678).
6. Lavi M, Milman U, Cabib D, et al. A new compact design interferometer based spectral imaging system for bio-medical applications. In: Cogswell CJ, Conchello J, Lerner JM, et al, eds. Three-dimensional and Multidimensional Microscopy: Image Acquisition and Processing V. Bellingham, WA: SPIE, 1998;313–21. (Proc SPIE; 3261).
Historical Image
Lorgnette, c 1900. In the early 20th century, lorgnettes with very long handles were called “dowagerduchess” types. This English example, made from carved tortoise shell, is enhanced by a gold pin decoration. Published through the courtesy of the Alcon Laboratories Museum of Ophthalmology, The Rosenthal Collection, Fort Worth, Texas.
1526