- Email: [email protected]
A Novel Noninvasive Detection Method for Retinal Nonperfusion Using Confocal Red-free Imaging
A Novel Noninvasive Detection Method for Retinal Nonperfusion Using Confocal Red-free Imaging Yong Un Shin, MD,1 Byung Ro Lee, MD, PhD,1 Sungmin Kim, PhD,2 Won June Lee, MD1 Purpose: To report confocal red-free blue reflectance imaging as a novel, noninvasive imaging modality for the detection of retinal nonperfusion and to compare its effectiveness with that of fluorescein angiography (FA) in diabetic retinopathy (DR) and retinal vein occlusion (RVO). Design: Evaluation of diagnostic technology, retrospective observational case series. Participants: We enrolled 54 eyes of 44 patients with DR or RVO that had definite retinal nonperfusion on FA. Methods: All patients underwent red-free blue reflectance imaging and FA using a confocal scanning laser ophthalmoscope. For all patients, macular and midperipheral retinal nonperfusion were identified on both confocal red-free and corresponding FA images, and were delineated by 2 independent readers. We evaluated the correspondence of the 2 imaging methods by comparing the sizes of the delineated areas and obtaining the overlapping ratio after image processing. Main Outcome Measures: Comparison of size and overlapping correspondence between delineated area of retinal nonperfusion obtained by FA and confocal red-free imaging. Results: Image analysis showed a high correlation (r⬎0.9) in the mean size of retinal nonperfusion between confocal red-free and corresponding FA images with DR or RVO. Reliable agreement between the 2 methods was confirmed by size comparisons (P ⫽ 0.563) and overlapping correspondence (overlapping ratio, 0.76) of the delineated area. Conclusions: Our findings suggest that confocal red-free imaging is a simple, reliable, safe, and noninvasive method for effectively plotting retinal nonperfusion. This procedure, first reported herein, has the potential to be used for the noninvasive detection and quantification of retinal nonperfusion in screening, initial evaluation, treatment, and follow-up of progressive ischemic retinopathy such as DR and RVO. Financial Disclosure(s): Proprietary or commercial disclosure may be found after the references. Ophthalmology 2012;119:1447–1454 © 2012 by the American Academy of Ophthalmology.
Diabetic retinopathy (DR) is the most common retinal vascular disease and one of the leading causes of blindness worldwide.1 Retinal vein occlusion (RVO), the second most common retinal vascular disease after DR,2,3 also leads to visual impairment through pathologic changes similar to those observed in DR. The important clinical issues of retinal vascular diseases are the development of retinal nonperfusion and ischemic maculopathy, both of which can affect vision. In most retinal vascular diseases, the area of retinal nonperfusion develops secondary to vascular occlusion. Subsequently, ischemic retina induces the release of vascular endothelial growth factor; a high vitreous level of vascular endothelial growth factor can cause complications that influence vision, such as macular edema, neovascularization, and ischemic maculopathy.2–7 In retinal vascular diseases including DR and RVO, the distribution of capillary nonperfusion is related to the severity and progression of the disease.5,8 –12 However, a more accurate understanding of nonperfusion status for the evaluation of disease © 2012 by the American Academy of Ophthalmology Published by Elsevier Inc.
severity is critical for the development of effective treatment plans and the prediction of prognosis. In addition, laser photocoagulation, the main treatment used to prevent neovascularization in retinal vascular diseases, is applied to the nonperfusion area.6,13 Periodic evaluation of nonperfusion status is necessary to determine whether the disease has progressed and if additional treatment is needed. Currently, fluorescein angiography (FA) is the gold standard method to detect areas of retinal nonperfusion. However, FA requires invasive intravenous dye injection. Fluorescein dye can cause side effects that range from a mild allergic reaction to rare but severe and life-threatening shock. The procedure is also time consuming; it requires 7 to 8 minutes to the end of the late phase; cost also prohibits its repeated use.14 Therefore, a simple, noninvasive method has been highly sought after to recognize areas of retinal nonperfusion. Several studies have assessed the nonperfusion status of the retina noninvasively by various techniques.15–22 HowISSN 0161-6420/12/$–see front matter doi:10.1016/j.ophtha.2012.01.036
1447
Ophthalmology Volume 119, Number 7, July 2012 ever, these techniques remain experimental and less effective. Retinal oximetry, which noninvasively measures the oxygen saturation of retinal vessels, has been studied to assess oxygen delivery to retinal tissues via spectrophotometry, phosphorescence imaging, and laser Doppler.15,23–26 These techniques are limited, however, because they provide only information regarding the oxygen tension of retinal vessels rather than retinal tissue perfusion status. The introduction of oral FA with the confocal scanning laser ophthalmoscope (cSLO) provides an FA image with the resolution of conventional intravenous FA; however, oral FA with cSLO is limited to detecting retinal nonperfusion because it only provides late-phase images, which show profuse dye leakage from peripheral neovascular complexes.21,22 Conventional red-free imaging captured with a fundus camera, which is usually performed after an FA examination, has been used for the detection of retinal nerve fiber layer defects or epiretinal membranes, which are often not easily visible using fundus photography, but is limited owing to its low contrast and resolution. Recently, with the advent of a commercially available cSLO,27 the quality and contrast of images is improving, which increases the likelihood of finding a new approach to diagnose retinal pathologies. In this study, confocal red-free imaging using blue reflectance was performed to visualize the area of retinal nonperfusion. This study reports confocal red-free imaging with a blue wavelength of light as a novel, noninvasive imaging modality for the detection of retinal nonperfusion areas and compares the effectiveness of this confocal redfree imaging with FA in DR and RVO.
Methods All subjects were enrolled after a review of the medical records of patients with DR and RVO in Hanyang University Medical Center. Data were collected from March 2011 to May 2011. Of all reviewed records, the only cases eligible were those in which the FA images of patients had definite retinal nonperfusion. This study was approved by the Institutional Review Board of Hanyang University Medical Center and followed the tenets of the Declaration of Helsinki.
Patients and Imaging Study A total of 79 eyes of 58 patients with DR or RVO were screened for this study. The stage of DR of the enrolled patients ranged from severe, nonproliferative DR to early proliferative DR. All patients involved in this study underwent red-free imaging and FA by cSLO (F-10; Nidek, Gamagori, Japan) on the same day, as well as comprehensive ophthalmologic examinations, including fundus photography and spectral-domain optical coherence tomography. Before performing FA, we obtained confocal red-free images with blue reflectance (490 nm) of the midperipheral retina as well as the area of the macula. In the DR cases, the midperipheral fundus images were taken based on an Early Treatment of Diabetic Retinopathy Study 7-standard field protocol28,29 at a 60-degree angle and a macular image subtended at an angle of 40 degrees, which applied to both the confocal red-free imaging and the FA examination. For RVO, the area of the macula and the midperipheral fundus adjacent to the occluded vessel were captured by both
1448
methods. We performed FA in the standard manner by rapidly injecting 5 ml of 10% sodium fluorescein into the antecubital vein. All images were obtained by a single, well-trained technician who maintained the same imaging protocol for all patients.
Image Selection and Comparison Image Selection. The FA images of the arteriovenous phase and the venous phase were selected for the assessment of the macula area and the midperipheral fundus, respectively. All images were interpreted by a single retinal specialist who evaluated them to determine whether or not they could be included in this study. There was no artificial manipulation for better detection of retinal nonperfusion, and only raw images were used. The following 2 criteria were used to select images for current study: (1) The enrolled eye should have only a single retinal disease (either DR or RVO) based on the FA examination; and (2) only FA images that definitely showed the area of nonperfusion were chosen. If the quality of images was too poor to detect the area of nonperfusion in the FA owing to media opacity, those images were excluded. After the selection of FA images, confocal red-free images that corresponded with the areas of FA images that demonstrated retinal nonperfusion were selected. Image Processing. For quantitative analysis, we carried out 2-staged image processing. Firstly, 2 masked readers manually outlined the area of nonperfusion of each image obtained by both imaging techniques. Then, the confocal red-free image was overlaid on the FA image with the help of commercially available software to create one registered image by matching the morphology of the vascular pattern at each image (MatLab2010a; The Math Works Inc.; Fig 1). Image Analysis for Comparison. The registered images were imported into the ImageJ program (software version 1.44p, National Institutes of Health, Bethesda, MD) to measure the size of the nonperfusion, which was expressed in pixels. The first outcome in the current study was to obtain the size of the delineated retinal nonperfusion of both confocal red-free and FA images and then to compare them. The second outcome was to calculate the overlapping ratio, the so-called Jaccard Index, to evaluate the overlapping correspondence between the 2 methods, which was used for image comparison in another report.30 The overlapping ratio is defined as the ratio of the shared area (intersection) to the combined area of retinal nonperfusion (union) observed by each imaging method using a registered image. The closer the value of this ratio is to 1, the more the 2 images overlap. To determine the overlapping ratio, the same masked readers delineated the areas of union and intersection of retinal nonperfusion, respectively, with the registered images made by second stage image processing. Then these areas were measured using the ImageJ program, and the overlapping ratio was calculated. All values measured by 2 examiners were averaged before our image analysis.
Statistics Interobserver repeatability was examined by calculating the interclass correlation (ICC). The independent t test and correlation analyses were undertaken to compare the extent of nonperfusion between the 2 methods. For all tests, a P⬍0.05 was considered significant.
Results Of all the reviewed records, we included 28 eyes (18 subjects) with DR and 26 eyes (26 subjects) with RVO in this study after the
Shin et al 䡠 Noninvasive Detection of Retinal Nonperfusion
Figure 1. Image processing for quantitative comparisons between confocal red-free and fluorescein angiographic (FA) images. The leftmost upper and lower images were captured by confocal red-free imaging and FA, respectively. The first step of image processing was the delineation of retinal nonperfusion shown in confocal red-free and FA images by 2 independent experts. After that, 2 images were registered using MatLab software. Using these registered images, the size of retinal nonperfusion shown by each imaging technique and the size of their union and intersection for calculating the overlapping ratio were obtained via Image J software.
image selection process. The images of 15 eyes were excluded because no definite retinal nonperfusion was observed on FA. Others (10 eyes) were also excluded owing to poor quality of image in the confocal red-free imaging although the FA demonstrated retinal nonperfusion. The basic clinical information from each study is shown in Table 1. The mean size of the delineated areas of each variable (FA, confocal red-free imaging, their union and intersection) measured by 2 experts is shown in Table 2. There were no differences between the delineations of the 2 experts (r⬎0.8 for all variables; P⬍0.001, intraclass coefficient). The mean sizes of retinal nonperfusion of the FA and confocal red-free images were 141 264.21 and 151 340.46 pixels, respectively, and Table 1. Clinical Characteristics Characteristics
Value
Number of eyes Age (yrs) Refraction (SE), diopters Diagnosis (eyes) DR Mild to moderate NPDR Severe NPDR Very severe NPDR Early PDR High-risk PDR RVO BRVO CRVO
54 57.0⫾10.73 ⫺0.66⫾1.64 * 28 0 11 10 7 0 26 23 3
BRVO ⫽ branch retinal vein occlusion; CRVO ⫽ central retinal vein occlusion; DR ⫽ diabetic retinopathy; NPDR ⫽ nonproliferative diabetic retinopathy; PDR ⫽ proliferative diabetic retinopathy; RVO ⫽ retinal vein occlusion; SE ⫽ spherical equivalent.
difference was not significant (P ⫽ 0.563, independent t test). The correlation coefficient between the size of retinal nonperfusion of the 2 imaging techniques was 0.973 (P⬍0.001, Spearman correlation coefficient). Figure 2 shows the agreement of confocal red-free imaging and FA based on the size of retinal nonperfusion by Bland–Altman plot. Most of the plots were present within 2 standard deviations of the mean area difference between the 2 methods. The overall overlapping ratio between 2 imaging methods was 0.76 (Figs 3 and 4). A difference in size and overlapping ratio according to disease was not observed.
Discussion Visualization of Retinal Nonperfusion by Confocal Red-free Imaging With cSLO, high-resolution and contrast images of specific planes in the fundus can be obtained owing to its confocal nature and the use of laser as a light source.27 We also used cSLO to take advantage of the increased reflectance caused by its confocal nature. Fundus cameras were less suited for the current study because the light scattered by other ocular structures (such as the lens and vitreous) dilutes the reflected light of the retina, which produces poor-quality images. This study demonstrated that retinal nonperfusion shown in FA could be visualized by confocal red-free imaging. Without adjusting the images, 2 experts were able to use the raw confocal red-free images to detect retinal nonperfusion. This correspondence was reinforced by the similar size of delineated retinal nonperfusion and the high overlapping ratio between the 2 imaging techniques. Retinal imaging using green reflectance as well as blue reflectance is among the red-free imaging modalities. We
1449
Ophthalmology Volume 119, Number 7, July 2012 Table 2. Mean Size and Overlapping Ratio of Delineated Retinal Nonperfusion Obtained by Fluorescein Angiography and Red-free Imaging Size of Delineated Area
DR RVO Total
Overlapping Correspondence
FA (Pixels ⫾ SD)
RF (Pixels ⫾ SD)
P Value*
FA 艚 RF (Pixels ⫾ SD)
FA 艛 RF (Pixels ⫾ SD)
Overlapping Ratio
135 190.36⫾92 736.32 148 072.24⫾966 647.02 141 264.21⫾92 034.37
146 825.44⫾103 784.75 156 511.55⫾104 319.98 151 340.46⫾101 339.61
0.631 0.541 0.563
97 518.31⫾82 139.54 128 279.37⫾72 464.13 111 500.61⫾78 253.21
133 066.53⫾115 741.30 163 629.60⫾90 761.48 146 958.83⫾104 687.41
0.74⫾0.11 0.81⫾0.12 0.76⫾0.12
DR ⫽ diabetic retinopathy; FA ⫽ Fluorescein angiograph; FA 艚 RF ⫽ the intersection area of retinal nonperfusion in FA and red-free imaging; FA 艛 RF ⫽ the union area of retinal nonperfusion shown in FA and red-free imaging; RF ⫽ red-free imaging; RVO ⫽ retinal vein occlusion; SD ⫽ standard deviation. All values were measured by 2 independent readers and were averaged for image analysis. The overlapping ratio is defined as the ratio of the intersection of 2 methods and their union. *The difference of the size of delineated area assessed by the independent t test.
investigated several cases using confocal red-free green reflectance imaging to detect retinal nonperfusion and to compare the results with those acquired by blue reflectance imaging. The demarcation line of retinal nonperfusion of confocal red-free green reflectance images is less definite than that of blue reflectance images. The absorption coefficient of hemoglobin increases at the movement of the spectrum from blue to yellow, and that of melanin decreases31,32; therefore, a green reflectance image may have lower contrast than a blue reflectance image. However, green reflectance may be useful for detecting retinal nonperfusion in cases that exhibit the same media opacity that produces poor-quality images with blue reflectance despite its low contrast. Further studies are required to evaluate the usefulness of confocal red-free green reflectance imaging.
Comparisons between Confocal Red-free Imaging and FA In this study, comparisons of area of demarcation lines showed that confocal red-free imaging is comparable with FA imaging for detecting retinal nonperfusion. The interobserver reliability of 2 studies as tested by the interclass coefficient was very high, which indicated that the process of delineation was reproducible and also that the boundaries of retinal nonperfusion that were captured by confocal red-
free imaging were observable. An overlapping ratio of ⬍0.75 was calculated, indicating high correspondence of the morphology between different images.
Retinal Nonperfusion Seems Dark on Confocal Red-free Imaging We hypothesized that retinal nonperfusion on confocal red-free blue reflectance imaging seems to be darker than the surrounding retinal tissue owing to differences in the absorption coefficients of hemoglobin and retinal chromophores. The absorption coefficient of tissue is proportional to optical density, which is calculated in tone pixels of retinal spectral images. If the absorption coefficient of the tissue is high, the captured retinal image seems darker owing to the high optical density. The retinal reflectance is determined by the combination of the absorption coefficient of hemoglobin and retinal tissue.33 The absorption coefficient of melanin, which is the major component of the retinal pigment epithelium, is very high in the blue region of the spectrum, although it quickly drops off at longer wavelengths.31 Normal perfused retinal tissue appears as a whitish area because oxygenated hemoglobin or deoxygenated hemoglobin partially reflects the blue wavelength of light. Their absorption coefficients are ⬍10% of melanin at 490 nm. However, if there was no blood flow in the retinal tissue, the avascular retina had a higher optical density than the surrounding retina because only retinal pigmentation, which absorbs most of the blue wavelength of light, influences retinal reflection.33 Therefore, if retinal blood flow is disturbed by retinal vascular diseases, the amount of hemoglobin is decreased, which makes the avascular retina seem dark owing to having a relatively greater absorption of the blue wavelength of light by retinal pigments. This phenomenon also occurs in the fovea avascular zone.34
Advantages and Limitations of Confocal Scanning Laser Ophthalmoscope Red-free Imaging Figure 2. Bland–Altman plot for comparison between fluorescein angiography (FA) and confocal red-free imaging (RF). This plot shows the agreement of the size of retinal nonperfusion between 2 imaging technique. SD ⫽ standard deviation.
1450
We hypothesize that confocal red-free blue reflectance imaging can be applied to various clinical situations because most cSLO devices have a red-free imaging mode. This technique may be useful for the screening and initial eval-
Shin et al 䡠 Noninvasive Detection of Retinal Nonperfusion
Figure 3. Case series of diabetic retinopathy (DR) with definite retinal nonperfusion. Cases 1 and 2 showed retinal nonperfusion of the posterior pole in confocal red-free imaging (white arrows) and fluorescein angiography (FA). Cases 3 and 4 showed midperipheral nonperfusion in red-free imaging and FA. NPDR ⫽ nonproliferative diabetic retinopathy; OR ⫽ overlapping ratio; PDR ⫽ proliferative diabetic retinopathy.
uation of DR, especially in large-scale health checkups or telemedicine. Until now, DR has been classified and screened by fundus photography, as suggested by the Early Treatment of Diabetic Retinopathy Study group.28,35 Confocal red-free imaging can provide more detailed information about disease status at screening and in the future may also create new classifications of DR, including nonperfusion status. The periodic evaluation of nonperfusion status is potentially possible, which is of great benefit because repeated FA to detect the progression of retinal nonperfusion is not recommended. Thus, confocal
red-free imaging may be the only feasible alternative method for patients who experience serious side effects when given fluorescein. This imaging technique is expected to aid in the decision-making process for the treatment of retinal vascular diseases, especially in cases that require laser therapy. Laser photocoagulation is usually performed in the area of nonperfusion seen on FA image. Confocal red-free imaging may provide the area of nonperfusion that needs further laser treatment without the help of FA (Fig 4, Case 1). Finally, confocal red-free imaging is superior to FA for detecting retinal nonperfu-
1451
Ophthalmology Volume 119, Number 7, July 2012
Figure 4. Case series of retinal vein occlusion (RVO) with definite retinal nonperfusion. Retinal nonperfusion shown in red-free imaging (white arrows) is comparable to fluorescein angiography (FA) in all cases. Case 1 had a history of laser photocoagulation treatment due to the development of new vessels. Six months later, the patient underwent red-free imaging alone at her follow-up evaluation because she refused the FA examination. The extension of retinal nonperfusion was found by confocal red-free imaging, and additional laser treatment was performed. BRVO ⫽ branch retinal vein occlusion; CRVO ⫽ central retinal vein occlusion; OR ⫽ overlapping ratio.
sion in cases with severe neovascularization because the profuse dye leakage from new vessels on mid-phase FA interferes with the plotting of retinal nonperfusion. Although the use of confocal red-free imaging is promising, several issues should be considered when interpreting retinal nonperfusion. Firstly, technicians can influence the demarcation line of retinal nonperfusion by controlling the brightness or contrast while performing the examination. Therefore, maintaining a consistent protocol is necessary at
1452
every examination. Secondly, the short blue wavelength is vulnerable to media opacity, which results in poor image quality. Even though media opacity (like severe cataracts or vitreous haziness) makes it difficult to interpret both FA and confocal red-free images, the image quality of FA with the help of fluorescein dye is superior to that of confocal redfree imaging. Poor-quality images owing to severe media opacity were excluded from this study. Thirdly, other retinal pathologies such as pigmented lesions and retinal hemor-
Shin et al 䡠 Noninvasive Detection of Retinal Nonperfusion rhage appear as dark regions that are actually perfused in FA. Therefore, when interpreting confocal red-free imaging, correlations with fundus photography may be required. Fourthly, this red-free imaging has shortcomings when visualizing retinal nonperfusion that is located in more peripheral areas. In some cases, the confocal red-free imaging showed lighter area in the peripheral retina, which seemed to be perfused, even if corresponding FA actually detected retinal nonperfusion. This phenomenon might be explained by retinal curvature, which leads to out-of-focus images. In the peripheral retina, the more anterior portions are not going to be on the same focal plane as the more posterior portions that are relatively flat because only 1 focal plane can be chosen for confocal imaging (Fig 3, Case 3). Finally, this noninvasive method cannot provide information about the development of new vessels. This limitation means that FA is still the gold standard for the evaluation of retinal vascular diseases. However, confocal red-free imaging is useful as a follow-up examination for monitoring until the next FA examination. The small sample size, retrospective study design, and subjective delineation of retinal nonperfusion are additional limitations of the current study. This study was conducted with the cases with only definite retinal nonperfusion on FA, thereby limiting generalization of this work. To generalize our study, we performed a small pilot study with 10 patients (20 eyes) who had DR (5 patients) and unilateral BRVO (5 patients) to investigate confocal red-free imaging for the detection of retinal nonperfusion according to the status of retinal nonperfusion such as normal perfused cases, questionable nonperfused cases, and definite nonperfused cases. In this pilot study, we chose study subjects first based on confocal red-free images and then red-free images of each group were compared with corresponding FA images (Fig 5, available at http://aaojournal.org). In all cases in the perfused red-free image group, corresponding FA revealed no retinal nonperfusion. In the definite nonperfused red-free image group, retinal nonperfusion was detected in all corresponding FA and correspondence between 2 methods was similar to our original study. In the questionable nonperfused red-free image group, corresponding FA showed retinal nonperfusion, but the size and shape of retinal nonperfusion delineated by masked readers were somewhat different owing to equivocal demarcation between perfused and nonperfused areas in red-free images, which resulted in lower overlapping correspondence than that of the definite nonperfused group (Table 3, available at http://aaojournal.org). Through this pilot study, we found that confocal red-free imaging could detect retinal nonperfusion in cases with it on FA. However, the exact detection of retinal nonperfusion using confocal red-free imaging remains a challenge in questionable nonperfused cases. Based on this pilot study, we plan to conduct a prospective study with large samples. To our knowledge, there have been no reports on the evaluation of midperipheral nonperfusion by confocal redfree imaging. We found that this confocal red-free imaging modality can reliably detect retinal nonperfusion and that red-free imaging is highly correlated with FA. This novel technique for visualizing the nonperfused retina is advanta-
geous because it is simple and safe, and images easy to obtain and are promptly available in current clinical practices. Clinically, this technique has the potential to be used for noninvasively detecting and quantifying retinal nonperfusion in screening, initial evaluation, treatment decisions, and follow-up of progressing ischemic retinopathy such as DR and RVO.
References 1. Fong DS, Aiello L, Gardner TW, et al. Retinopathy in diabetes. Diabetes Care 2004;27(suppl 1):S84 –7. 2. McIntosh RL, Rogers SL, Lim L, et al. Natural history of central retinal vein occlusion: an evidence-based systematic review. Ophthalmology 2010;117:1113–23 e15. 3. Rogers SL, McIntosh RL, Lim L, et al. Natural history of branch retinal vein occlusion: an evidence-based systematic review. Ophthalmology 2010;117:1094 –101 e5. 4. Aiello LP, Avery RL, Arrigg PG, et al. Vascular endothelial growth factor in ocular fluid of patients with diabetic retinopathy and other retinal disorders. N Engl J Med 1994; 331:1480 –7. 5. Boyd SR, Zachary I, Chakravarthy U, et al. Correlation of increased vascular endothelial growth factor with neovascularization and permeability in ischemic central vein occlusion. Arch Ophthalmol 2002;120:1644 –50. 6. Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lancet 2010;376:124 –36. 7. Pe’er J, Folberg R, Itin A, et al. Vascular endothelial growth factor upregulation in human central retinal vein occlusion. Ophthalmology 1998;105:412– 6. 8. Bresnick GH, De Venecia G, Myers FL, et al. Retinal ischemia in diabetic retinopathy. Arch Ophthalmol 1975;93:1300 –10. 9. Chan CK, Ip MS, Vanveldhuisen PC, et al. SCORE Study Report #11 Incidences of Neovascular Events in Eyes with Retinal Vein Occlusion. Ophthalmology 2011;118:1364 –72. 10. Shilling JS, Kohner EM. New vessel formation in retinal branch vein occlusion. Br J Ophthalmol 1976;60:810 –5. 11. Shimizu K, Kobayashi Y, Muraoka K. Midperipheral fundus involvement in diabetic retinopathy. Ophthalmology 1981;88: 601–12. 12. Niki T, Muraoka K, Shimizu K. Distribution of capillary nonperfusion in early-stage diabetic retinopathy. Ophthalmology 1984;91:1431–9. 13. Wong TY, Scott IU. Clinical practice. Retinal-vein occlusion. N Engl J Med 2010;363:2135– 44. 14. Yannuzzi LA, Rohrer KT, Tindel LJ, et al. Fluorescein angiography complication survey. Ophthalmology 1986;93: 611–7. 15. Delori FC. Noninvasive technique for oximetry of blood in retinal vessels. Appl Opt 1988;27:1113–25. 16. Hilmantel G, Applegate RA, van Heuven WA, et al. Entoptic foveal avascular zone measurement and diabetic retinopathy. Optom Vis Sci 1999;76:826 –31. 17. Martin JA, Roorda A. Direct and noninvasive assessment of parafoveal capillary leukocyte velocity. Ophthalmology 2005; 112:2219 –24. 18. Rumsey WL, Vanderkooi JM, Wilson DF. Imaging of phosphorescence: a novel method for measuring oxygen distribution in perfused tissue. Science 1988;241:1649 –51. 19. Tam J, Martin JA, Roorda A. Noninvasive visualization and analysis of parafoveal capillaries in humans. Invest Ophthalmol Vis Sci 2010;51:1691– 8.
1453
Ophthalmology Volume 119, Number 7, July 2012 20. Yoneya S, Saito T, Nishiyama Y, et al. Retinal oxygen saturation levels in patients with central retinal vein occlusion. Ophthalmology 2002;109:1521– 6. 21. Squirrell D, Dinakaran S, Dhingra S, et al. Oral fluorescein angiography with the scanning laser ophthalmoscope in diabetic retinopathy: a case controlled comparison with intravenous fluorescein angiography. Eye 2004;19:411–7. 22. Garcia C. Oral fluorescein angiography with the confocal scanning laser ophthalmoscope. Ophthalmology 1999;106: 1114 – 8. 23. Crittin M, Schmidt H, Riva CE. Hemoglobin oxygen saturation (So2) in the human ocular fundus measured by reflectance oximetry: preliminary data in retinal veins. Klin Monbl Augenheilkd 2002;219:289 –91. 24. Petrig BL, Riva CE, Hayreh SS. Laser Doppler flowmetry and optic nerve head blood flow. Am J Ophthalmol 1999; 127:413–25. 25. Shonat RD, Kight AC. Oxygen tension imaging in the mouse retina. Ann Biomed Eng 2003;31:1084 –96. 26. Shonat RD, Wilson DF, Riva CE, Pawlowski M. Oxygen distribution in the retinal and choroidal vessels of the cat as measured by a new phosphorescence imaging method. Appl Opt 1992;31:3711– 8. 27. Woon WH, Fitzke FW, Bird AC, Marshall J. Confocal imaging of the fundus using a scanning laser ophthalmoscope. Br J Ophthalmol 1992;76:470 – 4. 28. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy Study report number 1. Early
29.
30.
31.
32.
33. 34.
35.
Treatment Diabetic Retinopathy Study research group. Arch Ophthalmol 1985;103:1796 – 806. Li HK, Danis RP, Hubbard LD, et al. Comparability of digital photography with the ETDRS film protocol for evaluation of diabetic retinopathy severity. Invest Ophthalmol Vis Sci 2011; 52:4717–25. Zheng Y, Gandhi JS, Stangos AN, et al. Automated segmentation of foveal avascular zone in fundus fluorescein angiography. Invest Ophthalmol Vis Sci 2010;51:3653–9. Palanker D, Blumerkranz MS, Weiter JJ Retinal laser therapy: biophysics basis and applications. In: Ryan SJ, ed-in-chief; Hinton DR, Schachat AP, eds. Retina. vol. 1. Basic Science and Inherited Retinal Disease, Tumors. 4th ed. Philadelphia: Elsevier Mosby; 2006:541. Peli E, Hedges TR 3rd, McInnes T, et al. Nerve fiber layer photography. A comparative study. Acta Ophthalmol (Copenh) 1987;65:71– 80. Behrendt T, Wilson LA. Spectral reflectance photography of the retina. Am J Ophthalmol 1965;59:1079 – 88. Shin YU, Kim S, Lee BR, et al. Novel noninvasive detection of the fovea avascular zone using confocal red-free imaging in diabetic retinopathy and retinal vein occlusion. Invest Ophthalmol Vis Sci 2012;53:309 –15. Fundus photographic risk factors for progression of diabetic retinopathy. ETDRS report number 12. Early Treatment Diabetic Retinopathy Study Research Group. Ophthalmology 1991;98:823–33.
Footnotes and Financial Disclosures Originally received: September 2, 2011. Final revision: January 19, 2012. Accepted: January 19, 2012. Available online: April 4, 2012.
Financial Disclosures: The authors have made the following disclosures: Byung Ro Lee – Consultant – Nidek, Gamagori, Japan. Manuscript no. 2011-1307.
1
Department of Ophthalmology, College of Medicine, Hanyang University, Seoul, Korea.
2
Department of Biomedical Engineering, Hanyang University, Seoul, Korea.
1454
Correspondence: Byung Ro Lee, MD, Department of Ophthalmology, Hanyang University Hospital, #17 Seongdong-gu, Haengdang-dong, Seoul, 133-792, Korea. E-mail: [email protected].
Diabetic retinopathy (DR) is the most common retinal vascular disease and one of the leading causes of blindness worldwide.1 Retinal vein occlusion (RVO), the second most common retinal vascular disease after DR,2,3 also leads to visual impairment through pathologic changes similar to those observed in DR. The important clinical issues of retinal vascular diseases are the development of retinal nonperfusion and ischemic maculopathy, both of which can affect vision. In most retinal vascular diseases, the area of retinal nonperfusion develops secondary to vascular occlusion. Subsequently, ischemic retina induces the release of vascular endothelial growth factor; a high vitreous level of vascular endothelial growth factor can cause complications that influence vision, such as macular edema, neovascularization, and ischemic maculopathy.2–7 In retinal vascular diseases including DR and RVO, the distribution of capillary nonperfusion is related to the severity and progression of the disease.5,8 –12 However, a more accurate understanding of nonperfusion status for the evaluation of disease © 2012 by the American Academy of Ophthalmology Published by Elsevier Inc.
severity is critical for the development of effective treatment plans and the prediction of prognosis. In addition, laser photocoagulation, the main treatment used to prevent neovascularization in retinal vascular diseases, is applied to the nonperfusion area.6,13 Periodic evaluation of nonperfusion status is necessary to determine whether the disease has progressed and if additional treatment is needed. Currently, fluorescein angiography (FA) is the gold standard method to detect areas of retinal nonperfusion. However, FA requires invasive intravenous dye injection. Fluorescein dye can cause side effects that range from a mild allergic reaction to rare but severe and life-threatening shock. The procedure is also time consuming; it requires 7 to 8 minutes to the end of the late phase; cost also prohibits its repeated use.14 Therefore, a simple, noninvasive method has been highly sought after to recognize areas of retinal nonperfusion. Several studies have assessed the nonperfusion status of the retina noninvasively by various techniques.15–22 HowISSN 0161-6420/12/$–see front matter doi:10.1016/j.ophtha.2012.01.036
1447
Ophthalmology Volume 119, Number 7, July 2012 ever, these techniques remain experimental and less effective. Retinal oximetry, which noninvasively measures the oxygen saturation of retinal vessels, has been studied to assess oxygen delivery to retinal tissues via spectrophotometry, phosphorescence imaging, and laser Doppler.15,23–26 These techniques are limited, however, because they provide only information regarding the oxygen tension of retinal vessels rather than retinal tissue perfusion status. The introduction of oral FA with the confocal scanning laser ophthalmoscope (cSLO) provides an FA image with the resolution of conventional intravenous FA; however, oral FA with cSLO is limited to detecting retinal nonperfusion because it only provides late-phase images, which show profuse dye leakage from peripheral neovascular complexes.21,22 Conventional red-free imaging captured with a fundus camera, which is usually performed after an FA examination, has been used for the detection of retinal nerve fiber layer defects or epiretinal membranes, which are often not easily visible using fundus photography, but is limited owing to its low contrast and resolution. Recently, with the advent of a commercially available cSLO,27 the quality and contrast of images is improving, which increases the likelihood of finding a new approach to diagnose retinal pathologies. In this study, confocal red-free imaging using blue reflectance was performed to visualize the area of retinal nonperfusion. This study reports confocal red-free imaging with a blue wavelength of light as a novel, noninvasive imaging modality for the detection of retinal nonperfusion areas and compares the effectiveness of this confocal redfree imaging with FA in DR and RVO.
Methods All subjects were enrolled after a review of the medical records of patients with DR and RVO in Hanyang University Medical Center. Data were collected from March 2011 to May 2011. Of all reviewed records, the only cases eligible were those in which the FA images of patients had definite retinal nonperfusion. This study was approved by the Institutional Review Board of Hanyang University Medical Center and followed the tenets of the Declaration of Helsinki.
Patients and Imaging Study A total of 79 eyes of 58 patients with DR or RVO were screened for this study. The stage of DR of the enrolled patients ranged from severe, nonproliferative DR to early proliferative DR. All patients involved in this study underwent red-free imaging and FA by cSLO (F-10; Nidek, Gamagori, Japan) on the same day, as well as comprehensive ophthalmologic examinations, including fundus photography and spectral-domain optical coherence tomography. Before performing FA, we obtained confocal red-free images with blue reflectance (490 nm) of the midperipheral retina as well as the area of the macula. In the DR cases, the midperipheral fundus images were taken based on an Early Treatment of Diabetic Retinopathy Study 7-standard field protocol28,29 at a 60-degree angle and a macular image subtended at an angle of 40 degrees, which applied to both the confocal red-free imaging and the FA examination. For RVO, the area of the macula and the midperipheral fundus adjacent to the occluded vessel were captured by both
1448
methods. We performed FA in the standard manner by rapidly injecting 5 ml of 10% sodium fluorescein into the antecubital vein. All images were obtained by a single, well-trained technician who maintained the same imaging protocol for all patients.
Image Selection and Comparison Image Selection. The FA images of the arteriovenous phase and the venous phase were selected for the assessment of the macula area and the midperipheral fundus, respectively. All images were interpreted by a single retinal specialist who evaluated them to determine whether or not they could be included in this study. There was no artificial manipulation for better detection of retinal nonperfusion, and only raw images were used. The following 2 criteria were used to select images for current study: (1) The enrolled eye should have only a single retinal disease (either DR or RVO) based on the FA examination; and (2) only FA images that definitely showed the area of nonperfusion were chosen. If the quality of images was too poor to detect the area of nonperfusion in the FA owing to media opacity, those images were excluded. After the selection of FA images, confocal red-free images that corresponded with the areas of FA images that demonstrated retinal nonperfusion were selected. Image Processing. For quantitative analysis, we carried out 2-staged image processing. Firstly, 2 masked readers manually outlined the area of nonperfusion of each image obtained by both imaging techniques. Then, the confocal red-free image was overlaid on the FA image with the help of commercially available software to create one registered image by matching the morphology of the vascular pattern at each image (MatLab2010a; The Math Works Inc.; Fig 1). Image Analysis for Comparison. The registered images were imported into the ImageJ program (software version 1.44p, National Institutes of Health, Bethesda, MD) to measure the size of the nonperfusion, which was expressed in pixels. The first outcome in the current study was to obtain the size of the delineated retinal nonperfusion of both confocal red-free and FA images and then to compare them. The second outcome was to calculate the overlapping ratio, the so-called Jaccard Index, to evaluate the overlapping correspondence between the 2 methods, which was used for image comparison in another report.30 The overlapping ratio is defined as the ratio of the shared area (intersection) to the combined area of retinal nonperfusion (union) observed by each imaging method using a registered image. The closer the value of this ratio is to 1, the more the 2 images overlap. To determine the overlapping ratio, the same masked readers delineated the areas of union and intersection of retinal nonperfusion, respectively, with the registered images made by second stage image processing. Then these areas were measured using the ImageJ program, and the overlapping ratio was calculated. All values measured by 2 examiners were averaged before our image analysis.
Statistics Interobserver repeatability was examined by calculating the interclass correlation (ICC). The independent t test and correlation analyses were undertaken to compare the extent of nonperfusion between the 2 methods. For all tests, a P⬍0.05 was considered significant.
Results Of all the reviewed records, we included 28 eyes (18 subjects) with DR and 26 eyes (26 subjects) with RVO in this study after the
Shin et al 䡠 Noninvasive Detection of Retinal Nonperfusion
Figure 1. Image processing for quantitative comparisons between confocal red-free and fluorescein angiographic (FA) images. The leftmost upper and lower images were captured by confocal red-free imaging and FA, respectively. The first step of image processing was the delineation of retinal nonperfusion shown in confocal red-free and FA images by 2 independent experts. After that, 2 images were registered using MatLab software. Using these registered images, the size of retinal nonperfusion shown by each imaging technique and the size of their union and intersection for calculating the overlapping ratio were obtained via Image J software.
image selection process. The images of 15 eyes were excluded because no definite retinal nonperfusion was observed on FA. Others (10 eyes) were also excluded owing to poor quality of image in the confocal red-free imaging although the FA demonstrated retinal nonperfusion. The basic clinical information from each study is shown in Table 1. The mean size of the delineated areas of each variable (FA, confocal red-free imaging, their union and intersection) measured by 2 experts is shown in Table 2. There were no differences between the delineations of the 2 experts (r⬎0.8 for all variables; P⬍0.001, intraclass coefficient). The mean sizes of retinal nonperfusion of the FA and confocal red-free images were 141 264.21 and 151 340.46 pixels, respectively, and Table 1. Clinical Characteristics Characteristics
Value
Number of eyes Age (yrs) Refraction (SE), diopters Diagnosis (eyes) DR Mild to moderate NPDR Severe NPDR Very severe NPDR Early PDR High-risk PDR RVO BRVO CRVO
54 57.0⫾10.73 ⫺0.66⫾1.64 * 28 0 11 10 7 0 26 23 3
BRVO ⫽ branch retinal vein occlusion; CRVO ⫽ central retinal vein occlusion; DR ⫽ diabetic retinopathy; NPDR ⫽ nonproliferative diabetic retinopathy; PDR ⫽ proliferative diabetic retinopathy; RVO ⫽ retinal vein occlusion; SE ⫽ spherical equivalent.
difference was not significant (P ⫽ 0.563, independent t test). The correlation coefficient between the size of retinal nonperfusion of the 2 imaging techniques was 0.973 (P⬍0.001, Spearman correlation coefficient). Figure 2 shows the agreement of confocal red-free imaging and FA based on the size of retinal nonperfusion by Bland–Altman plot. Most of the plots were present within 2 standard deviations of the mean area difference between the 2 methods. The overall overlapping ratio between 2 imaging methods was 0.76 (Figs 3 and 4). A difference in size and overlapping ratio according to disease was not observed.
Discussion Visualization of Retinal Nonperfusion by Confocal Red-free Imaging With cSLO, high-resolution and contrast images of specific planes in the fundus can be obtained owing to its confocal nature and the use of laser as a light source.27 We also used cSLO to take advantage of the increased reflectance caused by its confocal nature. Fundus cameras were less suited for the current study because the light scattered by other ocular structures (such as the lens and vitreous) dilutes the reflected light of the retina, which produces poor-quality images. This study demonstrated that retinal nonperfusion shown in FA could be visualized by confocal red-free imaging. Without adjusting the images, 2 experts were able to use the raw confocal red-free images to detect retinal nonperfusion. This correspondence was reinforced by the similar size of delineated retinal nonperfusion and the high overlapping ratio between the 2 imaging techniques. Retinal imaging using green reflectance as well as blue reflectance is among the red-free imaging modalities. We
1449
Ophthalmology Volume 119, Number 7, July 2012 Table 2. Mean Size and Overlapping Ratio of Delineated Retinal Nonperfusion Obtained by Fluorescein Angiography and Red-free Imaging Size of Delineated Area
DR RVO Total
Overlapping Correspondence
FA (Pixels ⫾ SD)
RF (Pixels ⫾ SD)
P Value*
FA 艚 RF (Pixels ⫾ SD)
FA 艛 RF (Pixels ⫾ SD)
Overlapping Ratio
135 190.36⫾92 736.32 148 072.24⫾966 647.02 141 264.21⫾92 034.37
146 825.44⫾103 784.75 156 511.55⫾104 319.98 151 340.46⫾101 339.61
0.631 0.541 0.563
97 518.31⫾82 139.54 128 279.37⫾72 464.13 111 500.61⫾78 253.21
133 066.53⫾115 741.30 163 629.60⫾90 761.48 146 958.83⫾104 687.41
0.74⫾0.11 0.81⫾0.12 0.76⫾0.12
DR ⫽ diabetic retinopathy; FA ⫽ Fluorescein angiograph; FA 艚 RF ⫽ the intersection area of retinal nonperfusion in FA and red-free imaging; FA 艛 RF ⫽ the union area of retinal nonperfusion shown in FA and red-free imaging; RF ⫽ red-free imaging; RVO ⫽ retinal vein occlusion; SD ⫽ standard deviation. All values were measured by 2 independent readers and were averaged for image analysis. The overlapping ratio is defined as the ratio of the intersection of 2 methods and their union. *The difference of the size of delineated area assessed by the independent t test.
investigated several cases using confocal red-free green reflectance imaging to detect retinal nonperfusion and to compare the results with those acquired by blue reflectance imaging. The demarcation line of retinal nonperfusion of confocal red-free green reflectance images is less definite than that of blue reflectance images. The absorption coefficient of hemoglobin increases at the movement of the spectrum from blue to yellow, and that of melanin decreases31,32; therefore, a green reflectance image may have lower contrast than a blue reflectance image. However, green reflectance may be useful for detecting retinal nonperfusion in cases that exhibit the same media opacity that produces poor-quality images with blue reflectance despite its low contrast. Further studies are required to evaluate the usefulness of confocal red-free green reflectance imaging.
Comparisons between Confocal Red-free Imaging and FA In this study, comparisons of area of demarcation lines showed that confocal red-free imaging is comparable with FA imaging for detecting retinal nonperfusion. The interobserver reliability of 2 studies as tested by the interclass coefficient was very high, which indicated that the process of delineation was reproducible and also that the boundaries of retinal nonperfusion that were captured by confocal red-
free imaging were observable. An overlapping ratio of ⬍0.75 was calculated, indicating high correspondence of the morphology between different images.
Retinal Nonperfusion Seems Dark on Confocal Red-free Imaging We hypothesized that retinal nonperfusion on confocal red-free blue reflectance imaging seems to be darker than the surrounding retinal tissue owing to differences in the absorption coefficients of hemoglobin and retinal chromophores. The absorption coefficient of tissue is proportional to optical density, which is calculated in tone pixels of retinal spectral images. If the absorption coefficient of the tissue is high, the captured retinal image seems darker owing to the high optical density. The retinal reflectance is determined by the combination of the absorption coefficient of hemoglobin and retinal tissue.33 The absorption coefficient of melanin, which is the major component of the retinal pigment epithelium, is very high in the blue region of the spectrum, although it quickly drops off at longer wavelengths.31 Normal perfused retinal tissue appears as a whitish area because oxygenated hemoglobin or deoxygenated hemoglobin partially reflects the blue wavelength of light. Their absorption coefficients are ⬍10% of melanin at 490 nm. However, if there was no blood flow in the retinal tissue, the avascular retina had a higher optical density than the surrounding retina because only retinal pigmentation, which absorbs most of the blue wavelength of light, influences retinal reflection.33 Therefore, if retinal blood flow is disturbed by retinal vascular diseases, the amount of hemoglobin is decreased, which makes the avascular retina seem dark owing to having a relatively greater absorption of the blue wavelength of light by retinal pigments. This phenomenon also occurs in the fovea avascular zone.34
Advantages and Limitations of Confocal Scanning Laser Ophthalmoscope Red-free Imaging Figure 2. Bland–Altman plot for comparison between fluorescein angiography (FA) and confocal red-free imaging (RF). This plot shows the agreement of the size of retinal nonperfusion between 2 imaging technique. SD ⫽ standard deviation.
1450
We hypothesize that confocal red-free blue reflectance imaging can be applied to various clinical situations because most cSLO devices have a red-free imaging mode. This technique may be useful for the screening and initial eval-
Shin et al 䡠 Noninvasive Detection of Retinal Nonperfusion
Figure 3. Case series of diabetic retinopathy (DR) with definite retinal nonperfusion. Cases 1 and 2 showed retinal nonperfusion of the posterior pole in confocal red-free imaging (white arrows) and fluorescein angiography (FA). Cases 3 and 4 showed midperipheral nonperfusion in red-free imaging and FA. NPDR ⫽ nonproliferative diabetic retinopathy; OR ⫽ overlapping ratio; PDR ⫽ proliferative diabetic retinopathy.
uation of DR, especially in large-scale health checkups or telemedicine. Until now, DR has been classified and screened by fundus photography, as suggested by the Early Treatment of Diabetic Retinopathy Study group.28,35 Confocal red-free imaging can provide more detailed information about disease status at screening and in the future may also create new classifications of DR, including nonperfusion status. The periodic evaluation of nonperfusion status is potentially possible, which is of great benefit because repeated FA to detect the progression of retinal nonperfusion is not recommended. Thus, confocal
red-free imaging may be the only feasible alternative method for patients who experience serious side effects when given fluorescein. This imaging technique is expected to aid in the decision-making process for the treatment of retinal vascular diseases, especially in cases that require laser therapy. Laser photocoagulation is usually performed in the area of nonperfusion seen on FA image. Confocal red-free imaging may provide the area of nonperfusion that needs further laser treatment without the help of FA (Fig 4, Case 1). Finally, confocal red-free imaging is superior to FA for detecting retinal nonperfu-
1451
Ophthalmology Volume 119, Number 7, July 2012
Figure 4. Case series of retinal vein occlusion (RVO) with definite retinal nonperfusion. Retinal nonperfusion shown in red-free imaging (white arrows) is comparable to fluorescein angiography (FA) in all cases. Case 1 had a history of laser photocoagulation treatment due to the development of new vessels. Six months later, the patient underwent red-free imaging alone at her follow-up evaluation because she refused the FA examination. The extension of retinal nonperfusion was found by confocal red-free imaging, and additional laser treatment was performed. BRVO ⫽ branch retinal vein occlusion; CRVO ⫽ central retinal vein occlusion; OR ⫽ overlapping ratio.
sion in cases with severe neovascularization because the profuse dye leakage from new vessels on mid-phase FA interferes with the plotting of retinal nonperfusion. Although the use of confocal red-free imaging is promising, several issues should be considered when interpreting retinal nonperfusion. Firstly, technicians can influence the demarcation line of retinal nonperfusion by controlling the brightness or contrast while performing the examination. Therefore, maintaining a consistent protocol is necessary at
1452
every examination. Secondly, the short blue wavelength is vulnerable to media opacity, which results in poor image quality. Even though media opacity (like severe cataracts or vitreous haziness) makes it difficult to interpret both FA and confocal red-free images, the image quality of FA with the help of fluorescein dye is superior to that of confocal redfree imaging. Poor-quality images owing to severe media opacity were excluded from this study. Thirdly, other retinal pathologies such as pigmented lesions and retinal hemor-
Shin et al 䡠 Noninvasive Detection of Retinal Nonperfusion rhage appear as dark regions that are actually perfused in FA. Therefore, when interpreting confocal red-free imaging, correlations with fundus photography may be required. Fourthly, this red-free imaging has shortcomings when visualizing retinal nonperfusion that is located in more peripheral areas. In some cases, the confocal red-free imaging showed lighter area in the peripheral retina, which seemed to be perfused, even if corresponding FA actually detected retinal nonperfusion. This phenomenon might be explained by retinal curvature, which leads to out-of-focus images. In the peripheral retina, the more anterior portions are not going to be on the same focal plane as the more posterior portions that are relatively flat because only 1 focal plane can be chosen for confocal imaging (Fig 3, Case 3). Finally, this noninvasive method cannot provide information about the development of new vessels. This limitation means that FA is still the gold standard for the evaluation of retinal vascular diseases. However, confocal red-free imaging is useful as a follow-up examination for monitoring until the next FA examination. The small sample size, retrospective study design, and subjective delineation of retinal nonperfusion are additional limitations of the current study. This study was conducted with the cases with only definite retinal nonperfusion on FA, thereby limiting generalization of this work. To generalize our study, we performed a small pilot study with 10 patients (20 eyes) who had DR (5 patients) and unilateral BRVO (5 patients) to investigate confocal red-free imaging for the detection of retinal nonperfusion according to the status of retinal nonperfusion such as normal perfused cases, questionable nonperfused cases, and definite nonperfused cases. In this pilot study, we chose study subjects first based on confocal red-free images and then red-free images of each group were compared with corresponding FA images (Fig 5, available at http://aaojournal.org). In all cases in the perfused red-free image group, corresponding FA revealed no retinal nonperfusion. In the definite nonperfused red-free image group, retinal nonperfusion was detected in all corresponding FA and correspondence between 2 methods was similar to our original study. In the questionable nonperfused red-free image group, corresponding FA showed retinal nonperfusion, but the size and shape of retinal nonperfusion delineated by masked readers were somewhat different owing to equivocal demarcation between perfused and nonperfused areas in red-free images, which resulted in lower overlapping correspondence than that of the definite nonperfused group (Table 3, available at http://aaojournal.org). Through this pilot study, we found that confocal red-free imaging could detect retinal nonperfusion in cases with it on FA. However, the exact detection of retinal nonperfusion using confocal red-free imaging remains a challenge in questionable nonperfused cases. Based on this pilot study, we plan to conduct a prospective study with large samples. To our knowledge, there have been no reports on the evaluation of midperipheral nonperfusion by confocal redfree imaging. We found that this confocal red-free imaging modality can reliably detect retinal nonperfusion and that red-free imaging is highly correlated with FA. This novel technique for visualizing the nonperfused retina is advanta-
geous because it is simple and safe, and images easy to obtain and are promptly available in current clinical practices. Clinically, this technique has the potential to be used for noninvasively detecting and quantifying retinal nonperfusion in screening, initial evaluation, treatment decisions, and follow-up of progressing ischemic retinopathy such as DR and RVO.
References 1. Fong DS, Aiello L, Gardner TW, et al. Retinopathy in diabetes. Diabetes Care 2004;27(suppl 1):S84 –7. 2. McIntosh RL, Rogers SL, Lim L, et al. Natural history of central retinal vein occlusion: an evidence-based systematic review. Ophthalmology 2010;117:1113–23 e15. 3. Rogers SL, McIntosh RL, Lim L, et al. Natural history of branch retinal vein occlusion: an evidence-based systematic review. Ophthalmology 2010;117:1094 –101 e5. 4. Aiello LP, Avery RL, Arrigg PG, et al. Vascular endothelial growth factor in ocular fluid of patients with diabetic retinopathy and other retinal disorders. N Engl J Med 1994; 331:1480 –7. 5. Boyd SR, Zachary I, Chakravarthy U, et al. Correlation of increased vascular endothelial growth factor with neovascularization and permeability in ischemic central vein occlusion. Arch Ophthalmol 2002;120:1644 –50. 6. Cheung N, Mitchell P, Wong TY. Diabetic retinopathy. Lancet 2010;376:124 –36. 7. Pe’er J, Folberg R, Itin A, et al. Vascular endothelial growth factor upregulation in human central retinal vein occlusion. Ophthalmology 1998;105:412– 6. 8. Bresnick GH, De Venecia G, Myers FL, et al. Retinal ischemia in diabetic retinopathy. Arch Ophthalmol 1975;93:1300 –10. 9. Chan CK, Ip MS, Vanveldhuisen PC, et al. SCORE Study Report #11 Incidences of Neovascular Events in Eyes with Retinal Vein Occlusion. Ophthalmology 2011;118:1364 –72. 10. Shilling JS, Kohner EM. New vessel formation in retinal branch vein occlusion. Br J Ophthalmol 1976;60:810 –5. 11. Shimizu K, Kobayashi Y, Muraoka K. Midperipheral fundus involvement in diabetic retinopathy. Ophthalmology 1981;88: 601–12. 12. Niki T, Muraoka K, Shimizu K. Distribution of capillary nonperfusion in early-stage diabetic retinopathy. Ophthalmology 1984;91:1431–9. 13. Wong TY, Scott IU. Clinical practice. Retinal-vein occlusion. N Engl J Med 2010;363:2135– 44. 14. Yannuzzi LA, Rohrer KT, Tindel LJ, et al. Fluorescein angiography complication survey. Ophthalmology 1986;93: 611–7. 15. Delori FC. Noninvasive technique for oximetry of blood in retinal vessels. Appl Opt 1988;27:1113–25. 16. Hilmantel G, Applegate RA, van Heuven WA, et al. Entoptic foveal avascular zone measurement and diabetic retinopathy. Optom Vis Sci 1999;76:826 –31. 17. Martin JA, Roorda A. Direct and noninvasive assessment of parafoveal capillary leukocyte velocity. Ophthalmology 2005; 112:2219 –24. 18. Rumsey WL, Vanderkooi JM, Wilson DF. Imaging of phosphorescence: a novel method for measuring oxygen distribution in perfused tissue. Science 1988;241:1649 –51. 19. Tam J, Martin JA, Roorda A. Noninvasive visualization and analysis of parafoveal capillaries in humans. Invest Ophthalmol Vis Sci 2010;51:1691– 8.
1453
Ophthalmology Volume 119, Number 7, July 2012 20. Yoneya S, Saito T, Nishiyama Y, et al. Retinal oxygen saturation levels in patients with central retinal vein occlusion. Ophthalmology 2002;109:1521– 6. 21. Squirrell D, Dinakaran S, Dhingra S, et al. Oral fluorescein angiography with the scanning laser ophthalmoscope in diabetic retinopathy: a case controlled comparison with intravenous fluorescein angiography. Eye 2004;19:411–7. 22. Garcia C. Oral fluorescein angiography with the confocal scanning laser ophthalmoscope. Ophthalmology 1999;106: 1114 – 8. 23. Crittin M, Schmidt H, Riva CE. Hemoglobin oxygen saturation (So2) in the human ocular fundus measured by reflectance oximetry: preliminary data in retinal veins. Klin Monbl Augenheilkd 2002;219:289 –91. 24. Petrig BL, Riva CE, Hayreh SS. Laser Doppler flowmetry and optic nerve head blood flow. Am J Ophthalmol 1999; 127:413–25. 25. Shonat RD, Kight AC. Oxygen tension imaging in the mouse retina. Ann Biomed Eng 2003;31:1084 –96. 26. Shonat RD, Wilson DF, Riva CE, Pawlowski M. Oxygen distribution in the retinal and choroidal vessels of the cat as measured by a new phosphorescence imaging method. Appl Opt 1992;31:3711– 8. 27. Woon WH, Fitzke FW, Bird AC, Marshall J. Confocal imaging of the fundus using a scanning laser ophthalmoscope. Br J Ophthalmol 1992;76:470 – 4. 28. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy Study report number 1. Early
29.
30.
31.
32.
33. 34.
35.
Treatment Diabetic Retinopathy Study research group. Arch Ophthalmol 1985;103:1796 – 806. Li HK, Danis RP, Hubbard LD, et al. Comparability of digital photography with the ETDRS film protocol for evaluation of diabetic retinopathy severity. Invest Ophthalmol Vis Sci 2011; 52:4717–25. Zheng Y, Gandhi JS, Stangos AN, et al. Automated segmentation of foveal avascular zone in fundus fluorescein angiography. Invest Ophthalmol Vis Sci 2010;51:3653–9. Palanker D, Blumerkranz MS, Weiter JJ Retinal laser therapy: biophysics basis and applications. In: Ryan SJ, ed-in-chief; Hinton DR, Schachat AP, eds. Retina. vol. 1. Basic Science and Inherited Retinal Disease, Tumors. 4th ed. Philadelphia: Elsevier Mosby; 2006:541. Peli E, Hedges TR 3rd, McInnes T, et al. Nerve fiber layer photography. A comparative study. Acta Ophthalmol (Copenh) 1987;65:71– 80. Behrendt T, Wilson LA. Spectral reflectance photography of the retina. Am J Ophthalmol 1965;59:1079 – 88. Shin YU, Kim S, Lee BR, et al. Novel noninvasive detection of the fovea avascular zone using confocal red-free imaging in diabetic retinopathy and retinal vein occlusion. Invest Ophthalmol Vis Sci 2012;53:309 –15. Fundus photographic risk factors for progression of diabetic retinopathy. ETDRS report number 12. Early Treatment Diabetic Retinopathy Study Research Group. Ophthalmology 1991;98:823–33.
Footnotes and Financial Disclosures Originally received: September 2, 2011. Final revision: January 19, 2012. Accepted: January 19, 2012. Available online: April 4, 2012.
Financial Disclosures: The authors have made the following disclosures: Byung Ro Lee – Consultant – Nidek, Gamagori, Japan. Manuscript no. 2011-1307.
1
Department of Ophthalmology, College of Medicine, Hanyang University, Seoul, Korea.
2
Department of Biomedical Engineering, Hanyang University, Seoul, Korea.
1454
Correspondence: Byung Ro Lee, MD, Department of Ophthalmology, Hanyang University Hospital, #17 Seongdong-gu, Haengdang-dong, Seoul, 133-792, Korea. E-mail: [email protected].