Digital Color Fundus Photographs Research Articles

Detection of Macular Photocoagulation Scars With Confocal Infrared Reflection Imaging

To study the diagnostic reliability (specificity and sensitivity) of confocal infrared reflection in detecting threshold argon laser photocoagulation scars in the macula. Fifty-six maculae with diabetic macular edema were evaluated by biomicroscopic slit-lamp fundus examination, digital color fundus photography, digital fundus fluorescein angiography, and digital infrared reflection images. Three examiners evaluated whether the eye had undergone any laser photopexy in the macula. Sensitivity, specificity, and false-positive and false-negative results were calculated for each method. Fluorescein fundus angiography and infrared imaging, although using different approaches, both detect pigment epithelium changes such as laser scars. It seems that both methods are of equal specificity. On the other hand, both biomicroscopic fundus examination and digital color photography showed poor reliability. Infrared reflection imaging is an easy, noninvasive method with excellent sensitivity and specificity in detecting photocoagulation scars from previous threshold laser treatment. It may be useful in estimating photocoagulated areas, especially if threshold treatment has been applied.

A CLASSIFICATION SYSTEM FOR POLYPOIDAL CHOROIDAL VASCULOPATHY

To describe the demographic features and clinical characteristics of polypoidal choroidal vasculopathy (PCV) in Chinese patients and define a new classification system. Retrospective review of 138 eyes of 123 patients presenting to the Singapore National Eye Center with PCV. Patients underwent ophthalmologic examination including digital color fundus photography and stereoscopic indocyanine green angiography. Classification based on indocyanine green angiography findings. Mean age of patient 68.3 years and 62.4% were men. PCV was unilateral in 87.8% cases and age-related maculopathy was present in the unaffected fellow eye in 22.8%. Average largest size of polyp was 207 microm. PCV lesions were found in multiple discrete areas in 34.8%. Formation of lesion was cluster in 66.7%, single in 27.5%, and string in 5.8%. PCV lesions were found in the extrafoveal area in 63.0%, subfoveal in 29.7%, juxtafoveal in 15.9%, and peripapillary in 8.0%. Demographics of PCV, unilaterality and frequency of age-related maculopathy in fellow eye similar to other reports in Asians. We describe a classification system for PCV comprising polyp size, location, formation, and number of discrete polyp areas, which can be used for prospective interventional clinical studies and may aid in future prognosis and management of this condition.

Classification simplifiée de la rétinopathie diabétique adaptée au dépistage par photographies du fond d’œil

Fundus photographs using a nonmydriatic digital camera are the reference method for diabetic retinopathy screening today. The aim of this study was to validate a simplified diabetic retinopathy classification, adapted for diabetic retinopathy screening, including all diabetic retinopathy severity scales.One hundred and twenty-one diabetic patients had three digital color fundus photographs taken, which were graded by four independent ophthalmologists according to the proposed screening classification, and a reference examination (nine field fundus photographs and optical coherence tomography examination of the macula).The proposed diabetic retinopathy screening classification was easy to use, as it was based on a visual comparison between the three digital color fundus photographs and standard retinal photographs. This classification provided the diagnosis of severe levels of diabetic retinopathy with high sensitivity and accurate specificity (100% and 50%-58%,respectively, for a moderate nonproliferative diabetic retinopathy level or higher on the screening examination) and the diagnosis of macular edema with a high sensitivity and specificity (96%-97% and 89%-91%, respectively). The results of the screening examination using this classification were highly reproducible (weighted k for interobserver agreement: 0.78-0.93).These results suggest that this new simplified classification is accurate. The screening method should not be used above the threshold level of mild nonproliferative diabetic retinopathy. Thus, from the moderate nonproliferative level (inclusive), a complete fundus examination should be undertaken. Severe levels of diabetic retinopathy will be referred to an ophthalmologist without delay for laser treatment.

Ocular fundus auto-fluorescence observations at different wavelengths in patients with age-related macular degeneration and diabetic retinopathy

Post-translational protein modification by lipid peroxidation products or glycation is a feature of aging as well as pathologic processes in postmitotic cells at the ocular fundus exposed to an oxidative environment. The accumulation of modified proteins such as those found in lipofuscin and advanced glycation end products (AGEs) contribute greatly to the fundus auto-fluorescence. The distinct fluorescence spectra of lipofuscin and AGE enable their differentiation in multispectral fundus fluorescence imaging. A dual-centre consecutive case series of 78 pseudo-phacic patients is reported. Digital colour fundus photographs as well as auto-fluorescence images were taken from 33 patients with age related macular degeneration (AMD), 13 patients with diabetic retinopathy (RD), or from 32 cases without pathologic findings (controls). Fluorescence was excited at 475-515 nm or 476-604 nm and recorded in the emission bands 530-675 nm or 675-715 nm, respectively. Fluorescence images excited at 475-515 nm were taken by a colour CCD-camera (colour-fluorescence imaging) enabling the separate recording of green and red fluorescence. The ratio of green versus red fluorescence was calculated within a representative region of each image. The 530-675 nm auto-fluorescence in AMD patients was dominated by the red emission (green vs. red ratio, g/r = 0.861). In comparison, the fluorescence of the diabetics was green-shifted (g/r = 0.946; controls: g/r = 0.869). Atrophic areas (geographic atrophy, laser scars) showed massive hypo-fluorescence in both emission bands. Hyper-fluorescent drusen and exudates, unobtrusive in the colour fundus images as well as in the fluorescence images with emission >667 nm, showed an impressive green-shift in the colour-fluorescence image. Lipofuscin is the dominant fluorophore at long wavelengths (>675 nm or red channel of the colour fluorescence image). In the green spectral region, we found an additional emission of collagen and elastin (optic disc, sclera) as well as deposits in drusen and exudates. The green shift of the auto-fluorescence in RD may be a hint of increased AGE concentrations.

Autofluorescence Imaging in Age-Related Macular Degeneration Complicated by Choroidal Neovascularization: A Prospective Study

To determine if integrity of the retinal pigment epithelium (RPE)/photoreceptor complex as assessed by autofluorescence imaging can be predicted on the basis of visual acuity (VA), size, or fluorescein angiographic characteristics of the lesion in the early stage of choroidal neovascularization in age-related macular degeneration (AMD). Prospective, observational, consecutive case series. Seventy-nine eyes of 78 patients with untreated early-stage subfoveal neovascular AMD. Digital color fundus photography and fluorescein angiography were carried out by certified photographers using the same camera throughout the study. Confocal scanning laser ophthalmoscopy images were obtained using a retinal angiograph. Autofluorescence images were compared with digital fluorescein angiography and fundus color photographs using IMAGEnet. Autofluorescence at the macula was correlated with VA, angiographic lesion characteristics, lesion size, and length of symptoms. Of the 79 eyes studied, 40 had classic and predominantly classic choroidal neovascularization, 10 had minimally classic, 29 had occult, 75 were subfoveal, and 4 were juxtafoveal. In 54 eyes, autofluorescence was continuous at the central macula, and this correlated significantly with VA, lesion size, and symptom length but not choroidal neovascularization type. However, there was considerable overlap between the 2 groups such that the integrity of RPE autofluorescence could not be predicted on the basis of these criteria. Intact autofluorescence at the macula in early choroidal neovascularization correlates with VA, lesion size, and symptom length but not lesion type. None predict with any certainty the integrity of the outer retina. Our data suggest that the RPE/photoreceptor complex may be intact at the macula for several months in the presence of choroidal neovascularization, suggesting that VA might be rescued if treatment were effective in suppressing neovascular growth without damaging the RPE/retina complex, although this remains to be tested. It would be sensible to assess autofluorescence in treatment protocols to test this concept because it may be a marker for earlier disease and predict outcomes of treatment.

Automated detection and differentiation of drusen, exudates, and cotton-wool spots in digital color fundus photographs for diabetic retinopathy diagnosis.

To describe and evaluate a machine learning-based, automated system to detect exudates and cotton-wool spots in digital color fundus photographs and differentiate them from drusen, for early diagnosis of diabetic retinopathy. Three hundred retinal images from one eye of 300 patients with diabetes were selected from a diabetic retinopathy telediagnosis database (nonmydriatic camera, two-field photography): 100 with previously diagnosed bright lesions and 200 without. A machine learning computer program was developed that can identify and differentiate among drusen, (hard) exudates, and cotton-wool spots. A human expert standard for the 300 images was obtained by consensus annotation by two retinal specialists. Sensitivities and specificities of the annotations on the 300 images by the automated system and a third retinal specialist were determined. The system achieved an area under the receiver operating characteristic (ROC) curve of 0.95 and sensitivity/specificity pairs of 0.95/0.88 for the detection of bright lesions of any type, and 0.95/0.86, 0.70/0.93, and 0.77/0.88 for the detection of exudates, cotton-wool spots, and drusen, respectively. The third retinal specialist achieved pairs of 0.95/0.74 for bright lesions and 0.90/0.98, 0.87/0.98, and 0.92/0.79 per lesion type. A machine learning-based, automated system capable of detecting exudates and cotton-wool spots and differentiating them from drusen in color images obtained in community based diabetic patients has been developed and approaches the performance level of retinal experts. If the machine learning can be improved with additional training data sets, it may be useful for detecting clinically important bright lesions, enhancing early diagnosis, and reducing visual loss in patients with diabetes.

Maculopathy Due to the R345W Substitution in Fibulin-3: Distinct Clinical Features, Disease Variability, and Extent of Retinal Dysfunction

To determine (1) clinical features that distinguish maculopathy due to the R345W substitution in fibulin-3 from other forms of inherited or early-onset drusen, (2) the phenotypic variability, and (3) the extent of retinal disease in those with a positive molecular diagnosis. Affected individuals underwent ophthalmic examination, digital color fundus photography, fundus autofluorescence (AF) imaging, and psychophysical testing with automated photopic and dark-adapted perimetry and fine matrix mapping. Blood samples were taken for DNA extraction and screening for the R345W mutation in fibulin-3. Patients were subsequently divided into mutation-positive and -negative groups, to compare the identified phenotypic findings in these two sets of subjects. Twenty-nine subjects from 19 families were ascertained with inherited or early-onset drusen. Twenty-four (83%) subjects from 15 families were found to harbor the R345W fibulin-3 mutation. Peripapillary deposition and a radial distribution of macular drusen were consistent, distinguishing signs in the mutation-positive group. Subretinal neovascular membrane (SRNVM) was a rare occurrence, affecting only 1 of 48 eyes, whereas hyperpigmentation and atrophy of the retinal pigment epithelium (RPE) were common in older mutation-positive patients. Increased AF corresponding to the drusen was detected in both the mutation-positive and -negative groups. The phenotype in the group of patients positive for the R345W mutation was extremely variable, with evidence of interocular, intrafamilial, and interfamilial variability in visual loss, natural history, ophthalmoscopic findings, autofluorescence imaging, and psychophysical data. The novel finding of nonpenetrance was observed in a 62-year-old asymptomatic, mutation-positive man. The findings from detailed perimetry performed on a subset of subjects were consistent with the presence of widespread retinal dysfunction not isolated to the macula. Marked inter- and intrafamilial variation associated with the fibulin-3 R345W mutation in terms of retinal appearance, severity, progression, and nonpenetrance were identified. It was noted that SRNVM is a rare occurrence in R345W fibulin-3 maculopathy. These findings are helpful for advice regarding prognosis and for genetic counseling. The findings established that the presence of peripapillary deposit is highly likely to indicate that a patient carries the R345W mutation.

Automatic detection of red lesions in digital color fundus photographs

The robust detection of red lesions in digital color fundus photographs is a critical step in the development of automated screening systems for diabetic retinopathy. In this paper, a novel red lesion detection method is presented based on a hybrid approach, combining prior works by Spencer et al. (1996) and Frame et al. (1998) with two important new contributions. The first contribution is a new red lesion candidate detection system based on pixel classification. Using this technique, vasculature and red lesions are separated from the background of the image. After removal of the connected vasculature the remaining objects are considered possible red lesions. Second, an extensive number of new features are added to those proposed by Spencer-Frame. The detected candidate objects are classified using all features and a k-nearest neighbor classifier. An extensive evaluation was performed on a test set composed of images representative of those normally found in a screening set. When determining whether an image contains red lesions the system achieves a sensitivity of 100% at a specificity of 87%. The method is compared with several different automatic systems and is shown to outperform them all. Performance is close to that of a human expert examining the images for the presence of red lesions.

Prospective evaluation of digital non-stereo color fundus photography as a screening tool in age-related macular degeneration

To evaluate mydriatic, non-stereo digital color fundus photographs as a screening tool for identifying and classifying exudative age-related macular degeneration (AMD). Prospective case series. Digital color fundus photographs were obtained from patients seen in the AMD screening clinic over a 9-month period at the Ivey Eye Institute in London, Ontario. Photographs for eligible patients were separated by eye, cataloged, blinded, and randomly labeled before interpretation by an experienced vitreoretinal surgeon. Exact agreement, sensitivity, specificity, positive predictive value, and negative predictive value of the fundus photographs in diagnosing, classifying, and managing cases of suspected exudative AMD were then calculated against gold standard clinical examination and fluorescein angiography. A total of 223 images were used from 118 eligible patients. Exact agreement between photographic evaluation and gold standard ranged from 89.2% (presence of pigment epithelial detachment (PED)) to 82.5% (evidence of retinal pigment epithelium geographic atrophy). Sensitivities ranged from 89.2% (presence of choroidal neovascular membrane (CNVM)) to 40.0% (presence of PED). Specificities ranged from 94.1% (presence of PED) to 86.8% (presence of retinal pigment epithelium geographic atrophy). Positive predictive value ranged from 86.1% (presence of CNVM) to 40.0% (presence of PED). Negative predictive value ranged from 94.1% (presence of PED) to 88.9% (presence of CNVM). As a screening tool for high-risk dry changes and active exudative changes, overall sensitivity specificity, positive predictive value, and negative predictive value were 82.1%, 79.1%, 70.4%, and 88.0%, respectively. Digital, non-stereo color fundus photographs are highly sensitive and have high negative predictive value as a screening tool. Very few treatable lesions are missed using telemedicine in age-related macular degeneration.