MHANet: A hybrid attention mechanism for retinal diseases classification.

With the increase of patients with retinopathy, retinopathy recognition has become a research hotspot. In this article, we describe the etiology and symptoms of three kinds of retinal diseases, including drusen(DRUSEN), choroidal neovascularization(CNV) and diabetic macular edema(DME). In addition, we also propose a hybrid attention mechanism to classify and recognize different types of retinopathy images. In particular, the hybrid attention mechanism proposed in this paper includes parallel spatial attention mechanism and channel attention mechanism. It can extract the key features in the channel dimension and spatial dimension of retinopathy images, and reduce the negative impact of background information on classification results. The experimental results show that the hybrid attention mechanism proposed in this paper can better assist the network to focus on extracting thr fetures of the retinopathy area and enhance the adaptability to the differences of different data sets. Finally, the hybrid attention mechanism achieved 96.5% and 99.76% classification accuracy on two public OCT data sets of retinopathy, respectively.

Neovascularization in Fellow Eye of Unilateral Neovascular Age-related Macular Degeneration According to Different Drusen Types.

PURPOSE: To investigate the incidence of fellow eye (FE) neovascular age-related macular degeneration (nAMD) in patients with unilateral nAMD according to FE drusen type. DESIGN: Retrospective cohort study. METHODS: Between January 2013 and June 2016, 434 consecutive patients with naïve nAMD were enrolled. We selected 280 eligible patients with treatment-naïve, unilateral nAMD for analysis (280/280 = 100% patients were followed up at 2 years; 50/280 = 17.9% patients were followed up at 5 years). The incidence and hazard ratios (HR) of FE nAMD according to age, sex, choroidal thickness, nAMD subtype, and drusen type were analyzed. RESULTS: The 5-year incidence of FE nAMD was 20.9%. The incidences of the soft plus subretinal drusenoid deposits (SDD), soft drusen only, and SDD only groups were 76.4%, 46.2%, and 25.7%, respectively; they were significantly higher than the no drusen group (vs 3.6%; P < .001, P < .001, P < .001). There was no significant difference between the pachydrusen and no drusen groups (7.1% vs 3.6%; P = .101). The multivariate Cox regression hazard model revealed older age (HR, 1.053; P = .031) and drusen type were significant (P = .001). Compared with the no drusen group, the soft drusen plus SDD, soft drusen only, and SDD groups showed an HR of 18.460 (P = .001), 8.302 (P = .015), and 5.465 (P = .082), respectively. Pachydrusen was not shown to be a significant risk factor compared to the no drusen group (HR, 2.417; P = .281). CONCLUSION: The incidence of FE nAMD was significantly different with respect to drusen type. Soft drusen plus SDD had the highest risk of neovascular AMD, followed by soft drusen only and SDD only.

Manifestaciones retinianas en pacientes con glomerulonefritis membranoproliferativa mediada por complemento / Retinal manifestations in patients with complement-mediated membranoproliferative glomerulonephritis

La glomerulonefritis membranoproliferativa mediada por complemento es una enfermedad rara, pero puede asociarse a alteraciones retinianas. Por ello, el propósito de este estudio fue valorar una serie de casos con dicho diagnóstico en seguimiento por nefrología en nuestro centro. Se realizó un estudio transversal de 8 pacientes diagnosticados de glomerulonefritis membranoproliferativa mediada por complemento. Se realizó funduscopia, tomografía de coherencia óptica (OCT) y angio-OCT de dominio Swept Source. Uno de los 8 pacientes presentaba depósitos drusenoides que se localizaron bajo el epitelio pigmentario retiniano en la OCT, descartándose la presencia de neovascularización coroidea asociada en la angio-OCT. Por tanto, la glomerulonefritis membranoproliferativa puede producir alteraciones retinianas con drusas o desprendimientos del epitelio pigmentario retiniano, por lo que debe hacerse un adecuado diagnóstico diferencial con la degeneración macular asociada a la edad. Es crucial realizar un seguimiento en estos pacientes ante las posibles complicaciones que pueden desembocar en una pérdida de visión

Complement-mediated membranoproliferative glomerulonephritis is a rare progressive glomerular disease. In some patients it can be associated with retinal lesions. Therefore, the purpose of this study was to assess a case series with this diagnosis in our hospital. A cross-sectional study was conducted on 8 patients diagnosed with complement-mediated membranoproliferative glomerulonephritis. Funduscopy, optical coherence tomography (OCT) and Swept Source domain OCT angiography were performed. Only 1 of the 8 patients showed drusen-like deposits that were located under the retinal pigment epithelium in the OCT, with the presence of associated choroidal neovascularization being ruled out in OCT angiography. Therefore, membranoproliferative glomerulonephritis may produce retinal alterations with drusen or retinal pigment epithelium detachment, and requires an appropriate differential diagnosis to be made with age-related macular degeneration. The follow-up of these patients is important in order to detect vision-threatening complications

DeepSeeNet: A Deep Learning Model for Automated Classification of Patient-based Age-related Macular Degeneration Severity from Color Fundus Photographs.

PURPOSE: In assessing the severity of age-related macular degeneration (AMD), the Age-Related Eye Disease Study (AREDS) Simplified Severity Scale predicts the risk of progression to late AMD. However, its manual use requires the time-consuming participation of expert practitioners. Although several automated deep learning systems have been developed for classifying color fundus photographs (CFP) of individual eyes by AREDS severity score, none to date has used a patient-based scoring system that uses images from both eyes to assign a severity score. DESIGN: DeepSeeNet, a deep learning model, was developed to classify patients automatically by the AREDS Simplified Severity Scale (score 0-5) using bilateral CFP. PARTICIPANTS: DeepSeeNet was trained on 58 402 and tested on 900 images from the longitudinal follow-up of 4549 participants from AREDS. Gold standard labels were obtained using reading center grades. METHODS: DeepSeeNet simulates the human grading process by first detecting individual AMD risk factors (drusen size, pigmentary abnormalities) for each eye and then calculating a patient-based AMD severity score using the AREDS Simplified Severity Scale. MAIN OUTCOME MEASURES: Overall accuracy, specificity, sensitivity, Cohen's kappa, and area under the curve (AUC). The performance of DeepSeeNet was compared with that of retinal specialists. RESULTS: DeepSeeNet performed better on patient-based classification (accuracy = 0.671; kappa = 0.558) than retinal specialists (accuracy = 0.599; kappa = 0.467) with high AUC in the detection of large drusen (0.94), pigmentary abnormalities (0.93), and late AMD (0.97). DeepSeeNet also outperformed retinal specialists in the detection of large drusen (accuracy 0.742 vs. 0.696; kappa 0.601 vs. 0.517) and pigmentary abnormalities (accuracy 0.890 vs. 0.813; kappa 0.723 vs. 0.535) but showed lower performance in the detection of late AMD (accuracy 0.967 vs. 0.973; kappa 0.663 vs. 0.754). CONCLUSIONS: By simulating the human grading process, DeepSeeNet demonstrated high accuracy with increased transparency in the automated assignment of individual patients to AMD risk categories based on the AREDS Simplified Severity Scale. These results highlight the potential of deep learning to assist and enhance clinical decision-making in patients with AMD, such as early AMD detection and risk prediction for developing late AMD. DeepSeeNet is publicly available on https://github.com/ncbi-nlp/DeepSeeNet.

Evaluation of intraretinal migration of retinal pigment epithelial cells in age-related macular degeneration using polarimetric imaging.

The purpose of the present study was to evaluate the intraretinal migration of the retinal pigment epithelium (RPE) cells in age-related macular degeneration (AMD) using polarimetry. We evaluated 155 eyes at various AMD stages. Depolarized light images were computed using a polarization-sensitive scanning laser ophthalmoscope (PS-SLO), and the degree of polarization uniformity was calculated using polarization-sensitive optical coherence tomography (OCT). Each polarimetry image was compared with the corresponding autofluorescence (AF) images at 488 nm (SW-AF) and at 787 nm (NIR-AF). Intraretinal RPE migration was defined by the presence of depolarization at intraretinal hyperreflective foci on PS-SLO and PS-OCT images, and by the presence of hyper-AF on both NIR-AF and SW-AF images. RPE migration was detected in 52 of 155 eyes (33.5%) and was observed in drusenoid pigment epithelial detachment (PED) and serous PED with significantly higher frequencies than in other groups (P = 0.015). The volume of the migrated RPE cluster in serous PED was significantly correlated with the volume of the PED (R2 = 0.26; P = 0.011). Overall, our results showed that intraretinal RPE migrations occurred in various AMD stages, and that they occurred more commonly in eyes with serous and drusenoid PED.

[Macular drusen variability: multimodal imaging potential]. / Variabel'nost' makulyarnykh druz: vozmozhnosti mul'timodal'noi vizualizatsii.

Insufficient and controversial knowledge about the macular drusen (MD), a lack of scientifically proven management methods for drusen and their strong correlation with AMD active progression makes MD an important area of research. AIM: The purpose of the study ­ to assess clinical feature of MD using modern digital imaging technologies. MATERIAL AND METHODS: Patients with both hard and soft drusen were studied using fluorescein angiography, swept-source optical coherence tomography (OCT), OCT-angiography, autofluorescence (both short-wavelength and near infra-red), scanning laser ophthalmoscopy in Multicolor mode. The retina, choroid and vitreoretinal interface were assessed on 50 patients with AMD and drusen using different imaging modalities. An additional group of the study was presented by 5 patients with geographic atrophy (GA) formed as a result of soft drusen fading, where retrospective assessment of the OCT scans was performed with special attention to the signs of soft drusen regression associated with atrophy of the overlying RPE. RESULTS: Two types of hard drusen were defined as the reticular pseudodrusen and the cuticular drusen. The qualitative and comparative analysis of data for each type of MD was performed. Vitreoretinal interface evaluation demonstrated the correlation between vitreomacular adhesion and mixed reticular and cuticular drusen. The choroidal thickness assessment in 9 different macular sectors in drusenoid eyes does not reveal a significant difference with control group. All of the analysed drusen-faded-eyes initially had been presented with OCT patterns of "nascent" GA. CONCLUSION: The modern retinal imaging techniques enable new approach to the diagnostic differentiation and description of various macular drusen types. The value of these methods for AMD prognosis is yet to be further investigated.

An evaluation of fundus photography and fundus autofluorescence in the diagnosis of cuticular drusen.

PURPOSE: To examine non-mydriatic fundus photography (FP) and fundus autofluorescence (FAF) as alternative non-invasive imaging modalities to fluorescein angiography (FA) in the detection of cuticular drusen (CD). METHODS: Among 2953 adults from the Danish Rural Eye Study (DRES) with gradable FP, three study groups were selected: (1) All those with suspected CD without age-related macular degeneration (AMD) on FP, (2) all those with suspected CD with AMD on FP and (3) a randomly selected group with early AMD. Groups 1, 2 and 3 underwent FA and FAF and group 4 underwent FAF only as part of DRES CD substudy. Main outcome measures included percentage of correct positive and correct negative diagnoses, Cohen's κ and prevalence-adjusted and bias-adjusted κ (PABAK) coefficients of test and grader reliability. RESULTS: CD was correctly identified on FP 88.9% of the time and correctly identified as not being present 83.3% of the time. CD was correctly identified on FAF 62.0% of the time and correctly identified as not being present 100.0% of the time. Compared with FA, FP has a PABAK of 0.75 (0.60 to 1.5) and FAF a PABAK of 0.44 (0.23 to 0.95). CONCLUSIONS: FP is a promising, non-invasive substitute for FA in the diagnosis of CD. FAF was less reliable than FP to detect CD.

A Population-Based Ultra-Widefield Digital Image Grading Study for Age-Related Macular Degeneration-Like Lesions at the Peripheral Retina.

PURPOSE: Our understanding of the relevance of peripheral retinal abnormalities to disease in general and in age-related macular degeneration (AMD) in particular is limited by the lack of detailed peripheral imaging studies. The purpose of this study was to develop image grading protocols suited to ultra-widefield imaging (UWFI) in an aged population. DESIGN: A cross-sectional study of a random population sample in which UWFI was introduced at the 12-year review of the Reykjavik Eye Study in Iceland. PARTICIPANTS: Five hundred seventy-six subjects 62 years of age or older. METHODS: Ultra-widefield (up to 200°) color and autofluorescence images were obtained using the Optos P200CAF laser scanning ophthalmoscope (Optos plc, Dunfermline, Scotland). The images were graded at Moorfields Eye Hospital Reading Centre primarily based on the International Classification for AMD. Macular and peripheral changes were graded using a standardized grid developed for this imaging method. MAIN OUTCOME MEASURES: Presence or absence of hard, crystalline, and soft drusen; retinal pigment epithelial changes; choroidal neovascularization (CNV); atrophy; and hypoautofluorescence and hyperautofluorescence were graded in the peripheral retina. RESULTS: Of the eyes examined, 81.1% had AMD-like changes in the macula alone (13.6%), periphery alone (10.1%), and both periphery and macula (57.4%). There was no AMD-like CNV or pigment epithelial detachment in the periphery except in those cases in which these clearly originated from the macula. Seven patients had AMD-like atrophy in the periphery without end-stage disease in the macula. One patient with end-stage disease in the macula had normal periphery results on the color images. While analyzing the eyes, we detected pathologic appearances that were very reliably identified by graders. CONCLUSIONS: Phenotyping the retinal periphery using the categories defined by the International Classification confirmed the presence of wide-ranging AMD-like pathologic changes even in those without central sight-threatening macular disease. Based on our observations, we propose here new, reliably identifiable grading categories that may be more suited for population-based UWFI.

Prevalence of intermediate-stage age-related macular degeneration in patients with acquired immunodeficiency syndrome.

PURPOSE: To evaluate the prevalence of intermediate-stage age-related macular degeneration (AMD) in patients with acquired immunodeficiency syndrome (AIDS). DESIGN: Cross-sectional study of patients with AIDS enrolled in the Longitudinal Study of the Ocular Complications of AIDS. METHODS: Intermediate-stage AMD was determined from enrollment retinal photographs by graders at a centralized Reading Center, using the Age-Related Eye Disease Study grading system. Graders were masked as to clinical data. RESULTS: Of 1825 participants with AIDS and no ocular opportunistic infections, 9.9% had intermediate-stage AMD. Risk factors included age, with an odds ratio (OR) of 1.9 (95% confidence interval [CI] 1.6, 2.3, P < .001) for every decade of age; the prevalence of AMD ranged from 4.0% for participants 30-39 years old to 24.3% for participants ≥60 years old. Other risk factors included the human immunodeficiency virus (HIV) risk groups of injection drug use (OR = 2.4, 95% CI 1.5, 3.9, P < .001) or heterosexual contact (OR = 1.9, 95% CI 1.3, 2.8, P = .001). Compared with the HIV-uninfected population in the Beaver Dam Offspring Study, there was an approximate 4-fold increased age-adjusted prevalence of intermediate-stage AMD. CONCLUSIONS: Patients with AIDS have an increased age-adjusted prevalence of intermediate-stage AMD compared with that found in a non-HIV-infected cohort evaluated with similar methods. This increased prevalence is consistent with the increased prevalence of other age-related diseases in antiretroviral-treated, immune-restored, HIV-infected persons when compared to non-HIV-infected persons.

MORPHOLOGY SCORE AS A MARKER OF RETINAL FUNCTION IN DRUSENOID PIGMENT EPITHELIAL DETACHMENT.

PURPOSE: To evaluate a morphology score for drusenoid pigment epithelial detachment (dPED) regarding predictability of a decline in retinal function beyond best-corrected visual acuity. METHODS: Thirteen eyes of 10 patients with dPED due to age-related macular degeneration (AMD) were included (age 72.8 ± 4.2 years). All underwent volume spectral domain optical coherence tomography, fluorescence angiography, and confocal scanning laser ophthalmoscopy infrared imaging as well as multifocal electroretinography and microperimetry. The dPED morphology score suggested consists of five parameters: hyperreflective spots in infrared, lesion diameter, lesion height, presence of vitelliform-like material in the subretinal space or subretinal fluid, and integrity of the ellipsoid zone in spectral domain optical coherence tomography. Subsequently, a score value between 0 and 1 according to the extent of morphologic changes was correlated to foveal multifocal electroretinography and microperimetry measurements. RESULTS: The mean best-corrected visual acuity was 20/40. The mean height and mean diameter of dPED were 312.2 ± 111 µm and 2,535 ± 805 µm. Two dPED showed no hyperreflective spots in confocal scanning laser ophthalmoscopy infrared images, three displayed a moderate stage of hyperreflective spots, and eight had severe hyperreflective spots. Two eyes showed subretinal fluid, and five patients showed vitelliform-like material in the subretinal space. Eight eyes revealed a severe disruption of the ellipsoid zone. Although no correlation was found between dPED morphology score and best-corrected visual acuity, eyes with a dPED morphology score >0.5 revealed distinctly decreased values in functional measurements compared with those with a score ≤0.5. CONCLUSION: The dPED morphology score aggregates all currently known morphologic changes in dPED and represents a valuable tool for clinical lesion evaluation. Furthermore, it allows for assessing an estimate of functional decline beyond best-corrected visual acuity.

Pseudodrusen subtypes as delineated by multimodal imaging of the fundus.

PURPOSE: To subclassify pseudodrusen based on their appearance in multimodal imaging. DESIGN: Retrospective, observational series. METHODS: The color fundus photographs and infrared scanning laser ophthalmoscope (IR-SLO) images of patients with pseudodrusen were evaluated along with spectral-domain optical coherence tomography (SD OCT) by masked readers. Distinct types of pseudodrusen could be differentiated. RESULTS: There were 140 eyes of 93 patients with a mean age of 82.4 years. Multimodal imaging analysis showed 3 subtypes of pseudodrusen. One principal type was an orderly array of whitish discrete accumulations principally located in the perifovea, termed dot pseudodrusen. They appeared as hyporeflective spots, often with a target configuration, in IR-SLO images. The second type was interconnected bands of yellowish-white material forming a reticular pattern, called ribbon pseudodrusen, which were located in the perifovea. This subtype was faintly hyporeflective in IR-SLO imaging. Dot pseudodrusen were detected more commonly with IR-SLO imaging than in color photography (P = .014) and ribbon pseudodrusen were seen more frequently in color than in IR-SLO images (P < .001). An uncommon third type of pseudodrusen, yellow-white globules primarily located peripheral to the perifoveal region, appeared hyper-reflective in IR-SLO and were called peripheral pseudodrusen. All 3 types were seen as subretinal drusenoid deposits by SD OCT. CONCLUSION: Pseudodrusen may be classified into at least 3 categories, each with optimal methods of detection and only 1 that formed a reticular pattern. These findings suggest pseudodrusen could contain differing constituents and therefore may vary in conferred risk for progression to advanced age-related macular disease.

Subretinal drusenoid deposits with increased autofluorescence in eyes with reticular pseudodrusen.

PURPOSE: To characterize a variant type of drusenoid deposit with different imaging features in comparison to reticular pseudodrusen. METHODS: Retrospective observational consecutive case series. Eyes showing atypical drusenoid lesions were sorted out from 257 eyes of 133 patients previously diagnosed as reticular pseudodrusen. Eyes were evaluated using color fundus photography, confocal scanning laser ophthalmoscopy, and spectral domain optical coherence tomography. RESULTS: A variant type of drusenoid deposits showing different imaging features from reticular pseudodrusen was found in 17 eyes of 12 patients (6.6%). The mean age of patients was 62.7 ± 11.6 years, and all patients were women. These deposits were observed as yellowish white, round to oval lesions on color photographs, located under the sensory retina and above the retinal pigment epithelium on spectral domain optical coherence tomography similar to reticular pseudodrusen. However, they were present in a smaller number as discrete lesions and showed increased autofluorescence. None of them were accompanied by late age-related macular degeneration. CONCLUSION: Subretinal drusenoid deposits are not homogeneous and can be classified into two types according to the fundus autofluorescence. Multimodal imaging tests are needed for the differential diagnosis of subretinal drusenoid deposits.

Automated "disease/no disease" grading of age-related macular degeneration by an image mining approach.

PURPOSE: To describe and evaluate an automated grading system for age-related macular degeneration (AMD) by color fundus photography. METHODS: An automated "disease/no disease" grading system for AMD was developed based on image-mining techniques. First, image preprocessing was performed to normalize color and nonuniform illumination of the fundus images to define a region of interest and to identify and remove pixels belonging to retinal vessels. To represent images for the prediction task, a graph-based image representation using quadtrees was then adopted. Next, a graph-mining technique was applied to the generated graphs to extract relevant features (in the form of frequent subgraphs) from images of both AMD and healthy volunteers. Features of the training data were then fed into a classifier generator for training purposes before employing the trained classifiers to classify new "unseen" images. RESULTS: The algorithm was evaluated on two publically available fundus-image datasets comprising 258 images (160 AMD and 98 normal). Ten-fold cross validation was used. The experiments produced a best specificity of 100% and a best sensitivity of 99.4% with an overall accuracy of 99.6%. Our approach outperformed previous approaches reported in the literature. CONCLUSIONS: This study has demonstrated a proof-of-concept, image-mining technique for automated AMD grading. This technique has the potential to be further developed as an automated grading tool for future whole-scale AMD screening programs.

Accuracy and reproducibility of automated drusen segmentation in eyes with non-neovascular age-related macular degeneration.

PURPOSE: To evaluate the accuracy and reproducibility of drusen quantification by an automated drusen segmentation algorithm in spectral domain optical coherence tomography (SD-OCT) images of eyes with non-neovascular age-related macular degeneration (AMD). METHODS: Drusen segmentation was performed using both a commercial automated algorithm (Cirrus OCT RPE analysis tool) and manual segmentation in 44 eyes of 30 subjects with dry AMD who underwent volume OCT scanning. The drusen (space between outer RPE layer and Bruch's membrane) was segmented automatically using an automated RPE tool and manually by 3D-OCTOR software. Drusen area and volume were calculated in all eyes. Age and visual acuity data were also collected. Reproducibility of manual and automated measurements was assessed by intraclass correlation (ICC). RESULTS: The mean age of subjects was 78.24 (± 9.4; range, 56-97 years). The mean logMAR (logarithm of the minimum angle of resolution) visual acuity was 0.4 (Snellen equivalent, ~20/50) (standard deviation, 0.40; range, 0-1.3). The mean (standard deviation) drusen area was 5.05 (3.67) mm(2) with manual segmentation and 4.66 (3.51) mm(2) with the automated RPE tool; the absolute difference was 2.63 (2.5) mm(2). The mean drusen volume was 1.49 (0.42) mm(3) with manual segmentation and 1.42 (0.43) mm(3) with the automated RPE tool; the absolute difference was 1.42 (0.43) mm(3). The agreement between manual and automated measurements of drusen volume (highest ICC = 0.95) was better than the agreement for drusen area (ICC = 0.65). CONCLUSIONS: The quantification of drusen area and volume using an automated RPE yielded better agreement for volume than for area when compared with human expert manual segmentation. Using this software, drusen volume measurements may be a useful tool for quantifying drusen burden in clinical trials and clinical practice.

Drusen characterization with multimodal imaging.

PURPOSE: To characterize the known appearance of cuticular drusen, subretinal drusenoid deposits (reticular pseudodrusen), and soft drusen as revealed by multimodal fundus imaging and to create an explanatory model that accounts for these observations. METHODS: Reported color, fluorescein angiographic, autofluorescence, and spectral domain optical coherence tomography images of patients with cuticular drusen, soft drusen, and subretinal drusenoid deposits were reviewed, as were actual images from affected eyes. Representative histological sections were examined. The geometry, location, and imaging characteristics of these lesions were evaluated. A hypothesis based on the Beer-Lambert law of light absorption was generated to fit these observations. RESULTS: Cuticular drusen appear as numerous, uniform, round, yellow-white punctate accumulations under the retinal pigment epithelium (RPE). Soft drusen are larger, yellow-white dome-shaped mounds of deposit under the RPE. Subretinal drusenoid deposits are polymorphous light-gray interconnected accumulations above the RPE. Based on the model, both cuticular and soft drusen appear yellow because of the removal of shorter wavelength light by a double pass through the RPE. Subretinal drusenoid deposits, which are located on the RPE, are not subjected to short-wavelength attenuation and therefore are more prominent when viewed with blue light. The location and morphology of extracellular material in relationship to the RPE, and associated changes to RPE morphology and pigmentation, appeared to be the primary determinants of druse appearance in different imaging modalities. CONCLUSION: Although cuticular drusen, subretinal drusenoid deposits, and soft drusen are composed of common components, they are distinguishable by multimodal imaging because of differences in location, morphology, and optical filtering effects by drusenoid material and the RPE.

Dynamic soft drusen remodelling in age-related macular degeneration.

AIMS: To demonstrate and quantify the dynamic remodelling process of soft drusen resorption and new drusen formation in age-related macular degeneration (AMD) with novel interactive methods. METHODS: Twenty patients with large soft drusen bilaterally and without advanced AMD were imaged at baseline and again at a mean interval of 2 years (40 eyes, 80 images). Each of the 40 serial pairs of images was precisely registered by an automated technique. The drusen were segmented by a user-interactive method based on a background levelling algorithm and classified into three groups: new drusen (only in the final image), resorbed drusen (present initially but not in the final image) and stable drusen (present in both images). We measured each of these classes as well as the absolute change in drusen |D1-D0| and the dynamic drusen activity (creation and resorption) D(new)+D(resorbed). RESULTS: Mean dynamic activity for the right eye (OD) was 7.33±5.50%, significantly greater than mean absolute change (2.71±2.89%, p=0.0002, t test), with similar results for the left eye (OS). However, dynamic activity OD compared with OS (mean 7.33±5.50 vs 7.91±4.16%, NS) and absolute net change OD versus OS (2.71±2.89 vs 3.46±3.97%, NS) tended to be symmetrical between fellow eyes. CONCLUSIONS: Dynamic remodelling processes of drusen resorption and new drusen formation are distinct disease activities that can occur simultaneously and are not captured by change in total drusen load. Dynamic changes occur at rates more than twice that of net changes, and may be a useful marker of disease activity.

Description of the Age-Related Eye Disease Study 9-step severity scale applied to participants in the Complications of Age-related Macular Degeneration Prevention Trial.

OBJECTIVE: To describe characteristics of the Age-Related Eye Disease Study (AREDS) 9-step severity scale applied to participants in the Complications of Age-related Macular Degeneration Prevention Trial (CAPT). METHODS: Eligibility criteria for CAPT required 10 or more large (>or=125 microm) drusen in each eye. Readers graded baseline photographs from all participants and all follow-up photographs from 402 untreated eyes. Drusen and pigment characteristics were used to assign the AREDS scale score. Choroidal neovascularization was identified from fluorescein angiograms. Geographic atrophy involving the macular center was identified from color photographs. RESULTS: Among 1001 untreated eyes, 90% were at steps 5 to 7 at baseline. The 5-year incidence of advanced age-related macular degeneration (AMD) increased with each step from 8% (step 4) to 40% (steps 8 and 9 combined). These rates were similar to those reported in AREDS. Among 261 eyes with all 5 annual photograph gradings available and without progression to advanced AMD, 55% of eyes had scores that indicated improvement at least once. Before progression to advanced AMD, only 32% of 141 eyes either went through step 8 or 9 or had an increase of 2 or more steps from baseline. CONCLUSIONS: The AREDS 9-step severity scale was predictive of development of advanced AMD. The AREDS scale has deficiencies as a surrogate outcome for progression to advanced AMD.

Drusen ultrastructure imaging with spectral domain optical coherence tomography in age-related macular degeneration.

PURPOSE: To categorize drusen ultrastructure in age-related macular degeneration (AMD) using spectral domain optical coherence tomography (SDOCT) and correlate the tomographic and photographic drusen appearances. DESIGN: Prospective case series. PARTICIPANTS: Thirty-one eyes of 31 patients with non-neovascular AMD. METHODS: Subjects with drusen and a clinical diagnosis of AMD were enrolled in an SDOCT imaging study from August of 2005 to May of 2007. Foveal linear scans were acquired, and the image data were processed for analysis. Drusen were scored by 4 morphologic categories: shape, predominant internal reflectivity, homogeneity, and presence of overlying hyper-reflective foci. The prevalences of each morphologic pattern and combinations of morphologic patterns observed were calculated. The photographic appearance of each druse was compared with the tomographic classification. Interobserver and intraobserver agreement analysis was performed. MAIN OUTCOME MEASURES: Prevalence of morphologic parameters using SDOCT. RESULTS: Twenty-one eyes of 21 patients had SDOCT B-scans of adequate quality for analysis. On the basis of the above morphologic categories, 17 different drusen patterns were found in 120 total drusen. The most common was convex, homogeneous, with medium internal reflectivity, and without overlying hyper-reflective foci, present in 17 of 21 eyes (81%). Of the 16 eyes (76%) with nonhomogeneous drusen, 5 had a distinct hyper-reflective core. Hyper-reflective foci overlying drusen were in 7 eyes (33%). Although half of the photographically soft-indistinct drusen were convex with medium internal reflectivity and homogeneous without overlying hyper-reflective foci, the other half had significant variability in their tomographic appearance. Both interobserver and intraobserver agreement in drusen grading were high. Readers agreed the most when grading drusen shape and reflectivity, whereas the least agreement was for drusen homogeneity. CONCLUSIONS: Drusen ultrastructure can be imaged with SDOCT and characterized with a simple grading system. Photographic appearance may predict some but not all tomographic appearances. Trained observers have a high level of agreement with this grading system. These in vivo morphologic characteristics imaged with SDOCT may be distinct subclasses of drusen types, may relate closely to ultrastructural drusen elements identified in cadaveric eyes, and may be useful imaging biomarkers for disease severity or risk of progression. This will require validation from further studies.

Prevalence and morphology of druse types in the macula and periphery of eyes with age-related maculopathy.

PURPOSE: Macular drusen are hallmarks of age-related maculopathy (ARM), but these focal extracellular lesions also appear with age in the peripheral retina. The present study was conducted to determine regional differences in morphology that contribute to the higher vulnerability of the macula to advanced disease. METHODS: Drusen from the macula (n = 133) and periphery (n = 282) were isolated and concentrated from nine ARM-affected eyes. A semiquantitative light microscopic evaluation of 1-mum-thick sections included 12 parameters. RESULTS: Significant differences were found between the macula and periphery in ease of isolation, distribution of druse type, composition qualities, and substructures. On harvesting, macular drusen were friable, with liquefied or crystallized contents. Peripheral drusen were resilient and never crystallized. On examination, soft drusen appeared in the macula only, had homogeneous content without significant substructures, and had abundant basal laminar deposits (BlamD). Several substructures, previously postulated as signatures of druse biogenesis, were found primarily in hard drusen. Specific to hard drusen, which appeared everywhere, were central subregions and reduced RPE coverage. Macular hard drusen with a rich substructure profile differed from primarily homogeneous peripheral hard drusen. Compound drusen, found in the periphery only, exhibited a composition profile that was not intermediate between hard and soft. CONCLUSIONS: The data confirm regional differences in druse morphology, composition, and physical properties, most likely based on different formative mechanisms that may contribute to macular susceptibility for ARM progression. Two other reasons that only the macula is at high risk despite having relatively few drusen are the exclusive presence of soft drusen and the abundant BlamD in this region.

IDOCS: intelligent distributed ontology consensus system--the use of machine learning in retinal drusen phenotyping.

PURPOSE: To use the power of knowledge acquisition and machine learning in the development of a collaborative computer classification system based on the features of age-related macular degeneration (AMD). METHODS: A vocabulary was acquired from four AMD experts who examined 100 ophthalmoscopic images. The vocabulary was analyzed, hierarchically structured, and incorporated into a collaborative computer classification system called IDOCS. Using this system, three of the experts examined images from a second set of digital images compiled from more than 1000 patients with AMD. Images were annotated, and features were identified and defined. Decision trees, a machine learning method, were trained on the data collected and used to extract patterns. Interrelationships between the data from the different clinicians were investigated. RESULTS: Six drusen classes in the structured vocabulary were largely sufficient to describe all the identified features. The decision trees classified the data with 76.86% to 88.5% accuracy and distilled patterns in the form of hierarchical trees composed of 5 to 15 nodes. Experts were largely consistent in their characterization of soft, and to a lesser extent, hard drusen, but diverge in definition of other drusen. Size and crystalline morphology were the main determinants of drusen type across all experts. CONCLUSIONS: Machine learning is a powerful tool for the characterization of disease phenotypes. The creation of a defined feature set for AMD will facilitate the development of an IDOCS-based classification system.