Fundus Photography Research Articles

Evaluating the efficacy of AI systems in diabetic retinopathy detection: A comparative analysis of Mona DR and IDx-DR.

To compare two artificial intelligence (AI)-based Automated Diabetic Retinopathy Image Assessment (ARIA) softwares in terms of concordance with specialist human graders and referable diabetic retinopathy (DR) diagnostic capacity. Retrospective comparative study including 750 consecutive diabetes mellitus patients imaged for non-mydriatic fundus photographs. For each patient four images (45 degrees field of view) were captured, centered on the optic disc and macula. Images were manually graded for severity of DR as no DR, any DR (mild non-proliferative diabetic retinopathy [NPDR] or more), referable DR (RDR (more than mild DR)), or sight-threatening DR (severe NPDR or more severe disease and/or clinically significant diabetic macular edema [CSDME]). IDx-DR and MONA DR output was compared with manual grading and with each other. Total sample size was 750 patients, of which 55 were excluded due to ungradable images. Out of the remaining 695 patients 522 (75%) were considered as having no DR by manual consensus grading, and 106 (15%) as having RDR. Agreement between raters varied between moderate to substantial. IDx-DR showed moderate agreement with human grading (k = 0.4285) while MONA DR had substantial agreement (k = 0.6797). Out of 106 patients with a ground truth of RDR, IDx-DR identified 105 and MONA DR identified 99. The sensitivity and specificity rates for RDR detection of IDx-DR were 99.1 and 71.5% compared with MONA DR of 93.4 and 89.3% respectively. Of note, both ARIAs had 100% sensitivity for the detection of STDR. Both ARIAs performed well in this study population, both with sensitivity for RDR screening over 90%, with IDx-DR showing higher sensitivity and MONA DR higher specificity. MONA DR showed superior agreement with human certified graders.

Retromode imaging in vitreoretinal lymphoma.

The purpose of this study was to evaluate Retromode (Nidek Co., Gamagori, Japan) as an adjunct diagnostic tool in vitreoretinal lymphoma (VRL). By integrating Retromode findings with multimodal imaging, we aimed to enhance lesion detection of VRL lesions in the retina. This monocentric retrospective pilot study was conducted at the Ocular Oncology unit of San Raffaele Scientific Institute, Milan. Six consecutive patients with biopsy-proven vitreoretinal lymphoma (VRL) were analyzed. Comprehensive retinal examinations and multimodal imaging, encompassing Retromode, pseudocolor fundus photography, fundus autofluorescence (FAF), and optical coherence tomography (OCT). The study specifically focused on the role of Retromode findings used in conjunction with traditional imaging modalities like FAF and OCT to enhance lesion detection associated with VRL. Imaging analysis was conducted using ImageJ software. Retromode identified detailed structural abnormalities in the deep retinal layers and RPE, revealing additional or more widespread lesions compared to FAF imaging. Retromode pinpointed areas affected by VRL for further examination with OCT. However, in patients with vitritis-a common condition in VRL-Retromode did not yield quality images or meaningful information. Retromode imaging proved to be a valuable adjunct in the multimodal imaging approach to VRL. Its capacity to delineate subtle retinal changes facilitates a tailored diagnostic strategy, enhancing lesion detection and characterization in VRL.

The performance of optic disc-macula distance to disc-diameter ratio in qualitative and quantitative assessment of the optic disc size.

We aimed to evaluate the performance of the optic disc-macula distance to disc-diameter ratio (DM/DD) in qualitative and quantitative assessment of optic disc size. In 300 apparently normal eyes, we determined the correlation between DM/DD and the planimetric disc area (DA) and evaluated the performance of DM/DD in discriminating between small and large discs. The ability of DM/DD to predict the actual DA was validated in a separate cohort of 200 eyes. The disc size in eyes with glaucoma or related diseases was compared with DA- or DM/DD-based disc size. The DM/DD and its square of inverse (DD2/DM2) showed excellent non-linear and linear correlations, respectively, with DA (p < 0.001 and p < 0.001, respectively), demonstrating their performance in discriminating between small (<2.04 mm2) and large (>2.94 mm2) disc sizes with an area under the receiver operating characteristic curve of 0.901 and 0.959, respectively. The difference between the predicted DA and planimetric DA was within ±10% in more than 80% of the eyes. Eyes with suspected and normal-tension glaucoma exhibited larger discs (p < 0.01) when evaluated using both DM/DD and DA. The variation in disc-macula distance seemed to be a confounding factor for DM/DD in inferring the precise disc size. DM/DD, which can be easily calculated from various kinds of fundus photographs or images, was an excellent and interchangeable tool for qualitative and quantitative assessment of disc size, which did not require correction for ocular magnification.

The protective role of gamma zone peripapillary atrophy in diabetic retinopathy: insights from deep learning and SS-OCT angiography

PurposeTo explore the relationship between peripapillary atrophy (PPA) and diabetic retinopathy (DR), and to uncover potential mechanisms using swept-source optical coherence tomography (SS-OCT) angiography.MethodsThis cross-sectional study included 845 patients with type 2 diabetes (T2DM), who underwent detailed systemic and ophthalmic evaluations. A state-of-the-art deep learning method was employed to precisely identify the parapapillary beta and gamma zones. Based on PPA characteristics, eyes were categorized into four groups: without beta or gamma zone (Group A), isolated beta zone (Group B), isolated gamma zone (Group C), and with both beta and gamma zone (Group D). Digital fundus photography was utilized to diagnose and stage DR severity, while SS-OCT angiography quantified retinal and choroidal vasculature.ResultsParticipants had a mean age of 66 ± 8.8 years, with 437 (51.7%) male. Beta and gamma PPA zones were observed in 574 (67.9%) and 256 (30.3%) eyes, respectively. Beta zone PPA was associated with older age, whereas gamma zone PPA was correlated with longer axial length (AL), lower vessel density, and reduced choroidal thickness. Adjusted analyses revealed that eyes with isolated (Group C) or concurrent (Group D) gamma zone PPA conferred significantly lower DR grade, independent of known risk factors including systemic diabetes risk factors and AL.ConclusionThis study finds that gamma zone PPA is linked to a reduced risk of developing DR. These results imply that the gamma zone may reflect progressive myopia-associated thinning and microvascular depletion in posterior ocular tissues, potentially contributing to structural resilience against DR. This novel insight offers a promising avenue for understanding the interplay between PPA and DR.

Novel Multimodal Imaging in Bilateral Diffuse Uveal Melanocytic Proliferation and Change in the Findings During Follow-up and Treatment.

To describe the advanced multimodal imaging findings of patients with bilateral diffuse uveal melanocytic proliferation (BDUMP) and the changes in these findings over time. Fundus photography, fundus autofluorescence (FAF), fundus fluorescein angiography (FA), spectral domain optical coherence tomography (OCT), B-scan ultrasonography (US), and ultrasound biomicroscopy (UBM) images of the patients diagnosed with BDUMP at a single institution between years 2006-2023 were evaluated. Sixteen eyes of 8 patients with BDUMP were included. The median age was 68 years (Range: 49 - 77 years), and the median follow-up time was 11 months (Range: 1 - 79 months). The most common fundus photography findings were pigmentary changes and diffuse orange pigment patches in the posterior pole. The giraffe/leopard pattern of hyperfluorescence on the posterior pole was the most prominent finding in FAF and FA. Different from previous reports, intraretinal hyperreflective foci, outer retinal tubulations, subretinal fibrosis, and bacillary layer detachment in OCT; optic disc hyperfluorescence, and peripheral vascular leakage in FA; 360° ciliary body thickening, and anterior rotation of the ciliary body in UBM were detected. We presented various novel multimodal imaging features of BDUMP that may warrant updating current diagnostic criteria and their changes over time. Our newly defined findings also suggest a possible inflammatory component to BDUMP pathogenesis, which may have implications for treatment.

Phenotype and genetic spectrum of six Indian patients with bestrophinopathy

Abstract: The aim of this study is to describe genotype and phenotype of patients with bestrophinopathy. The case records were reviewed retrospectively, findings of multimodal imaging such as color fundus photograph, optical coherence tomography (OCT), fundus autofluorescence, electrophysiological, and genetic tests were noted. Twelve eyes of six patients from distinct Indian families with molecular diagnosis were enrolled. Exon 4 of BEST1 was mutated in 3 cases, while exons 2, 3, and 7 in others. Deletion is seen in Exon 7 and missense mutation in other exons. Sporadic autosomal dominant and recessive inheritance was observed in these families. Two patients had primary angle closure glaucoma with a history of consanguineous marriage and glaucoma in the family. Based on our findings, multifocal vitelliform subretinal deposits were the most common fundus finding in patients with autosomal recessive mutation while macular vitelliform lesion was seen with sporadic or autosomal dominant mutation; however, cosegregation analysis was not done. Baseline OCT showed macular and extramacular subretinal exudates, subretinal fluid, intraretinal cystic and schitic spaces, and thickened photoreceptors outer segment tips. Two patients developed abnormal vasculature and focal choroidal excavation in OCT. A severe reduction in the electro-oculogram Ardens ratio was noted while electroretinography was normal. Bestrophinopathy has a varied presentation with complex genotype-phenotype relationships. OCT is a noninvasive tool for monitoring and prognostication. Genetic testing of other family members should be facilitated.

Five-Year Incidence of Age-Related Macular Degeneration and Its Risk Factors in Adult Chinese Population: The Tongren Health Care Study.

To examine the 5-year incidence of age-related macular degeneration (AMD) and its associated factors in an adult Chinese population. The Tongren Health Care Study included individuals attending regular health care check-up examinations in the Beijing Tongren Hospital. Baseline examinations were performed from 2014 to 2015, with 5-year follow-up examinations conducted between 2019 and 2020. Fundus photographs were graded according to the Beckman Initiative guidelines. A total of 5658 participants with gradable photographs at both examinations were included in the study, comprising 58.0% women, with a mean age of 54.9 ± 11.0years. The 5-year incidence of any, early, intermediate, and late AMD were 6.1% (95% confidence interval [CI], 5.5%-6.8%), 5.0% (95% CI, 4.4%-5.6%), 3.4% (95% CI, 2.9%-3.9%), and 0.3% (95% CI, 0.2%-0.4%), respectively. In multivariate analysis, incident early AMD was associated with older age (P < 0.001; odds ratio [OR], 1.04; 95% CI, 1.02-1.06), female sex (P = 0.011; OR, 1.42; 95% CI, 1.08-1.86), and a higher estimated glomerular filtration rate (P = 0.020; OR, 1.15; 95% CI, 1.02-1.30), whereas having diabetes was a protective factor (P = 0.019; OR, 0.61; 95% CI, 0.41-0.92). Incident intermediate AMD was associated with older age (P < 0.001; OR, 1.05; 95% CI, 1.04-1.07), a higher high-density lipoprotein cholesterol level (P < 0.001; OR, 1.97; 95% CI, 1.38-2.83) and a lower triglyceride level (P = 0.008; OR, 0.77; 95% CI, 0.64-0.93). A higher estimated glomerular filtration rate level was a risk factor for incident early AMD. A higher high-density lipoprotein cholesterol level and lower triglyceride level were risk factors for incident intermediate AMD. This finding may point to the role of renal circulation and lipid metabolism in incident AMD. This community-based longitudinal study may provide a valuable understanding of AMD and its associated factors for targeted prevention and management strategies.

Disc-foveal angle and its correlation with retinal layers in cyclovertical strabismic subjects using fundus photography and optical coherence tomography.

Disc-foveal angle and its correlation with retinal layers in cyclovertical strabismic subjects using fundus photography and optical coherence tomography.

Multimodal imaging features of retinal neovascularization in response to 3 anti-VEGF injections

ObjectiveTo analyze longitudinal changes of retinal neovessels (NV) in eyes with proliferative diabetic retinopathy (PDR) after 3 monthly intravitreal injections (IVI) of ranibizumab by means of different imaging modalities, particularly focusing on optical coherence tomography (OCT) and OCT–angiography (OCT-A) features. MethodsProspective, monocentric study, Luigi Sacco Hospital, Milan. Consecutive patients with PDR were enrolled. All patients underwent color fundus photography (CFP), fluorescein angiography (FA), SD-OCT, and OCT-A at baseline (t0) and after 3 monthly IVI of ranibizumab (t1). Neovessels (NVs) were classified as flat, elevated, flat-elevated, or table-top using OCT scans and further analyzed on FIJI. Qualitative and quantitative NV characteristics were evaluated. Vessel density (VD) was calculated as the mean of all the white pixels of the NV and expressed as a percentage of the area of the NV (VD = NV area × 100). ResultsThirty-six NVs in 10 patients were studied. The regression of NVs at t1 was partial in 66.7% of cases and complete in 33.3%. Table-top NV demonstrated more frequently a complete regression (P = .03). A significant reduction of the NV area, perimeter, and VD was observed at t1 (P < .001). NVs that showed a complete regression had higher mean VD values at t0 compared with NVs, which showed a partial regression (P = .02). Flat NVs had more frequently a complete vitreous attachment, while table-top showed more often a partial vitreous detachment with focal NV adhesion (P < .05). A complete regression was more often observed for NVs with a partial vitreous detachment and focal NV adhesion at t0 (10/12), while most of the NVs with a complete vitreous attachment showed a partial regression (18/24) (P < .001). ConclusionsThe table-top configuration of the neovessel, higher VD values at baseline, and the presence of a focal adhesion of the vitreous over the NV were factors associated with a higher risk of complete regression of the NV in response to 3 monthly IVI of ranibizumab.

High dynamic range widefield fundus photography with transcranial illumination: A contrast agent-free method for non-mydriatic choroidal imaging

High dynamic range widefield fundus photography with transcranial illumination: A contrast agent-free method for non-mydriatic choroidal imaging

Glaucoma Screening and Referral Risk Factors in a High-Risk Population: Follow-up Study of the Manhattan Vision Screening Study (NYC-SIGHT)

Précis: Community-based eye health screenings that incorporated fundus photography and optometric exams in a high-risk NYC population effectively identified a higher than average number of participants that required an in-office glaucoma evaluation. Purpose: To report glaucoma screening rates and risk factors associated with referral for in-office glaucoma evaluation in the Manhattan Vision Screening and Follow-up Study (NYC-SIGHT). Methods: In this 5-year, cluster-randomized clinical trial, eligible individuals aged 40 and older were recruited from affordable housing developments and senior centers. Visual acuity with correction, intraocular pressure (IOP) measurements, and non-mydriatic fundus photography were conducted. Images were graded by a glaucoma specialist; those with an abnormal image were referred; those who failed the screening or had an unreadable fundus image were examined by the study optometrist. Chi-squared tests and stepwise multivariate logistic regression analyses were conducted to determine factors associated with glaucoma referral. Results: 708 participants were screened; 189 (26.6%) were referred for an in-office glaucoma evaluation due to an abnormal optic disc image (n=138) or abnormal optometric exam (n=51). Those referred had a mean age 68.5 ±11.7 years and were 60% female, 57% Black, and 37% Hispanic. Stepwise multivariate logistic regression showed participants with self-reported glaucoma (OR 8.096, 95% CI 4.706–13.928, P=0.000), IOP&gt;23 mmHg at the screening (OR 3.944, 95% CI 1.704–9.128, P=0.001), or wore prescription eyeglasses (OR 1.601, 95% CI 1.034–2.48, P=0.035) had higher odds of being referred for an in-office glaucoma evaluation. Of those referred, 106 (56%) attended, 36 participants (34%) were diagnosed with glaucoma and 38 participants (35.8%) as glaucoma suspects. Conclusion: Our findings support public health approaches that focus community-based eye health screenings on high-risk populations and prioritize underserved communities.

Evaluating the reproducibility of a deep learning algorithm for the prediction of retinal age.

Recently, a deep learning algorithm (DLA) has been developed to predict the chronological age from retinal images. The Retinal Age Gap (RAG), a deviation between predicted age from retinal images (Retinal Age, RA) and chronological age, correlates with mortality and age-related diseases. This study evaluated the reliability and accuracy of RA predictions and analyzed various factors that may influence them. We analyzed two groups of participants: Intravisit and Intervisit, both imaged by color fundus photography. RA was predicted using an established algorithm. The Intervisit group comprised 26 subjects, imaged in two sessions. The Intravisit group had 41 subjects, of whom each eye was photographed twice in one session. The mean absolute test-retest difference in predicted RA was 2.39years for Intervisit and 2.13years for Intravisit, with the latter showing higher prediction variability. The chronological age was predicted accurately from fundus photographs. Subsetting image pairs based on differential image quality reduced test-retest discrepancies by up to 50%, but mean image quality was not correlated with retest outcomes. Marked diurnal oscillations in RA predictions were observed, with a significant overestimation in the afternoon compared to the morning in the Intravisit cohort. The order of image acquisition across imaging sessions did not influence RA prediction and subjective age perception did not predict RAG. Inter-eye consistency exceeded 3years. Our study is the first to explore the reliability of RA predictions. Consistent image quality enhances retest outcomes. The observed diurnal variations in RA predictions highlight the need for standardized imaging protocols, but RAG could soon be a reliable metric in clinical investigations.

A Comprehensive Health Screening Program Reveals the Prevalence of and Risk Factors for Age-Related Macular Degeneration: A Cross-Sectional Analysis

Background/Objectives: We investigated the prevalence of age-related macular degeneration (AMD) and associated risk factors in Korean subjects who underwent comprehensive health screening examinations. Methods: This single health screening center-based cross-sectional study included a total of 73,574 consecutive participants older than 30 years who underwent a health screening examination, including fundus photography, between October 2003 and December 2010. Weighted prevalence and risk factors for AMD were evaluated. Logistic regression was used to identify AMD risk factors. Results: The weighted prevalence of AMD was 15.42%, with a prevalence of 3.34% among people in their 30s. Advanced age significantly increased the risk for both early/intermediate AMD (p &lt; 0.001 across the age groups of 40, 50, 60, and 70+ years) and advanced AMD (p &lt;0.001 for the age groups of 60 and 70+ years). The male sex was strongly associated with an increased risk of both early/intermediate and advanced AMD (p &lt; 0.001 for both). Retinal arteriosclerosis, whether low- or high-grade, was linked to early/intermediate AMD (p &lt; 0.001 for both grades), whereas only high-grade arteriosclerosis was linked to advanced AMD (p &lt; 0.001). Additionally, hypertension (p &lt; 0.001), the hepatitis B carrier status (p &lt; 0.001), elevated mean corpuscular volume (p &lt; 0.001), and lower serum uric acid levels (p = 0.014) were associated with early/intermediate AMD. Higher education levels protected against early/intermediate AMD (p = 0.004 for high school graduates, p &lt; 0.001 for ≥college graduates). Higher serum inorganic phosphate levels (p = 0.002) and lower total serum ALB levels (p = 0.005) were significant risk factors for advanced AMD. Conclusions: Korean individuals as young as 30 years old are at risk of AMD. This study newly identified associations between retinal arteriosclerosis and both early/intermediate and advanced AMD, as well as associations between serum inorganic phosphate levels and total ALB levels with advanced AMD.

Retinal vascular alterations are associated with cognitive function and neuroimaging in white matter hyperintensities

AimTo reveal alterations in retinal structure, vessels, and function, and their association with cognitive function and neuroimaging in white matter hyperintensities (WMH). MethodsThis study enlisted WMH and age-matched healthy controls (HC). All participants underwent six different tests: magnetic resonance imaging (MRI) of the brain, the Mini-Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), fundus photography, optical coherence tomography (OCT), and visual field testing. Visual field can reflect the function of optic nerve and retina. The peripapillary retinal nerve fiber layer (p-RNFL) was analyzed using OCT. Image J software was employed to measure retinal vascular caliber in fundus photographs and to compute the central retinal artery equivalent (CRAE), central retinal venous equivalent (CRVE) and arteriole-to-venule ratio (AVR). ResultsA total of 90 WMH patients and 93 HC participants. In comparison with the HC, the WMH group exhibited reduced cognitive function scores (MoCA: P < 0.001; MMSE: P < 0.001), narrower retinal arteries (P < 0.001), smaller AVR (P < 0.001) and thinner p-RNFL thickness (total: P = 0.026; temporal: P = 0.006). About visual field, both univariate and multivariate analysis showed that mean sensitivity decreased, and mean defect increased in WMH group (P < 0.05). Additionally, correlation analysis indicated a positive correlation between CRAE and AVR with MMSE and MoCA score (r = 0.424–0.57, P < 0.001) and a negative correlation with Fazekas score (CRAE: r = −0.515, P < 0.001; AVR: r = −0.554, P < 0.001), and p-RNFL was negatively correlated with Fazekas score (total p-RNFL: r = −0.192, P = 0.009; temporal p-RNFL: r = −0.217, P = 0.003). Notably, no significant correlation was found between cognitive function and p-RNFL. ConclusionWMH group exhibit narrower retinal arteries, smaller arteriole-to-venule ratio, damaged p-RNFL and visual function. These alterations in retinal vessels are associate with both neuroimaging and cognitive function. Our results suggest that retinal imaging could serve as a valuable instrument for evaluating WMH and provides some new approaches to study the characteristic markers of WMH.

Prediction of Visual Acuity After Cataract Surgery by Deep Learning Methods Using Clinical Information and Color Fundus Photography

Purpose To examine the performance of deep-learning models that predicts the visual acuity after cataract surgery using preoperative clinical information and color fundus photography (CFP). Methods We retrospectively collected the age, sex, and logMAR preoperative best corrected visual acuity (preoperative-BCVA) and CFP from patients who underwent cataract surgeries from 2020 to 2021 at National Taiwan University Hospital. Feature extraction of CFP was performed using a pre-existing image classification model, Xception. The CFP-extracted features and pre-operative clinical information were then fed to a downstream neural network for final prediction. We assessed the model performance by calculating the mean absolute error (MAE) between the predicted and the true logMAR of postoperative BCVA. A nested 10-fold cross-validation was performed for model validation. Results A total of 673 fundus images from 446 patients were collected. The mean preoperative BCVA and postoperative BCVA was 0.60 ± 0.39 and 0.14 ± 0.18, respectively. The model using age and sex as predictors achieved an MAE of 0.121 ± 0.016 in postoperative BCVA prediction. The inclusion of CFP as additional predictor in the model (predictors: age, sex and CFP) did not further improve the predictive performance (MAE = 0.120 ± 0.023, p = 0.375), while adding the preoperative BCVA as an additional predictor resulted in a 4.13% improvement (predictors: age, sex and preoperative BCVA, MAE = 0.116 ± 0.016, p = 0.013). Conclusions Our multimodal models including both CFP and clinical information achieved excellent accuracy in predicting BCVA after cataract surgery, while the learning models input with only clinical information performed similarly. Future studies are needed to clarify the effects of multimodal input on this task.

Characteristics of submacular hemorrhage with bacillary layer detachment and intrabacillary hemorrhage.

To study the outcomes of patients presenting with submacular hemorrhage (SMH) and bacillary layer detachment (BALAD) following intervention. This retrospective study examined fundus photographs and optical coherence tomography (OCT) scans to identify treatment-naive SMH and BALAD cases. Two groups were formed: SMH cases with and without BALAD. The treatment outcomes of these cases were assessed. Thirteen (65%) of the 20 eyes with SMH had BALAD. Blunt trauma was the most common cause of SMH (n = 10, 50%). Median age was 46 years (IQR range: 28-70). Demographic, clinical, or OCT imaging findings between the groups (p > 0.05) were comparable. Nine (45%) patients each underwent intravitreal gas injection alone or along with PPV and TpA injection. At 1-month post-treatment, VA improved (logMAR VA - 0.89; p = 0.017). BALADs and intrabacillary hemorrhage had resolved in eight (61%) cases. Intraretinal, subretinal, and sub-RPE fluids resolved in 100%, 70%, and 43% of cases, respectively. Eyes with resolved BALAD showed greater improvement in VA (logMAR VA - 0.98) than eyes with persisting BALAD (logMAR VA - 1.1) and resulted in significant decrease in central macular (p = 0.016) and retinal thicknesses (p = 0.031). SMH eyes without pre-treatment BALAD also observed statistically significant improvement in visual acuity following intervention (p = 0.031). BALAD and intrabacillary hemorrhage in SMH are relatively common. A significant proportion of cases had their BALAD and intrabacillary haemorrhage resolved following treatment. Persistence of BALAD had no effect on VA.

Evaluating the Impact of Retinal Vessel Segmentation Metrics on Retest Reliability in a Clinical Setting: A Comparative Analysis Using AutoMorph.

Current research on artificial intelligence-based fundus photography biomarkers has demonstrated inconsistent results. Consequently, we aimed to evaluate and predict the test-retest reliability of retinal parameters extracted from fundus photography. Two groups of patients were recruited for the study: an intervisit group (n = 28) to assess retest reliability over a period of 1 to 14days and an intravisit group (n = 44) to evaluate retest reliability within a single session. Using AutoMorph, we generated test and retest vessel segmentation maps; measured segmentation map agreement via accuracy, sensitivity, F1 score and Jaccard index; and calculated 76 metrics from each fundus image. The retest reliability of each metric was analyzed in terms of the Spearman correlation coefficient, intraclass correlation coefficient (ICC), and relative percentage change. A linear model with the input variables contrast-to-noise-ratio and fractal dimension, chosen by a P-value-based backward selection process, was developed to predict the median percentage difference on retest per image based on image-quality metrics. This model was trained on the intravisit dataset and validated using the intervisit dataset. In the intervisit group, retest reliability varied between Spearman correlation coefficients of 0.34 and 0.99, ICC values of 0.31 to 0.99, and mean absolute percentage differences of 0.96% to 223.67%. Similarly, in the intravisit group, the retest reliability ranged from Spearman correlation coefficients of 0.55 and 0.96, ICC values of 0.40 to 0.97, and mean percentage differences of 0.49% to 371.23%. Segmentation map accuracy between test and retest never dropped below 97%; the mean F1 scores were 0.85 for the intravisit dataset and 0.82 for the intervisit dataset. The best retest was achieved with disc-width regarding the Spearman correlation coefficient in both datasets. In terms of the Spearman correlation coefficient, the worst retests of the intervisit and intravisit groups were tortuosity density and artery tortuosity density, respectively. The intravisit group exhibited better retest reliability than the intervisit group (P < 0.001). Our linear model, with the two independent variables contrast-to-noise ratio and fractal dimension predicted the median retest reliability per image on its validation dataset, the intervisit group, with an R2 of 0.53 (P < 0.001). Our findings highlight a considerable volatility in the reliability of some retinal biomarkers. Improving retest could allow disease progression modeling in smaller datasets or an individualized treatment approach. Image quality is moderately predictive of retest reliability, and further work is warranted to understand the reasons behind our observations better and thus ensure consistent retest results.

Blood integrin- and cytokine-producing T cells are associated with stage and genetic risk score in age-related macular degeneration

Age-related macular degeneration (AMD) remains a leading cause of vision loss in the geriatric population. There are age-related changes in peripheral blood leukocyte composition, but their significance for AMD remains unclear. We aimed to determine changes in immune cell populations in the blood of AMD patients. A standardized 31-parameter flow cytometry analysis was conducted on peripheral blood mononuclear cells from 59 patients with early and advanced AMD and 39 controls without AMD, all older than 65 years. Fundus photography and optical coherence tomography were used to classify disease stages and a custom genotype array was used to compute an AMD genetic risk score based on 52 AMD disease risk variants (GRS-52). A generalized linear regression model corrected for age, sex, and smoking status revealed that AMD patients showed decreased frequencies of CD4+ T helper cell population expressing Integrin Alpha E (CD103) (Padj = 0.019). We further noted that early AMD was characterized by increased interleukin-4 (IL-4)-producing CD4+ T helper cells (Padj = 0.013; <0.001), as well as IL-4-producing cytotoxic CD8+ T cells (Padj = 0.016; <0.001). Reclassification of samples based on the GRS-52 revealed that IL-17-producing T cells decreased incrementally across GRS-52 categories. In AMD, alterations in peripheral blood leukocyte populations are associated with genetic risk score and disease stage and include specifically IL-4 and IL-17A cytokine-producing and CD103 integrin-expressing T cell populations.

Alcohol Consumption and Risk of Age-Related Macular Degeneration and Geographic Atrophy Progression: AREDS2 Report 34

PurposeTo examine potential relationships between alcohol consumption and age-related macular degeneration (AMD) progression, including progression to late AMD and geographic atrophy (GA) enlargement rate. DesignPost hoc analysis of cohorts within the Age-Related Eye Diseases Study 2 (AREDS2). Participants6670 eyes (of 3673 participants) with no late AMD at baseline; 1143 eyes (of 841 participants) with GA at ≥2 consecutive visits. MethodsColor fundus photographs were collected at annual study visits and graded centrally for late AMD, GA area, and GA proximity. Alcohol consumption was calculated by food frequency questionnaire. Regression analyses of disease progression were performed according to alcohol consumption.Main outcome measures:Progression to late AMD and its subtypes; GA area-based progression; GA proximity-based progression. ResultsOver mean follow-up of 3.8 years, 40.2% of eyes progressed to late AMD. In men, with alcohol tertile 1 (no regular consumption) as reference, hazard ratios for progression to late AMD were 0.69 (95% CI 0.55-0.87, p=0.0015) for tertile 2 and 0.85 (0.71-1.02, p=0.079) for tertile 3. In women, hazard ratios were 1.12 (0.95-1.31, p=0.17) and 0.85 (0.72-1.00, p=0.046), respectively. Over mean follow-up of 3.1 years, GA area-based progression was significantly faster in women than men, at 0.295 (95% CI 0.278-0.311) and 0.260 mm/year (0.241-0.279), respectively (p=0.007). In men, area-based progression differed significantly by alcohol tertile (p=0.0001), at 0.275 (0.248-0.303), 0.183 (0.143-0.223), and 0.280 mm/year (0.254-0.306) in tertiles 1-3, respectively. In women, the area-based rate did not differ significantly by alcohol tertile (p=0.11). In men only, CDC-defined heavy drinking was associated with faster progression (p=0.024), at 0.306 (0.262-0.349) vs 0.252 mm/year (0.233-0.270). In 808 eyes with non-central GA, GA proximity-based progression did not differ significantly by alcohol tertile (p=0.55). ConclusionsModerate alcohol consumption is associated with decreased risk of progression to late AMD in men. GA progression is faster in women, but its relationship with alcohol consumption is much stronger in men. In men, moderate consumption is associated with slower GA progression and higher consumption with faster progression. Although some of these associations may also relate to confounding, they might suggest that individuals with GA should avoid high alcohol consumption.

Performance of a Deep Learning System and Performance of Optometrists for the Detection of Glaucomatous Optic Neuropathy Using Colour Retinal Photographs.

Glaucoma is the leading cause of irreversible blindness, with a significant proportion of cases remaining undiagnosed globally. The interpretation of optic disc and retinal nerve fibre layer images poses challenges for optometrists and ophthalmologists, often leading to misdiagnosis. AI has the potential to improve diagnosis. This study aims to validate an AI system (a convolutional neural network based on the Inception-v3 architecture) for detecting glaucomatous optic neuropathy (GON) using colour fundus photographs from a UK population and to compare its performance against Australian optometrists. A retrospective external validation study was conducted, comparing AI's performance with that of 11 AHPRA-registered optometrists in Australia on colour retinal photographs, evaluated against a reference (gold) standard established by a panel of glaucoma specialists. Statistical analyses were performed using sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC). For referable GON, the sensitivity of the AI (33.3% [95%CI: 32.4-34.3) was significantly lower than that of optometrists (65.1% [95%CI: 64.1-66.0]), p < 0.0001, although with significantly higher specificity (AI: 97.4% [95%CI: 97.0-97.7]; optometrists: 85.5% [95%CI: 84.8-86.2], p < 0.0001). The optometrists demonstrated significantly higher AUROC (0.753 [95%CI: 0.744-0.762]) compared to AI (0.654 [95%CI: 0.645-0.662], p < 0.0001). The AI system exhibited lower performance than optometrists in detecting referable glaucoma. Our findings suggest that while AI can serve as a screening tool, both AI and optometrists have suboptimal performance for the nuanced diagnosis of glaucoma using fundus photographs alone. Enhanced training with diverse populations for AI is essential for improving GON detection and addressing the significant challenge of undiagnosed cases.