BACKGROUND/OBJECTIVE:Ophthalmologists rely on clinical practice guidelines to inform their practice. Not all guidelines, however, are considered trustworthy. For guidelines to be credible and evidence-based, internationally-accepted standards suggest that they must adhere to certain criteria. We aim to identify strengths and limitations of commonly used ophthalmology guidelines and provide insights to develop trustworthy, fit-for-purpose guidelines in the field. METHODS:We will search MEDLINE, Embase, Trip Database, the Royal College of Ophthalmologists ( https://www.rcophth.ac.uk/ ) and American Academy of Ophthalmology ( https://www.aao.org/clinical-statements ) websites. From these sources, we will include a purposeful sample 100 guidelines published from 2015 onwards addressing diverse areas in ophthalmology (e.g., cornea, glaucoma, retina, paediatrics, etc.) - forty guidelines from major eye care organisations and sixty from diverse ophthalmology subspecialties and regions. We will appraise each guideline according to the modified National Guideline Clearinghouse Extent Adherence to Trustworthy Standards (NEATS) instrument. CONCLUSION:By identifying best practices and common pitfalls in existing guidelines, this work will provide a roadmap for creating more robust, transparent, and patient-centred clinical practice guidelines in ophthalmology.

To evaluate the feasibility of integrating genetic, imaging, and demographic data for predictive modelling of treatment outcomes in neovascular age-related macular degeneration (nAMD). Proof-of-concept retrospective cohort study with prospective DNA collection. Patients with unilateral nAMD receiving anti-vascular endothelial growth factor (anti-VEGF) therapy on a treat-and-extend regimen at a single tertiary centre were recruited. Polygenic risk scores (PRS) for AMD were derived from genotyping data (NIHR Bioresource). Optical coherence tomography (OCT) biomarkers-intraretinal fluid (IRF), subretinal fluid (SRF), pigment epithelial detachment (PED), and subretinal hyperreflective material (SHRM)-were automatically quantified using a deep learning segmentation model. Predictors of treatment outcomes included PRS, age at first injection, and OCT feature volumes at baseline. XGBoost was used for binary outcomes and linear regression for continuous outcomes, employing five-fold cross-validation. (1) macular dryness (no IRF/SRF) at 24 months, (2) average treatment interval in year 2, and (3) age at first injection. 106 participants were included. The multimodal model integrating age, imaging, and PRS predicted macular dryness at 24 months with AUC = 0.903, outperforming imaging alone (AUC = 0.701). PRS was associated with younger age at first injection (β = –4.69, 95% CI [–8.93, –0.44], P = 0.031) but not with treatment burden (β = –6.39, P = 0.13). Integrating PRS with OCT-derived imaging biomarkers and patient age is technically feasible and improves predictive performance of modelling for anatomical treatment outcomes in nAMD. PRS reflects genetic susceptibility to nAMD and contextualizes the predictive value of imaging biomarkers for treatment response.

Provide expert-informed guidance on the integration of aflibercept 8 mg into routine clinical care for patients with neovascular age-related macular degeneration (nAMD) and diabetic macular oedema (DMO), based on clinical trial and real-world evidence. Sixteen retina specialists ('experts') met during the Association for Research in Vision and Ophthalmology (ARVO) annual meeting in May 2025 to discuss the clinical use of aflibercept 8 mg. A post-meeting online platform was used to collect additional independent input. Agreement was captured using binary responses and categorised as: all agreed (100%), majority agreed ( ≥ 75%), most agreed (50-74%), or non-consensus ( ≤ 50%). All experts agreed that aflibercept 8 mg may be considered as a first-line treatment in treatment-naïve patients with nAMD and DMO, and switching to aflibercept 8 mg can be considered for previously treated patients. The majority of experts agreed that switching patients with well-controlled disease on shorter or intermediate dosing intervals may enable further interval extension while maintaining disease control, and switching patients with poorly controlled disease may optimise anatomical outcomes and improve stability. Most experts agreed that three initial monthly loading doses are appropriate in newly diagnosed patients. No consensus was reached on optimal dosing-decision strategies after switching in patients with poorly controlled disease, nor on the timing of initial interval modification during treatment initiation. These expert opinions, informed by available, emerging clinical evidence and real-world clinical experience, support aflibercept 8 mg as a first-line and switch option for nAMD and DMO, using personalised treatment approaches for both treatment-naïve and previously treated patients.

Age-related macular degeneration (AMD) is a progressive retinal disease and a major cause of visual impairment in older adults. Non-neovascular AMD (non-nAMD), the more common phenotype, can substantially impair vision-related quality of life (QoL) even when visual acuity remains relatively preserved. The 25-item National Eye Institute Visual Function Questionnaire (NEI-VFQ-25) is the most widely used instrument for assessing vision-related QoL in AMD. However, QoL in non-nAMD has not been systematically synthesised across disease stages and settings. This review aims to summarise vision-related QoL in treatment-naive patients with non-nAMD, examine how QoL varies by disease stage and measurement instrument and identify study-level factors associated with QoL differences. We will search MEDLINE, Embase, EBM Reviews and APA PsycINFO from inception for original studies reporting mean values or mean differences, with corresponding variability, for overall or subscale scores from the NEI-VFQ-25 or other validated vision-related QoL instruments. Two reviewers will independently screen studies, extract data and assess risk of bias using the appropriate Joanna Briggs Institute critical appraisal checklist for analytical cross-sectional studies. Random-effects meta-analyses will be conducted to pool NEI-VFQ-25 composite and subscale scores by non-nAMD stage and, where available, to estimate differences between stages or versus controls. Instrument-specific meta-analyses will be performed for other QoL measures when feasible and standardised metrics will be used to combine instruments assessing similar constructs but using different scales. Meta-regression will explore whether QoL varies with study-level characteristics. The findings will clarify the functional burden of non-nAMD and support future clinical research and trial design.

PURPOSE:To explore the clinical and multimodal imaging characteristics of intraretinal fibrosis (IRFib) in macular telangiectasia type-2 (MacTel). METHODS:Macular telangiectasia type-2 eyes with IRFib that had multimodal imaging with color fundus photography, multicolor, blue reflectance, spectral-domain optical coherence tomography, optical coherence tomography angiography, and fluorescein angiography. RESULTS:Seven eyes of six patients were included. Clinically, all cases had grayish-white fibrosis at macula, with retinal vessel tortuosity in two cases best characterized on multicolor. Two eyes each belonged to Gass and Blodi stages 3 and 4, three eyes were stage 5 MacTel. The mean lesion area at presentation was 1.36 mm 2 ± 0.81 mm 2 (range 0.56-2.99 mm 2 ). The defining optical coherence tomography finding for IRFib was hyper-reflective distortion intraretinally (involving different layers). Other accompanying findings in decreasing order were collapse sign (seven eyes); epiretinal membrane (six eyes); pigment clumping and migration (five eyes); serrated internal limiting membrane and outer-retinal hyper-reflectivity (four eyes); and lamellar macular hole (one eye). On optical coherence tomography angiography, IRFib was associated with right-angled vessels, vascular invasion and distortion of foveal avascular zone, and vessel dilatation and tortuosity in superficial vascular complex; telangiectasia, foveal avascular zone distortion and invasion, vessel drag, capillary looping, and abnormal vascular structures in deep capillary plexus; and nodular tuft or twig-like hyper-reflective vascular lesions in outer-retina choriocapillaris slab. CONCLUSION:The incidence of IRFib on multicolor in our MacTel cohort was 1.75% with characteristic findings on optical coherence tomography and optical coherence tomography angiography. They are associated with advanced MacTel and probably implicate Müller cell dysfunction, activated glial cells, and vascular changes during the disease process.

BACKGROUND/OBJECTIVES:The aim was to evaluate the incidence of late AMD in eyes with intermediate AMD (iAMD) over two years. SUBJECTS/METHODS:In this multicentre cohort study, participants with iAMD were classified as: (i) iAMD without incomplete retinal pigment epithelium or outer retinal atrophy (iRORA) or subretinal drusenoid deposits (SDD); (ii) iAMD with SDD with no iRORA; (iii) iAMD with iRORA with no SDD and (iv) iAMD with iRORA and SDD. Factors associated with the incidence of late AMD were analysed using an interval-censored Weibull parametric proportional hazards model. RESULTS:983 eyes from 805 participants were analysed. Model-based cumulative incidence at nominal 2 years was 13.1% (95% CI 10.8-15.6%) for late AMD, 7.6% (95% CI 5.9-9.3%) for nAMD as the cause and 5.5% (95% CI 4.1-7.3%) for cRORA as the cause. Participants aged 75-84 (adjusted HR [aHR] 1.66, 95% CI 1.13-2.46; P = 0.01), 85 years+ (aHR 2.48, 95% CI 1.49-4.15; P = 0.001), presence of iRORA (aHR 1.50, 95% CI 1.05-2.16; P = 0.03), fellow-eye nAMD (aHR 3.23, 95% CI 2.04-5.12; P < 0.001) and BRVA 70-79 (vs 80 or better ETDRS letters: aHR 1.65, 95% CI 1.09-2.49; P = 0.02) had increased hazard of late AMD. Eyes that transitioned to neovascular AMD (nAMD) by 2 years (within tolerance window) displayed the greatest loss in BRVA compared to other phenotypes (-6.01 letters, SD 11.35). CONCLUSIONS:Approximately 13% of eyes with iAMD progressed to late AMD within 2 years. Eyes that transitioned to nAMD experienced the greatest loss in BRVA.

Importance Screening for diabetic retinopathy using fundus photographs is the global standard of care but results in high false-positive referrals to evaluate diabetic macular edema (DME), placing a substantial burden on specialist eye clinics. Integrating an AI-based optical coherence tomography (AI-OCT) system into screening pathways may reduce potentially unnecessary referrals. Objective To evaluate the diagnostic and referral performance of an AI-OCT system for DME detection within a diabetic retinopathy screening pathway in clinical settings. Design, Setting, and Participants Stepwise evaluation conducted in Hong Kong Special Administrative Region: a prospective silent-mode validation (February 2020 to July 2023) recruiting 603 patients with diabetes at a tertiary hospital triage unit, followed by a multicenter noninferiority RCT (September 2023 to April 2025), with follow-up completed in May 2025, recruiting 276 patients with suspected DME referred from a territory-wide diabetic retinopathy screening program. Interventions RCT participants were randomized to intervention (referral for DME evaluation based on both fundus photograph–based screening reports and AI-OCT reports [n = 137]) or control (automatic referral based solely on fundus photograph–based screening reports [n = 139]) groups. The AI-OCT system incorporated image-quality assessment, DME detection, and uncertainty flagging. Study outcomes focused on referral rates under the 2 pathways; for ethical reasons, all participants ultimately underwent specialist evaluation. Main Outcomes and Measures The primary outcome was false-positive DME referral rate, with a prespecified noninferiority margin of 20%. The secondary outcomes included sensitivity and specificity for DME detection and DME referral. Results In prospective silent-mode validation (mean age, 64.7 [SD, 9.4] years; 56.2% male), 86 of 1200 scans (7.2%) were identified as ungradable and 49 of 1114 gradable scans (4.4%) were classified as uncertain. The system achieved 98.8% (95% CI, 94.5%-100.0%) sensitivity and 90.7% (95% CI, 88.7%-92.4%) specificity for DME detection. In the RCT (mean age, 63.9 [SD, 10.9] years; 54.7% male), DME prevalence was similar in the intervention and control groups (30.9% vs 29.9%). The false-positive DME referral rate was 24.1% (95% CI, 14.6%-37.0%) and 69.1% (95% CI, 61.0%-76.1%), respectively (absolute difference, −45% [95% CI, −58.2% to −31.9%; P < .001 for noninferiority]; upper bound of the CI below the prespecified noninferiority margin of 20%). Sensitivity for DME referral was 100.0% (95% CI, 100.0%-100.0%) in both groups. Specificity for DME referral was 86.5% (95% CI, 79.3%-92.9%) in the intervention group and 0.0% (95% CI, 0.0%-0.0%) in the control group. No cases of DME occurred among nonreferred participants in the intervention group. Conclusions and Relevance Compared with standard practice, incorporation of the AI-OCT system as a secondary screening tool was noninferior with respect to false-positive referral rates and was associated with a substantial reduction in potentially unnecessary DME referrals without compromising sensitivity. Trial Registration Chinese Clinical Trial Registry: ChiCTR2300075087

BACKGROUND/OBJECTIVES:To determine the topographic distribution of retinal non-perfusion (RNP) progression in patients with severe non-proliferative diabetic retinopathy (sNPDR) and proliferative diabetic retinopathy (PDR). SUBJECTS/METHODS:This retrospective study was conducted at Moorfields Eye Hospital in the United Kingdom. Patients with diabetic retinopathy (DR) who had two sets of ultra-widefield fundus fluorescein angiography were included. A grid comprising seven concentric rings was superimposed on the images. RNP areas were annotated using Optos' proprietary tool. Total and zonal non-perfusion index (NPI) were calculated as the RNP area divided by the gradable retinal area. Adjusted linear mixed-effects models (LMEMs) were used to estimate progression, accounting for repeated measures and inter-eye correlation. RESULTS:A total of 53 patients (67 eyes) were included in the analyses. This cohort comprised 19 (28%) eyes with PDR, 31 (46%) with sNPDR, and 17 (25%) with mild-to-moderate NPDR. Focusing on sNPDR and PDR, the nasal 20-30 mm zone had the largest NPI (0.37 ± 0.26) at baseline. Over time, the total NPI increased significantly, with an estimated change of +0.027 (12.42 mm²) per year (95% confidence interval [CI] = 0.008 to 0.045, P = 0.005). The inferior 10-20 mm ring showed the fastest NPI progression ( + 0.049/year; P = 0.007). CONCLUSIONS:This study demonstrates that NPI increased with time in DR. Notably, the inferior 10-20 mm zone was most vulnerable to NPI progression. Of note, this study was limited by the confounding effect of anti-vascular endothelial growth factor (anti-VEGF) treatment; therefore, a future prospective study to control for treatment exposure would be warranted.

PURPOSE:To examine multimodal imaging findings in eyes before and after the onset of exudative macular neovascularization (ExMNV) bypassing the stage of pigmentation in Macular Telangiectasia Type-2. METHODS:From a total of 798 patients, Macular Telangiectasia Type-2 eyes with ExMNV that exhibited no signs of focal hyperpigmentation on color fundus photograph, fundus autofluorescence, and spectral-domain optical coherence tomography were retrospectively evaluated. RESULTS:There were 16 eyes of 16 patients (2%) with a mean age of 50.60 ± 9.82 years. Among them, 11 eyes had ExMNV at the time of examination, whereas the remaining five eyes developed ExMNV during the follow-up period, all lacking pigment when it occurred. The average time for ExMNV development in the five eyes was 27.6 months (range: 11-46 months). Preceding signs of ExMNV included right-angled vessels and telangiectatic leaks causing the following optical coherence tomography changes-temporal parafoveal thickening in three eyes, minimal intraretinal fluid pockets in three eyes, convex hyporeflective subretinal space in two eyes, and subretinal fluid in two eyes. With further declension/descent of inner into outer retina, progression of ellipsoid zone loss was seen and outer retinal hyperreflectivity formed in areas with preexisting ellipsoid zone loss with visual acuity decline. Multiple retinal vessels formed an anastomosis with each other and extended into the center of the subretinal neovascular lesion. CONCLUSION:Patients with early Macular Telangiectasia Type-2 can proceed directly to ExMNV without pigmentation. Recognizing the signs before ExMNV develop will allow for closer follow-up and earlier treatment of these patients to prevent irreversible visual loss.

Diabetic retinal disease remains a leading cause of vision loss, with current treatment limited by frequent injections and poor adherence. Ongoing clinical trials are evaluating topical therapies, sustained delivery systems, and gene therapy as alternatives to current gold standard treatments. The field is shifting toward earlier, more personalized, and more durable disease-modifying strategies.

Topic:This systematic review examined how vision-related patient-reported outcome measures (PROMs) were selected, operationalized, and reported at baseline in randomized controlled trials (RCTs) enrolling patients across the spectrum of age-related macular degeneration (AMD). Clinical Relevance:Age-related macular degeneration is a major cause of visual disability, and visual acuity alone does not fully capture its effects on everyday functioning and quality of life. Inconsistent selection and incomplete reporting of PROMs may limit interpretation, comparison across trials, and evidence synthesis. Methods:A systematic search of Medical Literature Analysis and Retrieval System Online, Embase, Web of Science Core Collection, and Scopus was conducted from database inception to September 27, 2023, supplemented by a targeted update for relevant 2024 and 2025 publications. Eligible studies were RCTs enrolling ≥20 participants with AMD and reporting both baseline visual acuity and baseline vision-related PROM data. Data on PROM instrument selection, scoring, baseline distributions, laterality context, and subgroup reporting were extracted and synthesized descriptively. Results:Thirty-eight RCTs met the inclusion criteria, spanning early through late-stage AMD. Across included trials, 11 distinct PROM instruments were reported. The National Eye Institute Visual Function Questionnaire family predominated (78.9%), with the National Eye Institute Visual Function Questionnaire‑25 being the most commonly used version (n = 17). Baseline patient-reported functional status varied by disease stage: early and intermediate AMD cohorts reported higher scores (75.3-89.1), whereas trials enrolling neovascular AMD patients showed lower and more heterogeneous scores (59.1-77.9). Subscale and domain reporting was inconsistent, with many trials presenting only composite scores or a limited subset of subscales. Baseline PROM results were not reported by subgroup, preventing comparisons across patient populations. Reporting of visual function was incomplete with only 13% of studies reporting both better-seeing and worse-seeing eyes. Conclusion:The interpretive value of PROMs in AMD trials is constrained by inconsistent baseline reporting and limited reporting of both-eye and binocular acuity needed to contextualize patient-level function. Enhanced consistency in stage-appropriate PROM selection and routine reporting of baseline distributions are essential to support patient-centered evaluation and evidence synthesis in AMD research. Financial Disclosures:Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

PURPOSE:To report the incidence of retinal pigment epithelial (RPE) tears following three monthly 2-mg aflibercept injections in treatment-naïve neovascular age-related macular degeneration (nAMD) and to identify associated risk factors. METHODS:Treatment-naïve patients aged ≥50 years with nAMD and baseline visual acuity (VA) between 24 and 78 Early Treatment of Diabetic Retinopathy Study (ETDRS) letters were included. RPE tears were diagnosed on optical coherence tomography (OCT) as a definite discontinuity of the RPE. Logistic regression using Generalized Estimating Equations (GEE) was performed to account for correlated eye data and identify factors associated with RPE tear development. Cluster bootstrap receiver operating characteristic (ROC) curve analysis assessed the predictive value of baseline pigment epithelial detachment (PED) height. RESULTS:A total of 1,929 eyes were analyzed; 67 eyes (3.5%) developed an RPE tear after the aflibercept loading phase. Multivariable analysis adjusted for age, ethnicity, and sex showed that male sex, poorer baseline VA, increased central subfield thickness, hemorrhagic PED, increased PED height, presence of subretinal hyperreflective material (SHRM), and polypoidal choroidal vasculopathy (PCV) were associated with higher odds of RPE tear and that presence of drusen with decreased odds of RPE tear. ROC analysis demonstrated that a baseline PED height of 244 µm optimally predicted RPE tear development with a sensitivity of 73.8% and a specificity of 73.6%. CONCLUSION:RPE tears occurred in 3.5% of treatment-naïve nAMD eyes after aflibercept loading phase. Several baseline clinical and imaging risk factors were identified to be significantly associated with RPE tear development.

IMPORTANCE:Nonexudative macular neovascularization (neMNV) is a risk factor for exudation in fellow eyes of patients with unilateral exudative age-related macular degeneration (AMD). Accurate estimation of prevalence of neMNV is required for sample size calculations for clinical trials evaluating novel interventions to delay exudation from neMNV. Certain dimensions of double-layer sign (DLS) on optical coherence tomography (OCT) are often regarded as a surrogate for neMNV. OBJECTIVE:To assess the prevalence of fellow-eye neMNV on OCT angiography (OCT-A) in eyes with AMD with DLS on OCT among patients with unilateral new-exudative AMD in the first eye. Design, Setting, and Participants:This prospective observational cohort multicenter study took place in the United Kingdom from January 2021 through June 2025. The study included 862 participants with OCT and OCT-A within 3 months of initiation of anti-vascular endothelial growth factor (VEGF) therapy, of whom 550 (63.8%) had both OCT and OCT-A performed within 30 days of the date of the first anti-VEGF injection to the fellow eye. These data were analyzed from July 2025 through October 2025. MAIN OUTCOMES AND MEASURES:Prevalence of DLS, neMNV, and assessment of univariable and adjusted associations with neMNV and DLS were assessed using logistic regression models. RESULTS:Among 550 eyes (mean [SD] participant age, 78.0 [7.6] years; 315 female [57.3%] and 235 male [42.7%]), 112 (20.4%; 95% CI [Wilson score], 17.1%-24.0%) had DLS and 47 eyes (8.5%; 95% CI [Wilson score], 6.4%-11.3%) had neMNV at baseline, including 42 (89.36%) within DLS, 3 (6.4%) in fibrovascular irregular shallow-pigment epithelial detachment, and 2 (4.3%) below drusen. Prevalence of neMNV was higher among eyes with thick DLS (n = 36 [48%]) compared with those with thin DLS (16.2%; n = 6; difference = -31.8%; 95% CI [Wilson score], -46.9% to -11.7%; P = .002). CONCLUSIONS AND RELEVANCE:While 1 in 5 fellow eyes had DLS on OCT among patients with unilateral new-onset exudative AMD in the first eye, the prevalence of neMNV on OCT-A was only 40% among these eyes with DLS. These results suggest DLS on OCT in fellow eyes of patients with new-onset exudative AMD in the first eye is not a good surrogate for neMNV. Trial Registration:ISRCTN registry Identifier: ISRTCTN13798759.

Macular disease is the leading cause of vision loss in the United Kingdom, affecting nearly 1.5 million people. Vascular endothelial growth factor (VEGF) has a major role in the onset and progression of vision-threatening macular conditions such as neovascular age-related macular degeneration, diabetic macular oedema and retinal vein occlusion. Anti-VEGF treatments, given by intravitreal injection, can prevent vision loss and may improve vision but must be started quickly before sight loss is irreversible. Informed by an observational study using Hospital Episode Statistics and the Emergency Care Data Set in England and a Costed Integrated Patient Scenario, we identify issues within the current pathway in England. Our key recommendations to optimise care and improve patient outcomes include: (1) increasing public awareness of macular disease to reduce late presentation; (2) ensuring the right patients are seen in the right place at the right time by improving quality and priority of optometrist referrals and by specialists directing patients with conditions such as diabetes and hypertension to community optometrists in the first instance; (3) planning for current and future demand and capacity using innovative solutions to manage expected increases in medical retina conditions and (4) ensuring patient-centred care. Our evidence should encourage National Health Service organisations at all levels to review macular disease pathways to address immediate issues and plan for the future. A shift in care delivery is needed, embracing scalable sustainable solutions, earlier diagnosis and treatment, improved access to more durable therapies, wider adoption of digital technologies and care closer to home.

BACKGROUND/AIMS:To characterise the distribution of retinal non-perfusion and its relationship with neovascularisation elsewhere (NVE) and visual acuity (VA) in referable diabetic retinopathy (DR) using ultra-widefield fluorescein angiography (UWFA). METHODS:Eyes with treatment-naïve moderately severe non-proliferative DR (NPDR) to proliferative diabetic retinopathy (PDR) (Diabetic Retinopathy Severity Scale (DRSS) level 43-60) with available UWFA were included in this study. Total and zonal non-perfusion and NVE area were manually delineated across concentric zones (central 1-10 mm, extended posterior >10-20 mm and mid-peripheral >20-30 mm diameter). The Non-Perfusion Index (NPI) was calculated as the ratio of non-perfused area to total area per zone. Associations between NPI, NVE, DR severity and best-recorded VA (BRVA) were assessed. RESULTS:A total of 133 eyes (60.9% NPDR, 39.1% PDR) were analysed. The highest NPI occurred in the mid-periphery (20-30 mm) and posterior pole (15-20 mm) zones, with nasal preponderance. NVE was most frequently observed in the posterior pole (66%) and nasal quadrants. Central and mid-peripheral NPI correlated with NVE area in corresponding zones, showing quadrant-specific associations. Receiver operating characteristic analysis identified a total NPI threshold of 0.23 to differentiate PDR from NPDR (area under the curve=0.798), with the lowest threshold in the superior quadrant. Neither NPI nor NVE area correlated with BRVA. CONCLUSION:NVE represents a localised response to adjacent ischaemia predominantly affecting the nasal and extended posterior retina. NPI threshold of 0.23 distinguishes proliferative from non-proliferative stages, with the superior quadrant showing greater susceptibility to NVE at lower ischaemic levels. No significant correlation found between NPI, NVE area and visual acuity.

Purpose:To investigate the feasibility of using color fundus photographs (CFPs) combined with artificial intelligence (AI) for detecting diabetic macular ischemia (DMI), a condition characterized by retinal capillary loss in the macula that leads to vision impairment in diabetic patients. Despite the widespread use of CFPs and AI in diagnosing other eye diseases like diabetic retinopathy (DR), their application in DMI detection remains unexplored because of skepticism regarding its viability. Methods:A graph neural network-based multispectral-view learning (GNN-MSVL) model was developed to detect DMI from CFPs. The model uses computational multispectral imaging to reconstruct 24-wavelength pseudo-multispectral fundus images from CFPs, enhancing sensitivity to subtle reflectance changes caused by ischemic tissue. ResNeXt101 served as the backbone for multi-view feature extraction, whereas a customized GNN with jumper connections improved cross-spectral relationship learning. The study included 1078 macula-centered CFPs from 1078 eyes of 592 diabetic patients, with 530 images from 300 patients confirmed as DMI cases. Results:The GNN-MSVL model achieved an accuracy of 84.7% and an area under the receiver operating characteristic curve of 0.900 (95% confidence interval, 0.852-0.937) at the eye level, significantly outperforming both baseline CFP-trained models and human experts (P < 0.01). Conclusions:AI-based analysis of CFPs shows promising potential for DMI detection, offering a feasible, early, and cost-effective screening method. This approach could address the current gap in DMI diagnosis and improve clinical outcomes for diabetic patients. Translational Relevance:This study will establish a feasible, CFP-based screening method for DMI.