Digital Fundus Photography Research Articles

Advancements in equine ophthalmic imaging enhance understanding of ocular and orbital anatomy and disease in standing sedated horses.

To review data on the advances in equine ophthalmic imaging that have been made during the past 5 years and highlight advantages of using multiple imaging modalities to improve clinical observational skills and improve diagnostic accuracy. A literature review from 2019 through 2024 of equine ophthalmic digital photography, fundus photography, ocular and orbital ultrasonography (US), ultrasound biomicroscopy (UBM), confocal microscopy (CM), spectral domain optical coherence tomography (SD-OCT), radiography, CT, and MRI. Digital photography remains the cornerstone of equine ophthalmic imaging for documenting examination findings, sharing information with colleagues, and consulting with specialists. Digital images also allow for in-depth postexamination review and evaluation, often revealing subtleties that may have otherwise gone undetected during the ophthalmic examination. Advanced imaging modalities are being used more frequently in equine ophthalmology, especially those that can be used with the horses standing under sedation, including US, UBM, SD-OCT, CM, and CT. Advances in equine ophthalmic imaging have led to many new clinical discoveries and to an increase in our knowledge of ocular anatomy and diseases in the horse. Many of these advanced diagnostic imaging modalities, such as MRI, CT, SD-OCT, and CM, are cost prohibitive and require substantial operator training to ensure proficiency. However, their availability in tertiary referral centers, such as veterinary teaching hospitals and large equine clinics/practices, is becoming more widespread. Advanced equine ophthalmic imaging data contributes substantially to our general understanding of clinical and applied anatomy and improves our understanding of the underlying pathogenesis associated with specific diseases.

Prevalence of Diabetic Retinopathy Among Diabetic Patients from Northeastern Bulgaria.

Diabetic retinopathy (DR) is a leading cause of visual impairment globally among working-aged individuals. This study aims to update data on DR prevalence in Bulgaria. The present cross-sectional study was conducted between 1 January 2022 and 1 January 2023, using a local diabetes registry from the city of Varna as a sampling framework. In total, 587 diabetic patients underwent DR examination. Data included demographics, diabetes type/duration, treatment, and ophthalmic history. DR status was assessed using indirect slit-lamp biomicroscopy or digital fundus photography, graded by the International Clinical Diabetic Retinopathy Scale. Of 587 participants, 13 were excluded due to cataract-related ungradable images. The median age was 65 years (IQR 56-73), with a slight female predominance (54%). The overall prevalence of any DR was 39.9% (95% CI 35.9-44.0), with non-proliferative DR (NPDR) at 27.5%, proliferative DR (PDR) at 7.3%, and macular edema (DME) at 5%. Type 1 diabetes patients had significantly higher DR prevalence (68.8%) than type 2 (34.1%, p < 0.001). Men exhibited higher DR prevalence. Age and diabetes duration correlated positively with DR prevalence. Insulin treatment was associated with higher DR prevalence (55.6%) than oral antidiabetic treatment (22.5%, p < 0.001) for type 2 diabetes patients. Among those diagnosed with DR, 70.9% received treatment, mainly laser therapy. These findings provide epidemiological insights for future research and emphasize the need for a comprehensive national DR screening program in Bulgaria. Technological advancements enable proactive measures to mitigate DR-related visual impairment and blindness, including widespread screening, even in rural areas.

Screening and monitoring of diabetic retinopathy in community care: the effectiveness of single-field vs multi-field fundus photography.

This study aimed to evaluate the effectiveness of single-field fundus photography for diabetic retinopathy (DR) screening and monitoring versus six-field imaging in community settings. Adults aged ≥30 years with type 2 diabetes from 15 communities in Northeast China were recruited for this prospective cohort study (n=2006 at baseline and n=1456 at follow-up). Participants underwent both single-field and six-field digital fundus photography at baseline and follow-up visits (mean duration of 21.2±3.2 months). Photographs were graded using international standards. Agreement in DR severity grading, referral recommendations, and detection of DR progression were compared between single-field and six-field fundus photography. Single-field grading showed substantial agreement with multi-field grading in classifying DR severity (81.9% identical at baseline, 80.6% at follow-up, Gwet's AC1 0.79 and 0.77). For referring eyes with moderate non-proliferative DR or worse, single-field grading had ∼70% sensitivity and 100% specificity compared to six-field grading. Single-field grading identified 74.9% or 79.7% of eyes progressing or regressing by six-field grading, respectively. Single-field fundus photography demonstrated reasonable effectiveness for DR screening and monitoring in a community setting, supporting its use for improving access to DR detection. However, reduced sensitivity compared to multi-field imaging should be acknowledged.

A Deep Learning Approach for Accurate Discrimination Between Optic Disc Drusen and Papilledema on Fundus Photographs.

Optic disc drusen (ODD) represent an important differential diagnosis of papilledema caused by intracranial hypertension, but their distinction may be difficult in clinical practice. The aim of this study was to train, validate, and test a dedicated deep learning system (DLS) for binary classification of ODD vs papilledema (including various subgroups within each category), on conventional mydriatic digital ocular fundus photographs collected in a large international multiethnic population. This retrospective study included 4,508 color fundus images in 2,180 patients from 30 neuro-ophthalmology centers (19 countries) participating in the Brain and Optic Nerve Study with Artificial Intelligence (BONSAI) Group. For training and internal validation, we used 857 ODD images and 3,230 papilledema images, in 1,959 patients. External testing was performed on an independent data set (221 patients), including 207 images with ODD (96 visible and 111 buried), provided by 3 centers of the Optic Disc Drusen Studies Consortium, and 214 images of papilledema (92 mild-to-moderate and 122 severe) from a previously validated study. The DLS could accurately distinguish between all ODD and papilledema (all severities included): area under the receiver operating characteristic curve (AUC) 0.97 (95% confidence interval [CI], 0.96-0.98), accuracy 90.5% (95% CI, 88.0%-92.9%), sensitivity 86.0% (95% CI, 82.1%-90.1%), and specificity 94.9% (95% CI, 92.3%-97.6%). The performance of the DLS remained high for discrimination of buried ODD from mild-to-moderate papilledema: AUC 0.93 (95% CI, 0.90-0.96), accuracy 84.2% (95% CI, 80.2%-88.6%), sensitivity 78.4% (95% CI, 72.2%-84.7%), and specificity 91.3% (95% CI, 87.0%-96.4%). A dedicated DLS can accurately distinguish between ODD and papilledema caused by intracranial hypertension, even when considering buried ODD vs mild-to-moderate papilledema.

High-normal serum bilirubin is a useful indicator to assess the risk of diabetic retinopathy in type 2 diabetes: A real-world study

BackgroundTo investigate the association of serum bilirubin within normal range, especially unconjugated bilirubin (UCB), with diabetic retinopathy (DR) in patients with type 2 diabetes mellitus (T2DM). MethodsIn this cross-sectional, real-world study, 7617 T2DM patients were stratified into quartiles based on serum UCB levels. DR was determined by digital fundus photography and further classified into non-proliferative diabetic retinopathy (NPDR) and PDR. The associations of serum bilirubin levels and UCB quartiles with DR were investigated by logistic regression analysis. ResultsAfter controlling for age, sex, and diabetes duration, the DR prevalence was significantly decreased across the serum UCB quartiles (40.4 %, 33.4 %, 29.7 %, 26.6 % for each quartile, respectively, p < 0.001 for trend). The subjects with DR had lower serum total bilirubin (TB) and UCB, rather than conjugated bilirubin (CB), compared with those without DR (p = 0.003 for TB, p < 0.001 for UCB, and p = 0.528 for CB, respectively), while all three types of serum bilirubin in the subjects with PDR were obviously lower than those with NPDR (p = 0.006 for TB, and p < 0.001 for UCB and CB, respectively). After adjustment for confounding factors, logistic regression demonstrated negative associations of serum TB and UCB levels, rather than CB, with the presence of DR (OR: 0.844, 95%CI: 0.774–0.920, p < 0.001 for TB; OR: 0.828, 95%CI: 0.763–0.899, p < 0.001 for UCB; and OR: 0.984, 95%CI: 0.900–1.074, p = 0.713 for CB, respectively). Additionally, a fully-adjusted analysis revealed a negative correlation between UCB quartiles and DR (p < 0.001). ConclusionHigh-normal serum TB and UCB were closely associated with the decreased odds of DR, while all types of serum bilirubin were negatively correlated with the severity of DR in T2DM patients. Serum bilirubin may be used as a potential indicator to assess the risk and severity of DR in T2DM.

Optic disc characteristics on digital fundus photographs in Saudi children.

To assess the optic disc parameters in healthy Saudi children. This study recruited 85 children who were medically free, born full-term, cooperative, and aged 3-17 years. The children underwent a thorough ophthalmological examination (visual acuity, refraction post-cycloplegia, fundus photography) at the ophthalmology clinic of King Abdulaziz University Hospital, Jeddah. Fundus photographs obtained by a fundus camera were evaluated by the Retinal Size Tool program. Forty-eight participants were male (56.5%). The mean birth weight was 2.97±0.8kg and the median gestational age was 39 weeks (range, 37-40 weeks). The median areas of the neuroretinal rim, cup, and optic disc were 1.82mm2 (range, 0.84-2.83mm2), 0.47mm2 (range, 0.18-1.25mm2), and 2.33mm2 (range, 1.15-3.52mm2), respectively. The older age group had smaller neuroretinal areas compared to the younger age groups. The variables demonstrated no apparent correlation to axial length, refraction, or birth parameters. The cup size increased together with the optic disc (r=0.659, p<0.001). Sex and refraction did not correlate with any of the studied factors. This study yielded normative data for the optic disc parameters of healthy Saudi children. The data can be used as a reference in the pediatric ophthalmology clinic to aid the identification of optic disc abnormalities.

Validation of using smartphone based non-mydriatic camera for retinal photography to diagnose diabetic retinopathy

Purpose: To assess the sensitivity and specificity of retinal imaging captured by portable non-mydriatic ‘fundus on phone’ camera as a screening tool against the conventional seven field digital fundus photography. Design: Prospective comparative study design. Methods: Study was performed over 300 diabetic patients (600 eyes). First fundus pictures were taken with non-mydriatic ‘fundus-on-phone’ camera and then with Zeiss seven field fundus camera after pupillary dilation. Images were analysed and compared by two observers. Results: Out of all the 600 undilated fundus images, 15.5% were ungradable, 76% were graded as poor and average quality by both observers. From the dilated fundus images taken using Zeiss camera, only 0.8% images were found ungradable by both observers. The sensitivity and specificity for detecting any diabetic retinopathy changes and its severity by using ‘fundus-on-phone’ camera against Zeiss fundus camera were found to be 54.92% and 85.5%, respectively, by observer one and 56.48% and 85.5%, respectively, by observer two. Conclusion: Non-mydriatic retinal imaging using fundus-on-phone camera is not a feasible method for screening diabetic retinopathy due to low sensitivity and high number of poor-quality images.

Prediction of cardiovascular risk factors from retinal fundus photographs: Validation of a deep learning algorithm in a prospective non-interventional study in Kenya.

Hypertension and diabetes mellitus (DM) are major causes of morbidity and mortality, with growing burdens in low-income countries where they are underdiagnosed and undertreated. Advances in machine learning may provide opportunities to enhance diagnostics in settings with limited medical infrastructure. A non-interventional study was conducted to develop and validate a machine learning algorithm to estimate cardiovascular clinical and laboratory parameters. At two sites in Kenya, digital retinal fundus photographs were collected alongside blood pressure (BP), laboratory measures and medical history. The performance of machine learning models, originally trained using data from the UK Biobank, were evaluated for their ability to estimate BP, glycated haemoglobin, estimated glomerular filtration rate and diagnoses from fundus images. In total, 301 participants were enrolled. Compared with the UK Biobank population used for algorithm development, participants from Kenya were younger and would probably report Black/African ethnicity, with a higher body mass index and prevalence of DM and hypertension. The mean absolute error was comparable or slightly greater for systolic BP, diastolic BP, glycated haemoglobin and estimated glomerular filtration rate. The model trained to identify DM had an area under the receiver operating curve of 0.762 (0.818 in the UK Biobank) and the hypertension model had an area under the receiver operating curve of 0.765 (0.738 in the UK Biobank). In a Kenyan population, machine learning models estimated cardiovascular parameters with comparable or slightly lower accuracy than in the population where they were trained, suggesting model recalibration may be appropriate. This study represents an incremental step toward leveraging machine learning to make early cardiovascular screening more accessible, particularly in resource-limited settings.

Age-related macular degeneration discordance in monozygotic twin pairs.

To examine age-related macular degeneration (AMD) and retinal pigment epithelium (RPE)-Bruch's membrane (BrM) complex volume associations in monozygotic twin pairs. In this study, 106 elderly twins (53 twin pairs) from the Finnish Twin Cohort study were recruited. Each participant underwent dilated 35-degree digital colour fundus photography (CFP), and spectral domain optical coherence tomography (OCT) and replied to a structured study questionnaire. The CFPs were graded according to the Age-Related Eye Disease Study (AREDS) classification. The OCT images were segmented and volumetric data of the RPE-BrM complex volume was calculated with the Orion™ software. The worse eye according to AREDS classification was used for the analysis. Twenty-nine (55%) of the twin pairs were discordant with regard to AREDS classification. Fourteen (26%) pairs were discordant with one twin participant having AMD (AREDS 2-4) and the other being unaffected (AREDS 1). Four (8%) pairs had one twin participant with intermediate or late AMD (AREDS 3-4) versus the other being unaffected (AREDS 1). The within-pair polychoric correlation for AREDS was 0.605 (95% confidence interval 0.418-0.792). In multivariate analysis intermediate and late AMD as well as age associated with RPE-BrM complex volume. RPE-BrM complex volume showed a within twin pair correlation, r = 0.430 (95% confidence interval 0.172-0.688, p < 0.01). A substantial proportion of monozygotic twin pairs are discordant with regard to age-related macular degeneration phenotype. RPE-BrM complex volume associated with age and intermediate and late AMD.

Rhegmatogenous retinal detachment in highly myopic eyes with implantable collamer lens.

To investigate the clinical characteristics, treatment methods and outcomes of rhegmatogenous retinal detachment (RRD) in highly myopic eyes with implantable collamer lens (ICL). High myopia patients who received treatment for nontraumatic RRD after ICL implantation surgery at the Retinal Department of Zhongshan Ophthalmic Center from Jan 2018 to Dec 2022 were reviewed. Comprehensive ophthalmologic examinations including visual acuity measurement and digital fundus photography were performed in each patient. A total of nine RRD eyes from nine patients who received V4c-ICL implantation were included. The mean time from ICL implantation surgery to the diagnosis of RRD was 32.44±22.56mo (range, 1-60mo). At the initial visit for RRD, giant retinal tear (GRT), horseshoe tear, simple round hole, and horseshoe tear combined with round hole were detected in 3, 3, 2, and 1 eye(s), respectively, with macula-off in eyes. Eight patients received surgical treatment, and one patient was treated by retinal laser photocoagulation alone. The ICL was preserved in 7 eyes. At the last follow-up, the mean best corrected visual acuity (BCVA) improved significantly from 1.76±1.06 logMAR at presentation to 0.81±1.01 logMAR (P=0.035), and no case of recurrent retinal detachment was found. The morphological presentation of retinal breaks is diverse in this study. The ICL can be preserved in most cases during the course of retinal detachment repair surgery in our data, companied with acceptable visual and anatomical outcomes.

Rapid detection of diabetic retinopathy in retinal images: a new approach using transfer learning and synthetic minority over-sampling technique

The challenge of early detection of diabetic retinopathy (DR), a leading cause of vision loss in working-age individuals in developed nations, was addressed in this study. Current manual analysis of digital color fundus photographs by clinicians, although thorough, suffers from slow result turnaround, delaying necessary treatment. To expedite detection and improve treatment timeliness, a novel automated detection system for DR was developed. This system utilized convolutional neural networks. Visual geometry group 16-layer network (VGG16), a pre-trained deep learning model, for feature extraction from retinal images and the synthetic minority over-sampling technique (SMOTE) to handle class imbalance in the dataset. The system was designed to classify images into five categories: normal, mild DR, moderate DR, severe DR, and proliferative DR (PDR). Assessment of the system using the Kaggle diabetic retinopathy dataset resulted in a promising 93.94% accuracy during the training phase and 88.19% during validation. These results highlight the system's potential to enhance DR diagnosis speed and efficiency, leading to improved patient outcomes. The study concluded that automation and artificial intelligence (AI) could play a significant role in timely and efficient disease detection and management.

Risk prediction of diabetic retinopathy based on visit-to-visit fasting blood glucose indices.

The long-term glucose monitoring is essential to the risk assessment of diabetic retinopathy (DR), the aim of this study was to investigate the predictive ability of visit-to-visit fasting blood glucose (FBG) indices on the risk of DR. This was a community-based, cohort study conducted from 2013 to 2021. DR was diagnosed by digital fundus photography. The FPG indices included FBG, var. Associations of each FBG indices and DR were estimated using multinomial logistic regression models adjusting for confounders, and discrimination was determined by area under the curve (AUC). Predictive utility of different models was compared by changes in AUC, integrated discrimination improvement (IDI), and net reclassification index (NRI). This study analyzed 5054 participants, the mean age was 46.26 ± 11.44 years, and 2620 (51.84%) were women. After adjustment for confounders, the adjusted odds ratios (ORs) with 95% confidence intervals (CIs) for FBG, SD, CV, VIM, ARV, M-FBG, and cumulative FBG load were 1.62 (1.52-1.73), 2.74 (2.38-3.16), 1.78 (1.62-1.95), 1.11 (0.95-1.29), 1.72 (1.56-1.91), 2.15 (1.96-2.36), and 2.57 (2.31-2.85), respectively. The AUC of the model with separate cumulative FBG load and classical risk factors was 0.9135 (95%CI 0.8890-0.9380), and no substantive improvement in discrimination was achieved with the addition of other FBG indices once cumulative FBG load was in the model. Cumulative FBG load is adequate for capturing the glucose-related DR risk, and the predictive utility of cumulative FBG load is not significantly improved by adding or replacing other FBG indices in the assessment of DR risk.

Risk Factors Associated with Diabetic Retinopathy with and without Macular Edema in Recently Diagnosed Patients with Type 2 Diabetes.

To evaluate the risk factors associated with diabetic macular edema (DME) in patients with a recent type 2 diabetes mellitus (T2DM) diagnosis. We conducted a case-control study at a third-level hospital in Mexico City. We enrolled patients ≥18 years old, with T2DM less than five years of diagnosis, without disabling complications, and non-smokers. The control group was patients with diabetic retinopathy and without macular edema (DR-DME). Cases were patients with DR+DME. We measured fasting glucose, creatinine, lipid profile, urinary albumin/creatinine ratio (ACR), and HbA1c. An ophthalmological examination consisted of visual acuity measurement, digital three-field fundus photography with an automatic non-mydriatic camera, slit lamp, and Optical coherence tomography (OCT) examination. 183 and 61 patients with DR-DME and DR+DME, respectively, were included in the analysis. The prevalence of mild DR was higher in the DR-DME group, but the frequencies of moderate and severe retinopathy were higher in the DR+DME group. Patients in the DR-DME group had better vision than those in the DR+DME group. Logistic regression analysis revealed that age (OR, 1.07), HbA1c (OR, 1.19), and Albumin-to-Creatinine Ratio (ACR) > 30 mg/g (OR, 3.37) were associated with an increased possibility of DME compared to DR-DME. Our study provides insights into the association between risk factors and DME. We found a statistically strong association between HbA1c levels, age, and ACR. Patients with poor metabolic control should undergo an extensive medical examination to screen for DME, which may be related to the chronicity of DM and renal damage.

Retinal pigment epithelium-Bruch's membrane volume in grading of age-related macular degeneration.

To assess the agreement of optical coherence tomography (OCT) algorithm-based retinal pigment epithelium -Bruch's membrane complex volume (RBV) with fundus photograph-based age-related macular degeneration (AMD) grading. Digital color fundus photographs (CFPs) and spectral domain OCT images were acquired from 96 elderly subjects. CFPs were graded according to Age-Related Eye Disease Study (AREDS) classification. OCT image segmentation and RBV data calculation were done with Orion™ software. Univariate and multivariate analyses were performed to find out whether AMD lesion features associated with higher RBVs. RBV correlated with AMD grading (rs=0.338, P=0.001), the correlation was slightly stronger in early AMD (n=52; rs=0.432, P=0.001). RBV was higher in subjects with early AMD compared with those with no AMD lesions evident in fundus photographs (1.05±0.20 vs 0.96±0.13 mm3, P=0.023). In multivariate analysis higher RBVs were associated significantly with higher total drusen (β=0.388, P=0.027) and pigmentation areas (β=0.319, P=0.020) in fundus photographs, whereas depigmentation area (β=-0.295, P=0.015) associated with lower RBV. RBV correlate with AMD grading status, with a stronger association in patients with moderate, non-late AMD grades. This effect is driven mostly by lesions with drusen or pigmentation. Lesions with depigmentation tend to have lower values. RBV is more comprehensive measurement of the key area of AMD pathogenesis, compared to sole drusen volume analysis. RBV measurements are independent on grader variations and offer a possibility to quantify early and middle grade AMD lesions in a research setting, but may not substitute fundus photograph-based grading in the whole range of AMD spectrum.

Diabetic Retinopathy Diagnosis Categorization Using Deep Learning

Diabetic Retinopathy (DR) is an eye disease associated with chronic diabetes. DR is the leading cause of blindness among working aged adults around the world and estimated it may affect more than 93 million people. Progression to vision impairment can be slowed or controlled if DR is detected in time, however this can be difficult as the disease often shows few symptoms until it is too late to provide effective treatment. Currently, detecting DR is a time-consuming and manual process, which requires an ophthalmologist or trained clinician to examine and evaluate digital color fundus photographs of the retina, to identify DR by the presence of lesions associated with the vascular abnormalities caused by the disease. The automated method of DR screening will speed up the detection and decision-making process, which will help to control or manage DR progression. This paper presents an automated classification system, in which it analyzes fundus images with varying illumination and fields of view and generates a severity grade for diabetic retinopathy (DR) using machine learning models such as CNN, VGG-16 and VGG-19.This system achieves 80% sensitivity, 82% accuracy, 82% specificity, and 0.904 AUC for classifying images into 5 categories ranging from 0 to 4, where 0 is no DR and 4 is proliferative DR.

Vitreomacular interface abnormalities in the Ghanaian African

Background/ObjectiveDescribe vitreomacular interface abnormalities (VMIA) using spectral-domain optical coherence tomography (SD-OCT), and correlations with age-related macular degeneration (AMD) grade in Ghanaian Africans.Subjects/MethodsProspective, cross-sectional study of adults aged ≥50 years recruited in Ghana AMD Study. Participant demographics, medical histories, ophthalmic examination, digital colour fundus photography (CFP) were obtained. High-resolution five-line raster OCT, Macular Cube 512 × 128 scans, and additional line scans in areas of clinical abnormality, were acquired. SD-OCT VMI features classified by International Vitreomacular Traction Study Group system and relationships to AMD grade were evaluated. Outcomes: VMIA prevalence, posterior vitreous detachment (PVD), vitreomacular adhesions (VMA), vitreomacular traction (VMT), epiretinal membranes (ERM), correlations with AMD grade.ResultsThe full Ghana AMD cohort included 718 participants; 624 participants (1248 eyes) aged ≥50 years (range = 50–101, mean = 68.8), 68.9% female were included in this analysis. CFP with OCT scans were available for 776 eyes (397 participants); 707 (91.1%) had gradable CFP and OCT scans for both AMD and VMI grading forming the dataset for this report. PVD was absent in 504 (71.3%); partial and complete PVD occurred in 16.7% and 12.0% respectively. PVD did not increase with age (p = 0.720). VMIA without traction and macular holes were observed in 12.2% of eyes; 87.8% had no abnormalities. VMIA was not significantly correlated with AMD grade (p = 0.819).ConclusionsThis provides the first assessment of VMIA in Ghanaian Africans. VMIA are common in Africans; PVD may be less common than in Caucasians. There was no significant association of AMD grade with VMIA.

Retinal vessel geometry in patients with idiopathic epiretinal membrane

We investigated the associations between retinal vascular geometric measurements and idiopathic epiretinal membrane (ERM). Whether changes in retinal vascular geometry are independent of systemic cardiovascular risk factors was also evaluated. This retrospective, cross sectional study included 98 patients with idiopathic ERM, and 99 healthy age-matched controls. Quantitative retinal vascular parameters were measured from digital retinal fundus photographs using a semi-automated computer-assisted program. Multivariate logistic regression analyses were performed to evaluate associations between retinal vascular geometric parameters and the presence of idiopathic ERM after adjusting for systemic cardiovascular risk factors. There was no significant difference in the baseline characteristics of the two groups, except that the ERM group had a higher proportion of females than the control group. In multivariate regression analyses, female sex (odds ratio [OR] 0.402; 95% CI 0.196–0.802; P = 0.011), wider retinal venular caliber (OR 16.852; 95% CI 5.384–58.997; P < 0.001) and decreased total fractal dimension (OR 0.156; 95% CI 0.052–0.440; P = 0.001) were associated with idiopathic ERM. Idiopathic ERM was associated with alterations in global retinal microvascular geometric parameters, wider retinal venules, and less complex vascular branching patterns, independent of cardiovascular risk factors.

Differentiating Glaucomatous Optic Neuropathy From Non-glaucomatous Optic Neuropathies Using Deep Learning Algorithms

A deep learning framework to differentiate glaucomatous optic disc changes due to glaucomatous optic neuropathy (GON) from non-glaucomatous optic disc changes due to non-glaucomatous optic neuropathies (NGONs). Cross-sectional study. A deep-learning system was trained, validated, and externally tested to classify optic discs as normal, GON, or NGON, using 2183 digital color fundus photographs. A Single-Center data set of 1822 images (660 images of NGON, 676 images of GON, and 486 images of normal optic discs) was used for training and validation, whereas 361 photographs from 4 different data sets were used for external testing. Our algorithm removed the redundant information from the images using an optic disc segmentation (OD-SEG) network, after which we performed transfer learning with various pre-trained networks. Finally, we calculated sensitivity, specificity, F1-score, and precision to show the performance of the discrimination network in the validation and independent external data set. For classification, the algorithm with the best performance for the Single-Center data set was DenseNet121, with a sensitivity of 95.36%, precision of 95.35%, specificity of 92.19%, and F1 score of 95.40%. For the external validation data, the sensitivity and specificity of our network for differentiating GON from NGON were 85.53% and 89.02%, respectively. The glaucoma specialist who diagnosed those cases in masked fashion had a sensitivity of 71.05% and a specificity of 82.21%. The proposed algorithm for the differentiation of GON from NGON yields results that have a higher sensitivity than those of a glaucoma specialist, and its application for unseen data thus is extremely promising.

Retinal blood vessel diameters in children and adults exposed to a simulated altitude of 3,000m.

Introduction: Technological advances have made high-altitude ski slopes easily accessible to skiers of all ages. However, research on the effects of hypoxia experienced during excursions to such altitudes on physiological systems, including the ocular system, in children is scarce. Retinal vessels are embryologically of the same origin as vessels in the brain, and have similar anatomical and physiological characteristics. Thus, any hypoxia-related changes in the morphology of the former may reflect the status of the latter. Objective: To compare the effect of one-day hypoxic exposure, equivalent to the elevation of high-altitude ski resorts in North America and Europe (∼3,000m), on retinal vessel diameter between adults and children. Methods: 11 adults (age: 40.1 ± 4.1years) and 8 children (age: 9.3 ± 1.3years) took part in the study. They spent 3days at the Olympic Sports Centre Planica (Slovenia; altitude: 940m). During days 1 and 2 they were exposed to normoxia (FiO2 = 0.209), and day 3 to normobaric hypoxia (FiO2 = 0.162 ± 0.03). Digital high-resolution retinal fundus photographs were obtained in normoxia (Day 2) and hypoxia (Day 3). Central retinal arteriolar equivalent (CRAE) and venular equivalents (CRVE) were determined using an Automated Retinal Image Analyser. Results: Central retinal arteriolar and venular equivalents increased with hypoxia in children (central retinal arteriolar equivalent: 105.32 ± 7.72µm, hypoxia: 110.13 ± 7.16µm, central retinal venular equivalent: normoxia: 123.39 ± 8.34µm, hypoxia: 130.11 ± 8.54µm) and adults (central retinal arteriolar equivalent: normoxia: 105.35 ± 10.67µm, hypoxia: 110.77 ± 8.36µm; central retinal venular equivalent: normoxia: 126.89 ± 7.24µm, hypoxia: 132.03 ± 9.72µm), with no main effect of group or group*condition interaction. A main effect of condition on central retinal arteriolar and venular equivalents was observed (central retinal arteriolar equivalent:normoxia: 105.34 ± 9.30µm, hypoxia: 110.50 ± 7.67µm, p < 0.001; central retinal venular equivalent: normoxia: 125.41 ± 7.70µm, hypoxia: 131.22 ± 9.05µm, p < 0.001). Conclusion: A 20-hour hypoxic exposure significantly increased central retinal arteriolar and venular equivalents in adults and children. These hypoxia-induced increases were not significantly different between the age groups, confirming that vasomotor sensitivity of the retinal vessels to acute hypoxia is comparable between adults and prepubertal children.

An efficient approach to estimate the risk of coronary artery disease for people living with HIV using machine-learning-based retinal image analysis.

People living with HIV (PLWH) have increased risks of non-communicable diseases, especially cardiovascular diseases. Current HIV clinical management guidelines recommend regular cardiovascular risk screening, but the risk equation models are not specific for PLWH. Better tools are needed to assess cardiovascular risk among PLWH accurately. We performed a prospective study to determine the performance of automatic retinal image analysis in assessing coronary artery disease (CAD) in PLWH. We enrolled PLWH with ≥1 cardiovascular risk factor. All participants had computerized tomography (CT) coronary angiogram and digital fundus photographs. The primary outcome was coronary atherosclerosis; secondary outcomes included obstructive CAD. In addition, we compared the performances of three models (traditional cardiovascular risk factors alone; retinal characteristics alone; and both traditional and retinal characteristics) by comparing the area under the curve (AUC) of receiver operating characteristic curves. Among the 115 participants included in the analyses, with a mean age of 54 years, 89% were male, 95% had undetectable HIV RNA, 45% had hypertension, 40% had diabetes, 45% had dyslipidemia, and 55% had obesity, 71 (61.7%) had coronary atherosclerosis, and 23 (20.0%) had obstructive CAD. The machine-learning models, including retinal characteristics with and without traditional cardiovascular risk factors, had AUC of 0.987 and 0.979, respectively and had significantly better performance than the model including traditional cardiovascular risk factors alone (AUC 0.746) in assessing coronary artery disease atherosclerosis. The sensitivity and specificity for risk of coronary atherosclerosis in the combined model were 93.0% and 93.2%, respectively. For the assessment of obstructive CAD, models using retinal characteristics alone (AUC 0.986) or in combination with traditional risk factors (AUC 0.991) performed significantly better than traditional risk factors alone (AUC 0.777). The sensitivity and specificity for risk of obstructive CAD in the combined model were 95.7% and 97.8%, respectively. In this cohort of Asian PLWH at risk of cardiovascular diseases, retinal characteristics, either alone or combined with traditional risk factors, had superior performance in assessing coronary atherosclerosis and obstructive CAD. People living with HIV in an Asian cohort with risk factors for cardiovascular disease had a high prevalence of coronary artery disease (CAD). A machine-learning-based retinal image analysis could increase the accuracy in assessing the risk of coronary atherosclerosis and obstructive CAD.