Comparison of standardized clinical classification with fundus photograph grading for the assessment of diabetic retinopathy and diabetic macular edema severity

Optimized deep CNN for detection and classification of diabetic retinopathy and diabetic macular edema

Diabetic Retinopathy (DR) and Diabetic Macular Edema (DME) are vision related complications prominently found in diabetic patients. The early identification of DR/DME grades facilitates the devising of an appropriate treatment plan, which ultimately prevents the probability of visual impairment in more than 90% of diabetic patients. Thereby, an automatic DR/DME grade detection approach is proposed in this work by utilizing image processing. In this work, the retinal fundus image provided as input is pre-processed using Discrete Wavelet Transform (DWT) with the aim of enhancing its visual quality. The precise detection of DR/DME is supported further with the application of suitable Artificial Neural Network (ANN) based segmentation technique. The segmented images are subsequently subjected to feature extraction using Adaptive Gabor Filter (AGF) and the feature selection using Random Forest (RF) technique. The former has excellent retinal vein recognition capability, while the latter has exceptional generalization capability. The RF approach also assists with the improvement of classification accuracy of Deep Convolutional Neural Network (CNN) classifier. Moreover, Chicken Swarm Algorithm (CSA) is used for further enhancing the classifier performance by optimizing the weights of both convolution and fully connected layer. The entire approach is validated for its accuracy in determination of grades of DR/DME using MATLAB software. The proposed DR/DME grade detection approach displays an excellent accuracy of 97.91%.

The additive effect of intravitreal dexamethasone combined with bevacizumab in refractory diabetic macular edema

PURPOSE

To evaluate the short-term structural and visual outcomes and side effects associated with intravitreal dexamethasone (IVD) combined with bevacizumab (IVB) in treating patients with diabetic macular edema (DME) and an inadequate response to anti-vascular endothelial growth factor (anti-VEGF) agents.

METHODS

In this prospective interventional case series, a total of 81 eyes of 81 patients with type 2 diabetes mellitus (T2DM) and refractory DME were included and assigned to one of two groups: I) those receiving three monthly intravitreal injections of combined bevacizumab and dexamethasone (IVB+IVD) and II) those receiving three monthly intravitreal injections of bevacizumab alone (IVB). The primary outcome was the inter-group difference in central macular thickness (CMT); secondary outcomes included best-corrected visual acuity (BCVA), baseline optical coherence tomography (OCT) biomarkers, and intraocular pressure (IOP) one month after the last injection.

RESULTS

Reduction in CMT and improvement in BCVA were significantly greater in the IVB+IVD group than the IVB group (109.88±156.25 vs. 43±113.67, respectively, P=0.03; and -0.13±0.23 vs. -0.01±0.17, respectively, P=0.008). Presence of neurosensory retinal detachment (NSD) (P<0.001) and complete inner segment/outer segment junction (IS-OS) disruption (P=0.049) on baseline OCT scans were associated with further CMT reductions in response to IVD. Conversely, identifiable epiretinal membrane (ERM) (P=0.002) and multiple hyperreflective foci (>20) (P=0.049) were associated with smaller reductions in CMT. Vitreomacular traction correlated with worse visual outcomes in the IVB+IVD group (P=0.003). The intergroup IOP difference was not clinically significant.

CONCLUSION

In patients with refractory DME, addition of IVD to the standard IVB regimen can improve visual and structural outcomes without increasing the risk of endophthalmitis, IOP rise, or intraocular inflammation. Patients with NSD are more likely to respond well to IVD. The presence of ERM may predict poor treatment response.

Diagnostic Accuracy of Automated Diabetic Retinopathy Image Assessment Softwares: IDx-DR and Medios Artificial Intelligence

INTRODUCTION

Numerous studies have demonstrated the use of artificial intelligence (AI) for early detection of referable diabetic retinopathy (RDR). A direct comparison of these multiple automated diabetic retinopathy (DR) image assessment softwares (ARIAs) is, however, challenging. We retrospectively compared the performance of two modern ARIAs, IDx-DR and Medios AI.

METHODS

In this retrospective-comparative study, retinal images with sufficient image quality were run on both ARIAs. They were captured in 811 consecutive patients with diabetes visiting diabetic clinics in Poland. For each patient, four non-mydriatic images, 45° field of view, i.e., two sets of one optic disc and one macula-centered image using Topcon NW400 were captured. Images were manually graded for severity of DR as no DR, any DR (mild non-proliferative diabetic retinopathy [NPDR] or more severe disease), RDR (moderate NPDR or more severe disease and/or clinically significant diabetic macular edema [CSDME]), or sight-threatening DR (severe NPDR or more severe disease and/or CSDME) by certified graders. The ARIA output was compared to manual consensus image grading (reference standard).

RESULTS

On 807 patients, based on consensus grading, there was no evidence of DR in 543 patients (67%). Any DR was seen in 264 (33%) patients, of which 174 (22%) were RDR and 41 (5%) were sight-threatening DR. The sensitivity of detecting RDR against reference standard grading was 95% (95% CI: 91, 98%) and the specificity was 80% (95% CI: 77, 83%) for Medios AI. They were 99% (95% CI: 96, 100%) and 68% (95% CI: 64, 72%) for IDx-DR, respectively.

CONCLUSION

Both the ARIAs achieved satisfactory accuracy, with few false negatives. Although false-positive results generate additional costs and workload, missed cases raise the most concern whenever automated screening is debated.

Comparisons of handheld retinal imaging devices with ultrawide field images for determining diabetic retinopathy severity

PURPOSE

To compare diabetic retinopathy (DR) severity identified on handheld retinal imaging with ultrawide field (UWF) images.

METHODS

Mydriatic images of 225 eyes of 118 diabetic patients were prospectively imaged with the Aurora (AU) handheld retinal camera [5-field protocol (macula-centred, disc-centred, temporal, superior, inferior)] and compared with UWF images. Images were classified based on the international classification for DR. Sensitivity, specificity, kappa statistics (K/Kw) were calculated on an eye and person-level.

RESULTS

Distribution of DR severity by AU/UWF images (%) by eye was no DR 41.3/36.0, mild non-proliferative DR (NPDR) 18.7/17.8, moderate 10.2/10.7, severe 16.4/15.1, proliferative DR (PDR) 13.3/20.4. Agreement between UWF and AU was exact in 64.4%, within 1-step 90.7%, k = 0.55 (95% CI:0.45-0.65), and kw = 0.79 (95% CI:0.73-0.85) by eye, and exact in 68%, within 1-step 92.9%, k = 0.58 (95% CI:0.50-0.66), and kw = 0.76 (95% CI:0.70-0.81) by person. Sensitivity/specificity for any DR, refDR, vtDR and PDR were as follows: 0.90/0.83, 0.90/0.97, 0.82/0.95 and 0.69/1.00 by person and 0.86/0.90, 0.84/0.98, 0.75/0.95 and 0.63/0.99 by eye. Handheld imaging missed 37% (17/46) eyes and 30.8% (8/26) persons with PDR. Only 3.9% (1/26) persons or 6.5% (3/46) eyes with PDR were missed if a referral threshold of moderate NPDR was used.

CONCLUSIONS

Data from this study show that comparing UWF and handheld images, when PDR was the referral threshold for handheld devices, 37.0% of eyes or 30.8% of patients with PDR were missed. Due to the identification of neovascular lesions outside of the handheld fields, lower referral thresholds are needed if handheld devices are used.

Validation of diagnostic accuracy of retinal image grading by trained non-ophthalmologist grader for detecting diabetic retinopathy and diabetic macular edema

PURPOSE

To validate the fundus image grading results by a trained grader (Non-ophthalmologist) and an ophthalmologist grader for detecting diabetic retinopathy (DR) and diabetic macular oedema (DMO) against fundus examination by a retina specialist (gold standard).

METHODS

A prospective diagnostic accuracy study was conducted using 2002 non-mydriatic colour fundus images from 1001 patients aged ≥40 years. Using the Aravind Diabetic Retinopathy Evaluation Software (ADRES) images were graded by both a trained non-ophthalmologist grader (grader-1) and an ophthalmologist (grader-2). Sensitivity, specificity, positive predictive value and negative predictive value were calculated for grader-1 and grader-2 against the grading results by an independent retina specialist who performed dilated fundus examination for every study participant.

RESULTS

Out of 1001 patients included, 42% were women and the mean ± (SD) age was 55.8 (8.39) years. For moderate or worse DR, the sensitivity and specificity for grading by grader-1 with respect to the gold standard was 66.9% and 91.0% respectively and the same for the ophthalmologist was 83.6% and 80.3% respectively. For referable DMO, grader-1 and grader-2 had a sensitivity of 74.6% and 85.6% respectively and a specificity of 83.7% and 79.8% respectively.

CONCLUSIONS

Our results demonstrate good level of accuracy for the fundus image grading performed by a trained non-ophthalmologist which was comparable with the grading by an ophthalmologist. Engaging trained non-ophthalmologists potentially can enhance the efficiency of DR diagnosis using fundus images. Further study with multiple non-ophthalmologist graders is needed to verify the results and strategies to improve agreement for DMO diagnosis are needed.

Blood pressure control for diabetic retinopathy

BACKGROUND

Diabetic retinopathy is a common complication of diabetes and a leading cause of visual impairment and blindness. Research has established the importance of blood glucose control to prevent development and progression of the ocular complications of diabetes. Concurrent blood pressure control has been advocated for this purpose, but individual studies have reported varying conclusions regarding the effects of this intervention.

OBJECTIVES

To summarize the existing evidence regarding the effect of interventions to control blood pressure levels among diabetics on incidence and progression of diabetic retinopathy, preservation of visual acuity, adverse events, quality of life, and costs.

SEARCH METHODS

We searched several electronic databases, including CENTRAL, and trial registries. We last searched the electronic databases on 3 September 2021. We also reviewed the reference lists of review articles and trial reports selected for inclusion.

SELECTION CRITERIA

We included randomized controlled trials (RCTs) in which either type 1 or type 2 diabetic participants, with or without hypertension, were assigned randomly to more intense versus less intense blood pressure control; to blood pressure control versus usual care or no intervention on blood pressure (placebo); or to one class of antihypertensive medication versus another or placebo.

DATA COLLECTION AND ANALYSIS

Pairs of review authors independently reviewed the titles and abstracts of records identified by the electronic and manual searches and the full-text reports of any records identified as potentially relevant. The included trials were independently assessed for risk of bias with respect to outcomes reported in this review.

MAIN RESULTS

We included 29 RCTs conducted in North America, Europe, Australia, Asia, Africa, and the Middle East that had enrolled a total of 4620 type 1 and 22,565 type 2 diabetic participants (sample sizes from 16 to 4477 participants). In all 7 RCTs for normotensive type 1 diabetic participants, 8 of 12 RCTs with normotensive type 2 diabetic participants, and 5 of 10 RCTs with hypertensive type 2 diabetic participants, one group was assigned to one or more antihypertensive agents and the control group to placebo. In the remaining 4 RCTs for normotensive participants with type 2 diabetes and 5 RCTs for hypertensive type 2 diabetic participants, methods of intense blood pressure control were compared to usual care. Eight trials were sponsored entirely and 10 trials partially by pharmaceutical companies; nine studies received support from other sources; and two studies did not report funding source. Study designs, populations, interventions, lengths of follow-up (range less than one year to nine years), and blood pressure targets varied among the included trials. For primary review outcomes after five years of treatment and follow-up, one of the seven trials for type 1 diabetics reported incidence of retinopathy and one trial reported progression of retinopathy; one trial reported a combined outcome of incidence and progression (as defined by study authors). Among normotensive type 2 diabetics, four of 12 trials reported incidence of diabetic retinopathy and two trials reported progression of retinopathy; two trials reported combined incidence and progression. Among hypertensive type 2 diabetics, six of the 10 trials reported incidence of diabetic retinopathy and two trials reported progression of retinopathy; five of the 10 trials reported combined incidence and progression. The evidence supports an overall benefit of more intensive blood pressure intervention for five-year incidence of diabetic retinopathy (11 studies; 4940 participants; risk ratio (RR) 0.82, 95% confidence interval (CI) 0.73 to 0.92; I2 = 15%; moderate certainty evidence) and the combined outcome of incidence and progression (8 studies; 6212 participants; RR 0.78, 95% CI 0.68 to 0.89; I2 = 42%; low certainty evidence). The available evidence did not support a benefit regarding five-year progression of diabetic retinopathy (5 studies; 5144 participants; RR 0.94, 95% CI 0.78 to 1.12; I2 = 57%; moderate certainty evidence), incidence of proliferative diabetic retinopathy, clinically significant macular edema, or vitreous hemorrhage (9 studies; 8237 participants; RR 0.92, 95% CI 0.82 to 1.04; I2 = 31%; low certainty evidence), or loss of 3 or more lines on a visual acuity chart with a logMAR scale (2 studies; 2326 participants; RR 1.15, 95% CI 0.63 to 2.08; I2 = 90%; very low certainty evidence). Hypertensive type 2 diabetic participants realized more benefit from intense blood pressure control for three of the four outcomes concerning incidence and progression of diabetic retinopathy. The adverse event reported most often (13 of 29 trials) was death, yielding an estimated RR 0.87 (95% CI 0.76 to 1.00; 13 studies; 13,979 participants; I2 = 0%; moderate certainty evidence). Hypotension was reported in two trials, with an RR of 2.04 (95% CI 1.63 to 2.55; 2 studies; 3323 participants; I2 = 37%; low certainty evidence), indicating an excess of hypotensive events among participants assigned to more intervention on blood pressure.

AUTHORS' CONCLUSIONS

Hypertension is a well-known risk factor for several chronic conditions for which lowering blood pressure has proven to be beneficial. The available evidence supports a modest beneficial effect of intervention to reduce blood pressure with respect to preventing diabetic retinopathy for up to five years, particularly for hypertensive type 2 diabetics. However, there was a paucity of evidence to support such intervention to slow progression of diabetic retinopathy or to affect other outcomes considered in this review among normotensive diabetics. This weakens any conclusion regarding an overall benefit of intervening on blood pressure in diabetic patients without hypertension for the sole purpose of preventing diabetic retinopathy or avoiding the need for treatment for advanced stages of diabetic retinopathy.

Detecting and staging diabetic retinopathy in retinal images using multi-branch CNN

Purpose

This paper aims to propose a solution for detecting and grading diabetic retinopathy (DR) in retinal images using a convolutional neural network (CNN)-based approach. It could classify input retinal images into a normal class or an abnormal class, which would be further split into four stages of abnormalities automatically.

Design/methodology/approach

The proposed solution is developed based on a newly proposed CNN architecture, namely, DeepRoot. It consists of one main branch, which is connected by two side branches. The main branch is responsible for the primary feature extractor of both high-level and low-level features of retinal images. Then, the side branches further extract more complex and detailed features from the features outputted from the main branch. They are designed to capture details of small traces of DR in retinal images, using modified zoom-in/zoom-out and attention layers.

Findings

The proposed method is trained, validated and tested on the Kaggle dataset. The regularization of the trained model is evaluated using unseen data samples, which were self-collected from a real scenario from a hospital. It achieves a promising performance with a sensitivity of 98.18% under the two classes scenario.

Originality/value

The new CNN-based architecture (i.e. DeepRoot) is introduced with the concept of a multi-branch network. It could assist in solving a problem of an unbalanced dataset, especially when there are common characteristics across different classes (i.e. four stages of DR). Different classes could be outputted at different depths of the network.

Automatic classification of brain magnetic resonance images with hypercolumn deep features and machine learning

Brain tumours are life-threatening and their early detection is very important in a patient's life. At the present time, magnetic resonance imaging is one of the methods used for detecting brain tumours. Expert decision support systems serve specialist physicians to make more accurate diagnoses by minimizing the errors arising from their subjective opinions in real clinical settings. The model proposed in this study detects important keypoints and then extracts hypercolumn deep features of these keypoints from some convolutional layers of VGG16. Finally, Random Forest and Logistic Regression classifiers are fed with a set of these features. Random Forest classifier offered the best performance with 94.51% accuracy, 91.61% sensitivity, 8.39% false-negative rate, 97.42% specificity, and 97.29% precision using fivefold cross-validation in this study. Consequently, it is thought that the proposed model could contribute to field experts by integrating it into computer-aided brain magnetic resonance imaging diagnosis systems.

Variability of Grading DR Screening Images among Non-Trained Retina Specialists

Poland has never had a widespread diabetic retinopathy (DR) screening program and subsequently has no purpose-trained graders and no established grader training scheme. Herein, we compare the performance and variability of three retinal specialists with no additional DR grading training in assessing images from 335 real-life screening encounters and contrast their performance against IDx-DR, a US Food and Drug Administration (FDA) approved DR screening suite. A total of 1501 fundus images from 670 eyes were assessed by each grader with a final grade on a per-eye level. Unanimous agreement between all graders was achieved for 385 eyes, and 110 patients, out of which 98% had a final grade of no DR. Thirty-six patients had final grades higher than mild DR, out of which only two had no grader disagreements regarding severity. A total of 28 eyes underwent adjudication due to complete grader disagreement. Four patients had discordant grades ranging from no DR to severe DR between the human graders and IDx-DR. Retina specialists achieved kappa scores of 0.52, 0.78, and 0.61. Retina specialists had relatively high grader variability and only a modest concordance with IDx-DR results. Focused training and verification are recommended for any potential DR graders before assessing DR screening images.

Quercetin: an effective polyphenol in alleviating diabetes and diabetic complications

Various studies, especially in recent years, have shown that quercetin has beneficial therapeutic effects in various human diseases, including diabetes. Quercetin has significant anti-diabetic effects and may be helpful in lowering blood sugar and increasing insulin sensitivity. Quercetin appears to affect many factors and signaling pathways involved in insulin resistance and the pathogenesis of type 2 of diabetes. TNFα, NFKB, AMPK, AKT, and NRF2 are among the factors that are affected by quercetin. In addition, quercetin can be effective in preventing and ameliorating the diabetic complications, including diabetic nephropathy, cardiovascular complications, neuropathy, delayed wound healing, and retinopathy, and affects the key mechanisms involved in the pathogenesis of these complications. These positive effects of quercetin may be related to its anti-inflammatory and anti-oxidant properties. In this article, after a brief review of the pathogenesis of insulin resistance and type 2 diabetes, we will review the latest findings on the anti-diabetic effects of quercetin with a molecular perspective. Then we will review the effects of quercetin on the key mechanisms of pathogenesis of diabetes complications including nephropathy, cardiovascular complications, neuropathy, delayed wound healing, and retinopathy. Finally, clinical trials investigating the effect of quercetin on diabetes and diabetes complications will be reviewed.

Automatic Detection of Covid-19 with Bidirectional LSTM Network Using Deep Features Extracted from Chest X-ray Images

Coronavirus disease, which comes up in China at the end of 2019 and showed different symptoms in people infected, affected millions of people. Computer-aided expert systems are needed due to the inadequacy of the reverse transcription-polymerase chain reaction kit, which is widely used in the diagnosis of this disease. Undoubtedly, expert systems that provide effective solutions to many problems will be very useful in the detection of Covid-19 disease, especially when unskilled personnel and financial deficiencies in underdeveloped countries are taken into consideration. In the literature, there are numerous machine learning approaches built with different classifiers in the detection of this disease. This paper proposes an approach based on deep learning which detects Covid-19 and no-finding cases using chest X-ray images. Here, the classification performance of the Bi-LSTM network on the deep features was compared with the Deep Neural Network within the frame of the fivefold cross-validation technique. Accuracy, sensitivity, specificity and precision metrics were used to evaluate the classification performance of the trained models. Bi-LSTM network presented better performance compare to DNN with 97.6% value of high accuracy despite the few numbers of Covid-19 images in the dataset. In addition, it is understood that concatenated deep features more meaningful than deep features obtained with pre-trained networks by one by, as well. Consequently, it is thought that the proposed study based on the Bi-LSTM network and concatenated deep features will be noteworthy in the design of highly sensitive automated Covid-19 monitoring systems.

LncRNA NEAT1 regulated diabetic retinal epithelial-mesenchymal transition through regulating miR-204/SOX4 axis

AIM

Epithelial-mesenchymal transition (EMT) of retinal pigment epithelium (RPE) cells is the key of the development of diabetic retinopathy (DR), and lncRNA NEAT1 could accelerate EMT in diabetic nephropathy. Meanwhile, as a diabetes susceptibility gene, whether sex-determining region Y-related (SRY) high-mobility group box 4 (SOX4) has relationship with lncRNA NEAT1 in DR remains unclear.

METHODS

Firstly, NEAT1, SOX4 and miR-204 were evaluated by qRT-PCR (quantitative reverse-transcriptase PCR) under high glucose condition. Then, cell viability, proliferation, migration and invasion were respectively detected by MTT, BrdU staining, wound healing and transwell assay after NEAT1 knockdown or miR-204 overexpression. Also, the EMT-related proteins were examined by western blot and cell immunofluorescence assay. In order to confirm the relationship between miR-204 and NEAT1 or SOX4, dual luciferase reporter gene assay was conducted. At the same time, the protein levels of SOX4 and EMT-related proteins were investigated by immunohistochemistry in vivo.

RESULTS

High glucose upregulated NEAT1 and SOX4 and downregulated miR-204 in ARPE19 cells. NEAT1 knockdown or miR-204 overexpression inhibited the proliferation and EMT progression of ARPE19 cells induced by high glucose. NEAT1 was identified as a molecular sponge of miR-204 to increase the level of SOX4. The effect of NEAT1 knockdown on the progression of EMT under high glucose condition in ARPE19 cells could be reversed by miR-204 inhibitor. Also, NEAT1 knockdown inhibited retinal EMT in diabetic mice.

CONCLUSION

NEAT1 regulated the development of EMT in DR through miR-204/SOX4 pathway, which could provide reference for clinical prevention and treatment.

Evidence-based Canadian guidelines for tele-retina screening for diabetic retinopathy: recommendations from the Canadian Retina Research Network (CR2N) Tele-Retina Steering Committee

OBJECTIVE

The purpose of this report is to develop a consensus for Canadian national guidelines specific to a tele-medicine approach to screening for diabetic retinopathy (DR) using evidence-based and clinical data.

METHODS

Canadian Tele-Screening Grading Scales for DR and diabetic macular edema (DME) were created primarily based on severity grading scales outlined by the International Clinical Diabetic Retinopathy Disease Severity Scale (ICDR) and the Scottish DR Grading Scheme 2007. Other grading scales used in international screening programs and the clinical expertise of the Canadian Retina Research Network members and retina specialists nationwide were also used in the creation of the guidelines.

RESULTS

National Tele-Screening Guidelines for DR and DME with and without optical coherence tomography (OCT) images are proposed. These outline a diagnosis and management algorithm for patients presenting with different stages of DR and/or DME. General guidelines detailing the requirements for imaged retina fields, image quality, quality control, and follow-up care and the role of visual acuity, pupil dilation, OCT, ultra-wide-field imaging, and artificial intelligence are discussed.

CONCLUSIONS

Tele-retina screening can help to address the need for timely and effective screening for DR, whose prevalence continues to rise. A standardized and evidence-based national approach to DR tele-screening has been proposed, based on DR/DME grading using two 45° image fields or a single widefield or ultra-wide-field image, preferable use of OCT imaging, and a focus on local quality control measures.

Neurovascular regulation in diabetic retinopathy and emerging therapies

Diabetic retinopathy (DR) is the leading cause of vision loss in working adults in developed countries. The disease traditionally classified as a microvascular complication of diabetes is now widely recognized as a neurovascular disorder resulting from disruption of the retinal neurovascular unit (NVU). The NVU comprising retinal neurons, glia and vascular cells coordinately regulates blood flow, vascular density and permeability to maintain homeostasis. Disturbance of the NVU during DR can lead to vision-threatening clinical manifestations. A limited number of signaling pathways have been identified for intercellular communication within the NVU, including vascular endothelial growth factor (VEGF), the master switch for angiogenesis. VEGF inhibitors are now widely used to treat DR, but their limited efficacy implies that other signaling molecules are involved in the pathogenesis of DR. By applying a novel screening technology called comparative ligandomics, we recently discovered secretogranin III (Scg3) as a unique DR-selective angiogenic and vascular leakage factor with therapeutic potential for DR. This review proposes neuron-derived Scg3 as the first diabetes-selective neurovascular regulator and discusses important features of Scg3 inhibition for next-generation disease-targeted anti-angiogenic therapies of DR.

Advancing Diabetic Retinopathy Research: Analysis of the Neurovascular Unit in Zebrafish

Diabetic retinopathy is one of the most important microvascular complications associated with diabetes mellitus, and a leading cause of vision loss or blindness worldwide. Hyperglycaemic conditions disrupt microvascular integrity at the level of the neurovascular unit. In recent years, zebrafish (Danio rerio) have come into focus as a model organism for various metabolic diseases such as diabetes. In both mammals and vertebrates, the anatomy and the function of the retina and the neurovascular unit have been highly conserved. In this review, we focus on the advances that have been made through studying pathologies associated with retinopathy in zebrafish models of diabetes. We discuss the different cell types that form the neurovascular unit, their role in diabetic retinopathy and how to study them in zebrafish. We then present new insights gained through zebrafish studies. The advantages of using zebrafish for diabetic retinopathy are summarised, including the fact that the zebrafish has, so far, provided the only animal model in which hyperglycaemia-induced retinal angiogenesis can be observed. Based on currently available data, we propose potential investigations that could advance the field further.

Reversed Neurovascular Coupling on Optical Coherence Tomography Angiography Is the Earliest Detectable Abnormality before Clinical Diabetic Retinopathy

Diabetic retinopathy (DR) has traditionally been viewed as either a microvasculopathy or a neuropathy, though neurovascular coupling deficits have also been reported and could potentially be the earliest derangement in DR. To better understand neurovascular coupling in the diabetic retina, we investigated retinal hemodynamics by optical coherence tomography angiography (OCTA) in individuals with diabetes mellitus (DM) but without DR (DM no DR) and mild non-proliferative DR (mild NPDR) compared to healthy eyes. Using an experimental design to monitor the capillary responses during transition from dark adaptation to light, we examined 19 healthy, 14 DM no DR and 11 mild NPDR individuals. We found that the only structural vascular abnormality in the DM no DR group was increased superficial capillary plexus (SCP) vessel density (VD) compared to healthy eyes, while mild NPDR eyes showed significant vessel loss in the SCP at baseline. There was no significant difference in inner retinal thickness between the groups. During dark adaptation, the deep capillary plexus (DCP) VD was lower in mild NPDR individuals compared to the other two groups, which may leave the photoreceptors more susceptible to ischemia in the dark. When transitioning from dark to ambient light, both diabetic groups showed a qualitative reversal of VD trends in the SCP and middle capillary plexus (MCP), with significantly decreased SCP at 5 min and increased MCP VD at 50 s compared to healthy eyes, which may impede metabolic supply to the inner retina during light adaptation. Mild NPDR eyes also demonstrated DCP dilation at 50 s and 5 min and decreased adjusted flow index at 5 min in light. Our results show altered neurovascular responses in all three macular vascular plexuses in diabetic subjects in the absence of structural neuronal changes on high resolution imaging, suggesting that neurovascular uncoupling may be a key mechanism in the early pathogenesis of DR, well before the clinical appearance of vascular or neuronal loss.

Diabetic Retinopathy and Risk of Stroke: A Secondary Analysis of the ACCORD Eye Study

BACKGROUND AND PURPOSE

Diabetic retinopathy (DR) is a common microvascular complication of diabetes, which causes damage to the retina and may lead to rapid vision loss. Previous research has shown that the macrovascular complications of diabetes, including stroke, are often comorbid with DR. We sought to explore the association between DR and subsequent stroke events.

METHODS

This is a secondary analysis of patients enrolled in the ACCORD Eye study (Action to Control Cardiovascular Risk in Diabetes). The primary outcome was stroke during follow-up. The exposure was presence of DR at study baseline. We fit adjusted Cox proportional hazards models to provide hazard ratios for stroke and included interaction terms with the ACCORD randomization arms.

RESULTS

We included 2828 patients, in whom the primary outcome of stroke was met by 117 (4.1%) patients during a mean (SD) of 5.4 (1.8) years of follow-up. DR was present in 874 of 2828 (30.9%) patients at baseline and was more common in patients with than without incident stroke (41.0% versus 30.5%; P=0.016). In an adjusted Cox regression model, DR was independently associated with incident stroke (hazard ratio, 1.52 [95% CI, 1.05-2.20]; P=0.026). This association was not affected by randomization arm in the ACCORD glucose (P=0.300), lipid (P=0.660), or blood pressure interventions (P=0.469).

CONCLUSIONS

DR is associated with an increased risk of stroke, which suggests that the microvascular pathology inherent to DR has larger cerebrovascular implications. This association appears not to be mediated by serum glucose, lipid, and blood pressure interventions.

Pediatric diabetic retinopathy telescreening

PURPOSE

To describe the role of telemedicine screening for pediatric diabetic retinopathy (DR) and to identify risk factors for pediatric DR.

METHODS

The medical records of a telemedicine program at a tertiary, academic medical center over 17 months were reviewed retrospectively. Patients visiting an academic pediatric endocrinology clinic who met guidelines underwent telescreening. Presence of pediatric DR and risk factors for retinopathy were evaluated.

RESULTS

The fundus photographs of 852 patients 10-23 years of age were reviewed. Diabetic retinopathy was noted in 51 (6%). Patients with an abnormal screening photograph were compared to patients with diabetes who had normal screening photographs (n = 64). Older age, longer diabetes duration, type 1 diabetes, and higher average glycated hemoglobin (HbA1c) from the year prior to the photograph were associated with increased risk of retinopathy. Of these, longer duration (P = 0.003) and higher average A1c (P = 0.02) were significant after adjusting for sex, race, and age.

CONCLUSIONS

Our telemedicine program found a higher percentage of diabetic retinopathy screening non-mydriatic photographs than prior studies found through standard ophthalmic examinations. In this relatively small sample size, longer duration of disease and higher average A1c were associated with increased risk of having diabetic retinopathy in our study.

Remote Tool-Based Adjudication for Grading Diabetic Retinopathy

Purpose

To present and evaluate a remote, tool-based system and structured grading rubric for adjudicating image-based diabetic retinopathy (DR) grades.

Methods

We compared three different procedures for adjudicating DR severity assessments among retina specialist panels, including (1) in-person adjudication based on a previously described procedure (Baseline), (2) remote, tool-based adjudication for assessing DR severity alone (TA), and (3) remote, tool-based adjudication using a feature-based rubric (TA-F). We developed a system allowing graders to review images remotely and asynchronously. For both TA and TA-F approaches, images with disagreement were reviewed by all graders in a round-robin fashion until disagreements were resolved. Five panels of three retina specialists each adjudicated a set of 499 retinal fundus images (1 panel using Baseline, 2 using TA, and 2 using TA-F adjudication). Reliability was measured as grade agreement among the panels using Cohen's quadratically weighted kappa. Efficiency was measured as the number of rounds needed to reach a consensus for tool-based adjudication.

Results

The grades from remote, tool-based adjudication showed high agreement with the Baseline procedure, with Cohen's kappa scores of 0.948 and 0.943 for the two TA panels, and 0.921 and 0.963 for the two TA-F panels. Cases adjudicated using TA-F were resolved in fewer rounds compared with TA (P < 0.001; standard permutation test).

Conclusions

Remote, tool-based adjudication presents a flexible and reliable alternative to in-person adjudication for DR diagnosis. Feature-based rubrics can help accelerate consensus for tool-based adjudication of DR without compromising label quality.

Translational Relevance

This approach can generate reference standards to validate automated methods, and resolve ambiguous diagnoses by integrating into existing telemedical workflows.

Comparison between Tear Film Osmolar Cocentration and Other Tear Film Function Parameters in Patients with Diabetes Mellitus

Purpose

To evaluate tear film function in patients with diabetes mellitus (DM) using tear film osmolarity (TFO) measurements compared to other tear film function tests.

Methods

DM patients without any history of ocular surface disorder but with potential effects on the tear film were enrolled in this cross-sectional study. Data including dry eye symptoms, duration of DM, stage of diabetic retinopathy and blood hemoglobin A1c levels were recorded. Tear film break-up time (TBUT) and basic tear secretion (Schirmer test) were assessed. TFO was determined using the Tearlab Osmolarity System. The outcome measures were the difference between the mean values of TBUT, basic tear secretion and TFO in both the study and control groups.

Results

We recruited 51 DM patients and 20 control subjects with a mean age of 51.2 (range, 21 to 70) and 48.5 (range, 24 to 70) years, respectively. A total of 27 patients (53%) and 11 controls (55%) reported dry eye symptoms (p = 0.668). The mean TBUT was 10.2 ± 4.8 seconds in the study group versus 10.5 ± 2.8 seconds in controls, which was not significantly different (p = 0.747). The mean Schirmer test score was 8.1 ± 4.3 mm in the patients versus 10.1 ± 3.0 mm in the controls (p = 0.069). The mean TFO was 294.1 ± 12.9 mosmol/L in the patients versus 291.4 ± 14.5 mosmol/L in the controls (p = 0.456). It was significantly higher in patients with poor glycemic control determined by hemoglobin A1c > 8% (p = 0.003). TFO had a positive correlation with the duration of DM (p = 0.030) but not with the stage of diabetic retinopathy (p = 0.944). However, TFO showed a significant relationship with dry eye symptoms (p = 0.001).

Conclusions

TFO is impaired in patients with uncontrolled DM and is better correlated with glycemic control and dry eye symptoms than the TBUT and Schirmer tests.

Update on Screening for Sight-Threatening Diabetic Retinopathy

PURPOSE

The aim of this article was to describe recent advances in the use of new technology in diabetic retinopathy screening by looking at studies that assessed the effectiveness and cost-effectiveness of these technologies.

METHODS

The author conducts an ongoing search for articles relating to screening or management of diabetic retinopathy utilising Zetoc with keywords and contents page lists from relevant journals.

RESULTS

The areas discussed in this article are reference standards, alternatives to digital photography, area of retina covered by the screening method, size of the device and hand-held cameras, mydriasis versus non-mydriasis or a combination, measurement of distance visual acuity, grading of images, use of automated grading analysis and cost-effectiveness of the new technologies.

CONCLUSIONS

There have been many recent advances in technology that may be adopted in the future by screening programmes for sight-threatening diabetic retinopathy but each device will need to demonstrate effectiveness and cost-effectiveness before more widespread adoption.

Fundus photograph-based deep learning algorithms in detecting diabetic retinopathy

Remarkable advances in biomedical research have led to the generation of large amounts of data. Using artificial intelligence, it has become possible to extract meaningful information from large volumes of data, in a shorter frame of time, with very less human interference. In effect, convolutional neural networks (a deep learning method) have been taught to recognize pathological lesions from images. Diabetes has high morbidity, with millions of people who need to be screened for diabetic retinopathy (DR). Deep neural networks offer a great advantage of screening for DR from retinal images, in improved identification of DR lesions and risk factors for diseases, with high accuracy and reliability. This review aims to compare the current evidences on various deep learning models for diagnosis of diabetic retinopathy (DR).

Deep Capillary Macular Perfusion Indices Obtained with OCT Angiography Correlate with Degree of Nonproliferative Diabetic Retinopathy

PURPOSE

To evaluate the integrity of macular and temporomacular vasculature in nonproliferative diabetic retinopathy (NPDR) with noninvasive optical coherence tomography angiography (OCTA) and correlate perfusion indices with degree of NPDR.

METHODS

In this prospective observational cross-sectional study, 102 eyes with newly diagnosed NPDR (mild NPDR, 36; moderate NPDR, 21; severe NPDR, 13; NPDR with diabetic macular edema [DME], 32) underwent OCTA. Sixty eyes of normal subjects served as control. Degree of NPDR (based on Early Treatment Diabetic Retinopathy Study criteria) was confirmed with fluorescein angiography. Automated OCTA/split-spectrum amplitude decorrelation angiography software generated perfusion indices (vessel density and flow index) from images of the retina. The perfusion index of superficial and deep retinal plexuses was obtained in both perifoveal (central 1-3 mm) and parafoveal (3-6 mm) areas.

RESULTS

Deep plexus parafoveal vessel density was 25.23% (±6.1) in mild NPDR, 20.16% (±6.16) in moderate NPDR, 11.16% (±4.16) in severe NPDR, and 17.91% (±4.42) in NPDR + DME compared to normal subjects (36.93% [±8.1]; (p<0.01). Spearman correlation coefficient (r_s) between vessel density and level of NPDR severity in the parafoveal region showed inverse correlation for both superficial (r_s -0.87; p = 0.083) and deep (r_s -0.96; p = 0.017) plexus. Similarly, when vessel density of the perifoveal region was compared with level of NPDR severity, inverse correlation was noted in both superficial (r_s -0.85; p = 0.08) and deep (r_s -0.98; p = 0.011) plexus.

CONCLUSIONS

Optical coherence tomography angiography clearly delineated the retinal microcirculation and allowed quantification of vascular perfusion of each layer. As diabetic retinopathy progressed, a decrease in perfusion index is more pronounced in the deep retinal plexus and precedes changes in superficial plexus.

Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices

Artificial Intelligence (AI) has long promised to increase healthcare affordability, quality and accessibility but FDA, until recently, had never authorized an autonomous AI diagnostic system. This pivotal trial of an AI system to detect diabetic retinopathy (DR) in people with diabetes enrolled 900 subjects, with no history of DR at primary care clinics, by comparing to Wisconsin Fundus Photograph Reading Center (FPRC) widefield stereoscopic photography and macular Optical Coherence Tomography (OCT), by FPRC certified photographers, and FPRC grading of Early Treatment Diabetic Retinopathy Study Severity Scale (ETDRS) and Diabetic Macular Edema (DME). More than mild DR (mtmDR) was defined as ETDRS level 35 or higher, and/or DME, in at least one eye. AI system operators underwent a standardized training protocol before study start. Median age was 59 years (range, 22–84 years); among participants, 47.5% of participants were male; 16.1% were Hispanic, 83.3% not Hispanic; 28.6% African American and 63.4% were not; 198 (23.8%) had mtmDR. The AI system exceeded all pre-specified superiority endpoints at sensitivity of 87.2% (95% CI, 81.8–91.2%) (>85%), specificity of 90.7% (95% CI, 88.3–92.7%) (>82.5%), and imageability rate of 96.1% (95% CI, 94.6–97.3%), demonstrating AI’s ability to bring specialty-level diagnostics to primary care settings. Based on these results, FDA authorized the system for use by health care providers to detect more than mild DR and diabetic macular edema, making it, the first FDA authorized autonomous AI diagnostic system in any field of medicine, with the potential to help prevent vision loss in thousands of people with diabetes annually. ClinicalTrials.gov NCT02963441

Development and Validation of a Diabetic Retinopathy Referral Algorithm Based on Single-Field Fundus Photography

Purpose

To develop a simplified algorithm to identify and refer diabetic retinopathy (DR) from single-field retinal images specifically for sight-threatening diabetic retinopathy for appropriate care (ii) to determine the agreement and diagnostic accuracy of the algorithm as a pilot study among optometrists versus “gold standard” (retinal specialist grading).

Methods

The severity of DR was scored based on colour photo using a colour coded algorithm, which included the lesions of DR and number of quadrants involved. A total of 99 participants underwent training followed by evaluation. Data of the 99 participants were analyzed. Fifty posterior pole 45 degree retinal images with all stages of DR were presented. Kappa scores (κ), areas under the receiver operating characteristic curves (AUCs), sensitivity and specificity were determined, with further comparison between working optometrists and optometry students.

Results

Mean age of the participants was 22 years (range: 19–43 years), 87% being women. Participants correctly identified 91.5% images that required immediate referral (κ) = 0.696), 62.5% of images as requiring review after 6 months (κ = 0.462), and 51.2% of those requiring review after 1 year (κ = 0.532). The sensitivity and specificity of the optometrists were 91% and 78% for immediate referral, 62% and 84% for review after 6 months, and 51% and 95% for review after 1 year, respectively. The AUC was the highest (0.855) for immediate referral, second highest (0.824) for review after 1 year, and 0.727 for review after 6 months criteria. Optometry students performed better than the working optometrists for all grades of referral.

Conclusions

The diabetic retinopathy algorithm assessed in this work is a simple and a fairly accurate method for appropriate referral based on single-field 45 degree posterior pole retinal images.

Association of the Fractal Dimension of Retinal Arteries and Veins with Quantitative Brain MRI Measures in HIV-Infected and Uninfected Women

Objective

The fractal dimension of retinal arteries and veins is a measure of the complexity of the vascular tree. We hypothesized that retinal fractal dimension would be associated with brain volume and white matter integrity in HIV-infected women.

Design

Nested case-control within longitudinal cohort study.

Methods

Women were recruited from the Brooklyn site of the Women’s Interagency HIV study (WIHS); 34 HIV-infected and 21 HIV-uninfected women with analyzable MRIs and retinal photographs were included. Fractal dimension was determined using the SIVA software program on skeletonized retinal images. The relationship between predictors (retinal vascular measures) and outcomes (quantitative MRI measures) were analyzed with linear regression models. All models included age, intracranial volume, and both arterial and venous fractal dimension. Some models were adjusted for blood pressure, race/ethnicity, and HIV-infection.

Results

The women were 45.6 ± 7.3 years of age. Higher arterial dimension was associated with larger cortical volumes, but higher venous dimension was associated with smaller cortical volumes. In fully adjusted models, venous dimension was significantly associated with fractional anisotropy (standardized β = -0.41, p = 0.009) and total gray matter volume (β = -0.24, p = 0.03), and arterial dimension with mean diffusivity (β = -0.33,.p = 0.04) and fractional anisotropy (β = 0.34, p = 0.03). HIV-infection was not associated with any retinal or MRI measure.

Conclusions

Higher venous fractal dimension was associated with smaller cortical volumes and lower fractional anisotropy, whereas higher arterial fractal dimension was associated with the opposite patterns. Longitudinal studies are needed to validate this finding.

Genome-wide association study for sight-threatening diabetic retinopathy reveals association with genetic variation near the GRB2 gene

AIMS/HYPOTHESIS

Diabetic retinopathy is a serious complication of diabetes mellitus and can lead to blindness. A genetic component, in addition to traditional risk factors, has been well described although strong genetic factors have not yet been identified. Here, we aimed to identify novel genetic risk factors for sight-threatening diabetic retinopathy using a genome-wide association study.

METHODS

Retinopathy was assessed in white Australians with type 2 diabetes mellitus. Genome-wide association analysis was conducted for comparison of cases of sight-threatening diabetic retinopathy (n = 336) with diabetic controls with no retinopathy (n = 508). Top ranking single nucleotide polymorphisms were typed in a type 2 diabetes replication cohort, a type 1 diabetes cohort and an Indian type 2 cohort. A mouse model of proliferative retinopathy was used to assess differential expression of the nearby candidate gene GRB2 by immunohistochemistry and quantitative western blot.

RESULTS

The top ranked variant was rs3805931 with p = 2.66 × 10(-7), but no association was found in the replication cohort. Only rs9896052 (p = 6.55 × 10(-5)) was associated with sight-threatening diabetic retinopathy in both the type 2 (p = 0.035) and the type 1 (p = 0.041) replication cohorts, as well as in the Indian cohort (p = 0.016). The study-wide meta-analysis reached genome-wide significance (p = 4.15 × 10(-8)). The GRB2 gene is located downstream of this variant and a mouse model of retinopathy showed increased GRB2 expression in the retina.

CONCLUSIONS/INTERPRETATION

Genetic variation near GRB2 on chromosome 17q25.1 is associated with sight-threatening diabetic retinopathy. Several genes in this region are promising candidates and in particular GRB2 is upregulated during retinal stress and neovascularisation.

Risk Factors for Retinopathy and DME in Type 2 Diabetes-Results from the German/Austrian DPV Database

To assess the prevalence and risk factors for early and severe diabetic retinopathy and macular edema in a large cohort of patients with type 2 diabetes Retinopathy grading (any retinopathy, severe retinopathy, diabetic macular edema) and risk factors of 64784 were prospectively recorded between January 2000 and March 2013 and analyzed by Kaplan–Meier analysis and logistic regression. Retinopathy was present in 20.12% of subjects, maculopathy was found in 0.77%. HbA1c > 8%, microalbuminuria, hypertension, BMI > 35 kg/m² and male sex were significantly associated with any retinopathy, while HbA1c and micro- and macroalbuminuria were the strongest risk predictors for severe retinopathy. Presence of macroalbuminuria increased the risk for DME by 177%. Retinopathy remains a significant clinical problem in patients with type 2 diabetes. Metabolic control and blood pressure are relevant factors amenable to treatment. Concomitant kidney disease identifies high risk patients and should be emphasized in interdisciplinary communication.