DETECTION OF MORPHOLOGIC PATTERNS OF DIABETIC MACULAR EDEMA USING A DEEP LEARNING APPROACH BASED ON OPTICAL COHERENCE TOMOGRAPHY IMAGES

Artificial intelligence for early detection of diabetes mellitus complications via retinal imaging

Background

Diabetes mellitus (DM) increases the risk of vascular complications, and retinal vasculature imaging serves as a valuable indicator of both microvascular and macrovascular health. Moreover, artificial intelligence (AI)-enabled systems developed for high-throughput detection of diabetic retinopathy (DR) using digitized retinal images have become clinically adopted. This study reviews AI applications using retinal images for DM-related complications, highlighting advancements beyond DR screening, diagnosis, and prognosis, and addresses implementation challenges, such as ethics, data privacy, equitable access, and explainability.

Methods

We conducted a thorough literature search across several databases, including PubMed, Scopus, and Web of Science, focusing on studies involving diabetes, the retina, and artificial intelligence. We reviewed the original research based on their methodology, AI algorithms, data processing techniques, and validation procedures to ensure a detailed analysis of AI applications in diabetic retinal imaging.

Results

Retinal images can be used to diagnose DM complications including DR, neuropathy, nephropathy, and atherosclerotic cardiovascular disease, as well as to predict the risk of cardiovascular events. Beyond DR screening, AI integration also offers significant potential to address the challenges in the comprehensive care of patients with DM.

Conclusion

With the ability to evaluate the patient's health status in relation to DM complications as well as risk prognostication of future cardiovascular complications, AI-assisted retinal image analysis has the potential to become a central tool for modern personalized medicine in patients with DM.

Unsupervised machine learning analysis of optical coherence tomography radiomics features for predicting treatment outcomes in diabetic macular edema

This study aimed to identify distinct clusters of diabetic macular edema (DME) patients with differential anti-vascular endothelial growth factor (VEGF) treatment outcomes using an unsupervised machine learning (ML) approach based on radiomic features extracted from pre-treatment optical coherence tomography (OCT) images. Retrospective data from 234 eyes with DME treated with three anti-VEGF therapies between January 2020 and March 2024 were collected from two clinical centers. Radiomic analysis was conducted on pre-treatment OCT images. Following principal component analysis (PCA) for dimensionality reduction, two unsupervised clustering methods (K-means and hierarchical clustering) were applied. Baseline characteristics and treatment outcomes were compared across clusters to assess clustering efficacy. Feature selection employed a three-stage pipeline: exclusion of collinear features (Pearson's r > 0.8); sequential filtering through ANOVA (P < 0.05) and Boruta algorithm (500 iterations); multivariate stepwise regression (entry criteria: univariate P < 0.1) to identify outcome-associated predictors. From 1165 extracted radiomic features, four distinct DME clusters were identified. Cluster 4 exhibited a significantly lower incidence of residual/recurrent DME (RDME) (34.29%) compared to Clusters 1-3 (P = 0.003, P = 0.005 and P = 0.002, respectively). This cluster also demonstrated the highest proportion of eyes (71.43%) with best-corrected visual acuity (BCVA) exceeding 20/63 (P = 0.003, P = 0.005 and P = 0.002, respectively). Multivariate analysis identified logarithm_gldm_DependenceVariance as an independent risk factor for RDME (OR 1.75, 95% CI 1.28-2.40; P < 0.001), while Wavelet-LH_Firstorder_Mean correlated with worse visual outcomes (OR 8.76, 95% CI 1.22-62.84; P = 0.031). Unsupervised ML leveraging pre-treatment OCT radiomics successfully stratifies DME eyes into clinically distinct subgroups with divergent therapeutic responses. These quantitative features may serve as non-invasive biomarkers for personalized outcome prediction and retinal pathology assessment.

Detection of diabetic macular oedema patterns with fine-grained image categorisation on optical coherence tomography

PURPOSE

To develop an artificial intelligence (AI) system for detecting pathological patterns of diabetic macular oedema (DME) with fine-grained image categorisation using optical coherence tomography (OCT) images.

METHODS

The development of the AI system consists of two parts, a pretraining process on a public dataset (Asia Pacific Tele-Ophthalmology Society (APTOS)), and the training process on the local dataset. The local dataset was partitioned into the training set, validation set and test set in the ratio of 6:2:2. The Split Subspace Attention Network (SSA-Net) architecture was adopted to train independent models to detect the seven pathological patterns of DME: intraretinal fluid (IRF), subretinal fluid (SRF), pigment epithelial detachment (PED), hyper-reflective retinal foci (HRF), Müller cell cone disruption (MCCD), subretinal hyper-reflective material (SHRM) and intra-cystic hyper-reflective material (ICHRM). The confusion matrix, sensitivity, specificity and receiver operating characteristic (ROC) were used to evaluate the performance of the models.

RESULTS

The APTOS public dataset consists of 33 853 OCT images and the local dataset consists of 1346 OCT images with DME. In the pretraining process on the public dataset, the accuracy was 0.652 for IRF, 0.928 for SRF, 0.936 for PED and 0.975 for HRF. After the training process on the local dataset, the SSA-Net architecture showed better performance in fine-grained image categorisation on the test set. The area under the ROC curve was 0.980 for IRF, 0.995 for SRF, 0.999 for PED, 0.908 for MCCD, 0.887 for HRF, 0.990 for SHRM and 0.972 for ICHRM. The AI system outputs a heatmap for each result, which can give a visual explanation for the automated identification process. The heatmaps revealed that the regions of interest of the AI system were consistent with the retinal experts.

CONCLUSIONS

The proposed SSA-Net architecture for detecting the pathological patterns of DME achieved satisfactory accuracy. However, this study was conducted in one medical centre, and multicentre trials will be needed in the future.

Comparison of the effect of ranibizumab in retinal vein occlusion and macular edema with different optical coherence tomographic patterns

AIM

To explore the morphological and functional parameters to evaluate the effectiveness of intravitreal injections of ranibizumab (IVR) in treating macular edema (ME) secondary to retinal vein occlusion (RVO).

METHODS

This retrospective study involved 65 RVO patients (65 eyes) who received IVR and were followed-up for more than 3mo. ME was categorized into cystoid macular edema (CME), diffuse retinal thickening (DRT), and serous retinal detachment (SRD) according to optical coherence tomography (OCT) images. The comparison of best corrected visual acuity (BCVA; logMAR) and central macular thickness (CMT) among different follow-up points and those among 3 groups were performed by Kruskal-Wallis test. The correlation between BCVA and baseline parameters during treatment was analyzed using Spearman correlation analysis.

RESULTS

BCVA tended to improve in all groups, with marked improvement in CME and DRT groups. CMT showed the greatest reduction after 1wk, and remained stable over the following 3mo. DRT patients had the worst BCVA and the highest CMT at baseline, but the differences became smaller after IVR treatment. CMT in SRD group was significantly better than in CME and DRT groups 3mo after IVR. Most patients of CME and SRD groups transitioned to a normal pattern at 3mo follow-up. DRT patients were most likely to transform into the other morphological groups, while SRD patients showed minimal transitions. BCVA at baseline was identified as the most important prognostic indicator in all 3 groups. Additionally, DRT patients with a longer clinical course, higher CMT and central retinal vein occlusion (CRVO) tend to exhibit worse BCVA after treatment. In addition, CRVO patients are more likely to have worse BCVA at 2 and 3mo follow-up compared with branch retinal vein occlusion (BRVO) patients in CME group. SRD patients with higher baseline CMT were prone to experiencing worse BCVA after treatment.

CONCLUSION

The effectiveness of IVR is strongly correlated with baseline BCVA in all 3 groups. Baseline parameters including clinical course, CMT, and RVO position are also useful in predicting the BCVA at different time points after treatment.

Enhancing AI reliability: A foundation model with uncertainty estimation for optical coherence tomography-based retinal disease diagnosis

Inability to express the confidence level and detect unseen disease classes limits the clinical implementation of artificial intelligence in the real world. We develop a foundation model with uncertainty estimation (FMUE) to detect 16 retinal conditions on optical coherence tomography (OCT). In the internal test set, FMUE achieves a higher F1 score of 95.74% than other state-of-the-art algorithms (92.03%-93.66%) and improves to 97.44% with threshold strategy. The model achieves similar excellent performance on two external test sets from the same and different OCT machines. In human-model comparison, FMUE achieves a higher F1 score of 96.30% than retinal experts (86.95%, p = 0.004), senior doctors (82.71%, p < 0.001), junior doctors (66.55%, p < 0.001), and generative pretrained transformer 4 with vision (GPT-4V) (32.39%, p < 0.001). Besides, FMUE predicts high uncertainty scores for >85% images of non-target-category diseases or with low quality to prompt manual checks and prevent misdiagnosis. Our FMUE provides a trustworthy method for automatic retinal anomaly detection in a clinical open-set environment.

Multimodal imaging in diabetic retinopathy and macular edema: An update about biomarkers

Diabetic macular edema (DME), defined as retinal thickening near, or involving the fovea caused by fluid accumulation in the retina, can lead to vision impairment and blindness in patients with diabetes. Current knowledge of retina anatomy and function and DME pathophysiology has taken great advantage of the availability of several techniques for visualizing the retina. Combining these techniques in a multimodal imaging approach to DME is recommended to improve diagnosis and to guide treatment decisions. We review the recent literature about the following retinal imaging technologies: optical coherence tomography (OCT), OCT angiography (OCTA), wide-field and ultrawide-field techniques applied to fundus photography, fluorescein angiography, and OCTA. The emphasis will be on characteristic DME features identified by these imaging technologies and their potential or established role as diagnostic, prognostic, or predictive biomarkers. The role of artificial intelligence in the assessment and interpretation of retina images is also discussed.

Effect of anti-vascular endothelial growth factor on early-stage post-vitrectomy macular edema in patients with proliferative diabetic retinopathy

PURPOSE

To investigate the effectiveness of anti-vascular endothelial growth factor (VEGF) therapy on post-vitrectomy macular edema (PVME) and determine the risk factors for PVME recovery.

METHODS

This retrospective study included 179 eyes of 179 patients who underwent pars plana vitrectomy for proliferative diabetic retinopathy and developed PVME within 3 months after surgery. Eyes were grouped according to postoperative anti-VEGF treatment.

RESULTS

Central retinal thickness (CRT) decreased significantly from baseline to 3-month follow-up in groups with (509.9 ± 157.2 μm vs. 401.2 ± 172.1 μm, P < 0.001) or without (406.1 ± 96.1 μm vs. 355.1 ± 126.0 μm, P = 0.008) postoperative anti-VEGF treatment. Best-corrected visual acuity (BCVA) did not differ between the two groups during follow-up. In the group not receiving anti-VEGF therapy, BCVA was significantly improved at 1, 2, and 3 months (P = 0.007, P < 0.001, and P < 0.001, respectively), while in the anti-VEGF group, BCVA was significantly improved at 1 and 3 months (P = 0.03 and P < 0.001). A thicker baseline CRT (β = 0.44; 95% confidence interval, 0.26-0.61; P < 0.001) was significantly associated with decreasing CRT.

CONCLUSION

PVME tends to spontaneously resolve in the early postoperative period. The effect of anti-VEGF therapy in the first 3 months after diagnosis appears to be limited.

Fenofibrate and diabetic retinopathy

Background

Diabetic retinopathy (DR), a sight-threatening ocular complication of diabetes mellitus, is one of the main causes of blindness in the working-age population. Dyslipidemia is a potential risk factor for the development or worsening of DR, with conflicting evidence in epidemiological studies. Fenofibrate, an antihyperlipidemic agent, has lipid-modifying and pleiotropic (non-lipid) effects that may lessen the incidence of microvascular events.

Methods

Relevant studies were identified through a PubMed/MEDLINE search spanning the last 20 years, using the broad term "diabetic retinopathy" and specific terms "fenofibrate" and "dyslipidemia". References cited in these studies were further examined to compile this mini-review. These pivotal investigations underwent meticulous scrutiny and synthesis, focusing on methodological approaches and clinical outcomes. Furthermore, we provided the main findings of the seminal studies in a table to enhance comprehension and comparison.

Results

Growing evidence indicates that fenofibrate treatment slows DR advancement owing to its possible protective effects on the blood-retinal barrier. The protective attributes of fenofibrate against DR progression and development can be broadly classified into two categories: lipid-modifying effects and non-lipid-related (pleiotropic) effects. The lipid-modifying effect is mediated through peroxisome proliferator-activated receptor-α activation, while the pleiotropic effects involve the reduction in serum levels of C-reactive protein, fibrinogen, and pro-inflammatory markers, and improvement in flow-mediated dilatation. In patients with DR, the lipid-modifying effects of fenofibrate primarily involve a reduction in lipoprotein-associated phospholipase A2 levels and the upregulation of apolipoprotein A1 levels. These changes contribute to the anti-inflammatory and anti-angiogenic effects of fenofibrate. Fenofibrate elicits a diverse array of pleiotropic effects, including anti-apoptotic, antioxidant, anti-inflammatory, and anti-angiogenic properties, along with the indirect consequences of these effects. Two randomized controlled trials-the Fenofibrate Intervention and Event Lowering in Diabetes and Action to Control Cardiovascular Risk in Diabetes studies-noted that fenofibrate treatment protected against DR progression, independent of serum lipid levels.

Conclusions

Fenofibrate, an oral antihyperlipidemic agent that is effective in decreasing DR progression, may reduce the number of patients who develop vision-threatening complications and require invasive treatment. Despite its proven protection against DR progression, fenofibrate treatment has not yet gained wide clinical acceptance in DR management. Ongoing and future clinical trials may clarify the role of fenofibrate treatment in DR management.

Classification of diabetic maculopathy based on optical coherence tomography images using a Vision Transformer model

PURPOSE

To develop a Vision Transformer model to detect different stages of diabetic maculopathy (DM) based on optical coherence tomography (OCT) images.

METHODS

After removing images with poor quality, a total of 3319 OCT images were extracted from the Eye Center of the Renmin Hospital of Wuhan University and randomly split the images into training and validation sets in a 7:3 ratio. All macular cross-sectional scan OCT images were collected retrospectively from the eyes of DM patients from 2016 to 2022. One of the OCT stages of DM, including early diabetic macular oedema (DME), advanced DME, severe DME and atrophic maculopathy, was labelled on the collected images, respectively. A deep learning (DL) model based on Vision Transformer was trained to detect four OCT grading of DM.

RESULTS

The model proposed in our paper can provide an impressive detection performance. We achieved an accuracy of 82.00%, an F1 score of 83.11%, an area under the receiver operating characteristic curve (AUC) of 0.96. The AUC for the detection of four OCT grading (ie, early DME, advanced DME, severe DME and atrophic maculopathy) was 0.96, 0.95, 0.87 and 0.98, respectively, with an accuracy of 90.87%, 89.96%, 94.42% and 95.13%, respectively, a precision of 88.46%, 80.31%, 89.42% and 87.74%, respectively, a sensitivity of 87.03%, 88.18%, 63.39% and 89.42%, respectively, a specificity of 93.02%, 90.72%, 98.40% and 96.66%, respectively and an F1 score of 87.74%, 84.06%, 88.18% and 88.57%, respectively.

CONCLUSION

Our DL model based on Vision Transformer demonstrated a relatively high accuracy in the detection of OCT grading of DM, which can help with patients in a preliminary screening to identify groups with serious conditions. These patients need a further test for an accurate diagnosis, and a timely treatment to obtain a good visual prognosis. These results emphasised the potential of artificial intelligence in assisting clinicians in developing therapeutic strategies with DM in the future.

Detection and Classification of Diabetic Macular Edema with a Desktop-Based Code-Free Machine Learning Tool

Objectives

To evaluate the effectiveness of the Lobe application, a machine learning (ML) tool that can be used on a personal computer without requiring coding expertise, in the recognition and classification of diabetic macular edema (DME) in spectral-domain optical coherence tomography (SD-OCT) scans.

Materials and Methods

A total of 695 cross-sectional SD-OCT images from 336 patients with DME and 200 OCT images of 200 healthy controls were included. Images with DME were classified into three main types: diffuse retinal edema (DRE), cystoid macular edema (CME), and cystoid macular degeneration (CMD). To develop the ML model, we used the desktop-based code-free Lobe application, which includes a pre-trained ResNet-50 V2 convolutional neural network and is available free of charge. The performance of the trained model in recognizing and classifying DME was evaluated with 41 DRE, 28 CMD, 70 CME, and 40 normal SD-OCT images that were not used in the training.

Results

The developed model showed 99.28% sensitivity and 100% specificity for class-independent detection of DME. Sensitivity and specificity by labels were 87.80% and 98.57% for DRE, 96.43% and 99.29% for CME, and 95.71% and 95.41% for CMD, respectively.

Conclusion

To our knowledge, this is the first evaluation of the effectiveness of Lobe with ophthalmological images, and the results indicate that it can be used with high efficiency in the recognition and classification of DME from SD-OCT images by ophthalmologists without coding expertise.

DME-DeepLabV3+: a lightweight model for diabetic macular edema extraction based on DeepLabV3+ architecture

Introduction

Diabetic macular edema (DME) is a major cause of vision impairment in the patients with diabetes. Optical Coherence Tomography (OCT) is an important ophthalmic imaging method, which can enable early detection of DME. However, it is difficult to achieve high-efficiency and high-precision extraction of DME in OCT images because the sources of OCT images are diverse and the quality of OCT images is not stable. Thus, it is still required to design a model to improve the accuracy of DME extraction in OCT images.

Methods

A lightweight model (DME-DeepLabV3+) was proposed for DME extraction using a DeepLabV3+ architecture. In this model, MobileNetV2 model was used as the backbone for extracting low-level features of DME. The improved ASPP with sawtooth wave-like dilation rate was used for extracting high-level features of DME. Then, the decoder was used to fuse and refine low-level and high-level features of DME. Finally, 1711 OCT images were collected from the Kermany dataset and the Affiliated Eye Hospital. 1369, 171, and 171 OCT images were randomly selected for training, validation, and testing, respectively.

Conclusion

In ablation experiment, the proposed DME-DeepLabV3+ model was compared against DeepLabV3+ model with different setting to evaluate the effects of MobileNetV2 and improved ASPP on DME extraction. DME-DeepLabV3+ had better extraction performance, especially in small-scale macular edema regions. The extraction results of DME-DeepLabV3+ were close to ground truth. In comparative experiment, the proposed DME-DeepLabV3+ model was compared against other models, including FCN, UNet, PSPNet, ICNet, and DANet, to evaluate DME extraction performance. DME-DeepLabV3+ model had better DME extraction performance than other models as shown by greater pixel accuracy (PA), mean pixel accuracy (MPA), precision (Pre), recall (Re), F1-score (F1), and mean Intersection over Union (MIoU), which were 98.71%, 95.23%, 91.19%, 91.12%, 91.15%, and 91.18%, respectively.

Discussion

DME-DeepLabV3+ model is suitable for DME extraction in OCT images and can assist the ophthalmologists in the management of ocular diseases.

A bibliometric analysis of artificial intelligence applications in macular edema: exploring research hotspots and Frontiers

Background: Artificial intelligence (AI) is used in ophthalmological disease screening and diagnostics, medical image diagnostics, and predicting late-disease progression rates. We reviewed all AI publications associated with macular edema (ME) research Between 2011 and 2022 and performed modeling, quantitative, and qualitative investigations. Methods: On 1st February 2023, we screened the Web of Science Core Collection for AI applications related to ME, from which 297 studies were identified and analyzed (2011-2022). We collected information on: publications, institutions, country/region, keywords, journal name, references, and research hotspots. Literature clustering networks and Frontier knowledge bases were investigated using bibliometrix-BiblioShiny, VOSviewer, and CiteSpace bibliometric platforms. We used the R "bibliometrix" package to synopsize our observations, enumerate keywords, visualize collaboration networks between countries/regions, and generate a topic trends plot. VOSviewer was used to examine cooperation between institutions and identify citation relationships between journals. We used CiteSpace to identify clustering keywords over the timeline and identify keywords with the strongest citation bursts. Results: In total, 47 countries published AI studies related to ME; the United States had the highest H-index, thus the greatest influence. China and the United States cooperated most closely between all countries. Also, 613 institutions generated publications - the Medical University of Vienna had the highest number of studies. This publication record and H-index meant the university was the most influential in the ME field. Reference clusters were also categorized into 10 headings: retinal Optical Coherence Tomography (OCT) fluid detection, convolutional network models, deep learning (DL)-based single-shot predictions, retinal vascular disease, diabetic retinopathy (DR), convolutional neural networks (CNNs), automated macular pathology diagnosis, dry age-related macular degeneration (DARMD), class weight, and advanced DL architecture systems. Frontier keywords were represented by diabetic macular edema (DME) (2021-2022). Conclusion: Our review of the AI-related ME literature was comprehensive, systematic, and objective, and identified future trends and current hotspots. With increased DL outputs, the ME research focus has gradually shifted from manual ME examinations to automatic ME detection and associated symptoms. In this review, we present a comprehensive and dynamic overview of AI in ME and identify future research areas.

Association between Aqueous Humor Cytokines and Structural Characteristics Based on Optical Coherence Tomography in Patients with Diabetic Macular Edema

Purpose

To investigate the relationship between aqueous humor cytokines and structural characteristics based on optical coherence tomography (OCT) in patients with diabetic macular edema (DME).

Methods

Forty eyes of 28 patients with DME diagnosed in the Affiliated Eye Hospital of Wenzhou Medical University at Hangzhou were included. All patients collected aqueous humor during anti-VEGF treatment, and the IL-6, IL-8, IL-10, VEGF, VCAM-1, ICAM-1, TGF-β1, FGF, and MCP-1 concentrations were detected. OCT examination was performed before anti-VEGF treatment and 1 month after anti-VEGF operation. Central macular thickness (CMT), macular volume (MV), choroidal thickness (CT), and the number of hyperreflective foci (HRF) were obtained for analysis. Each eye was determined whether there is subretinal effusion (SRD), cystoid macular edema (CME), and diffuse retinal thickening (DRT).

Results

The levels of IL-6 and FGF in DME patients with SRD were significantly higher than those without SRD (all P < 0.05). The level of VEGF in DME patients with CME was significantly higher than that in DME patients without CME (P = 0.005); IL-6, TGF-β1, and MCP-1 were significantly higher in DME patients with DRT than that without DRT (all P < 0.05). There was no significant correlation between aqueous humor cytokines and retinal thickness and retinal volume. However, the thinner the CT, the higher the level of aqueous humor cytokines IL-6 (r = -0.313, P = 0.049) and FGF (r = -0.361, P = 0.022). A multivariate linear regression analysis showed that IL-6 was significantly correlated with CT (P = 0.002) and SRD (P = 0.017), FGF was also significantly correlated with CT (P = 0.002) and SRD (P = 0.005), and TGF-β1 was correlated with triglycerides (P = 0.030) and HRF (P = 0.021).

Conclusion

DME patients with significant macular cystoid edema changes may be related to high VEGF concentrations and thin CT; meanwhile, the presence of SRD or a high number of HRF on OCT macular scans in DME patients may indicate high levels of intraocular inflammatory factors. Thus, OCT morphology characteristics to some extent reflect intraocular inflammatory factors and VEGF levels and may guide treatment alternatives.

The Classification of Common Macular Diseases Using Deep Learning on Optical Coherence Tomography Images with and without Prior Automated Segmentation

We compared the performance of deep learning (DL) in the classification of optical coherence tomography (OCT) images of macular diseases between automated classification alone and in combination with automated segmentation. OCT images were collected from patients with neovascular age-related macular degeneration, polypoidal choroidal vasculopathy, diabetic macular edema, retinal vein occlusion, cystoid macular edema in Irvine-Gass syndrome, and other macular diseases, along with the normal fellow eyes. A total of 14,327 OCT images were used to train DL models. Three experiments were conducted: classification alone (CA), use of automated segmentation of the OCT images by RelayNet, and the graph-cut technique before the classification (combination method 1 (CM1) and 2 (CM2), respectively). For validation of classification of the macular diseases, the sensitivity, specificity, and accuracy of CA were found at 62.55%, 95.16%, and 93.14%, respectively, whereas the sensitivity, specificity, and accuracy of CM1 were found at 72.90%, 96.20%, and 93.92%, respectively, and of CM2 at 71.36%, 96.42%, and 94.80%, respectively. The accuracy of CM2 was statistically higher than that of CA (p = 0.05878). All three methods achieved AUC at 97%. Applying DL for segmentation of OCT images prior to classification of the images by another DL model may improve the performance of the classification.

Deep learning algorithms for detection of diabetic macular edema in OCT images: A systematic review and meta-analysis

PURPOSE

Artificial intelligence (AI) can detect diabetic macular edema (DME) from optical coherence tomography (OCT) images. We aimed to evaluate the performance of deep learning neural networks in DME detection.

METHODS

Embase, Pubmed, the Cochrane Library, and IEEE Xplore were searched up to August 14, 2021. We included studies using deep learning algorithms to detect DME from OCT images. Two reviewers extracted the data independently, and the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool was applied to assess the risk of bias. The study is reported according to Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies (PRISMA-DTA).

RESULTS

Ninteen studies involving 41005 subjects were included. The pooled sensitivity and specificity were 96.0% (95% confidence interval (CI): 93.9% to 97.3%) and 99.3% (95% CI: 98.2% to 99.7%), respectively. Subgroup analyses found that data set selection, sample size of training set and the choice of OCT devices contributed to the heterogeneity (all P < 0.05). While there was no association between the diagnostic accuracy and transfer learning adoption or image management (all P > 0.05).

CONCLUSIONS

Deep learning methods, particularly the convolutional neural networks (CNNs) could effectively detect clinically significant DME, which can provide referral suggestions to the patients.

Generative adversarial network-based deep learning approach in classification of retinal conditions with optical coherence tomography images

PURPOSE

To determine whether a deep learning approach using generative adversarial networks (GANs) is beneficial for the classification of retinal conditions with Optical coherence tomography (OCT) images.

METHODS

Our study utilized 84,452 retinal OCT images obtained from a publicly available dataset (Kermany Dataset). Employing GAN, synthetic OCT images are produced to balance classes of retinal disorders. A deep learning classification model is constructed using pretrained deep neural networks (DNNs), and outcomes are evaluated using 2082 images collected from patients who visited the Department of Ophthalmology and the Department of Endocrinology and Metabolism at the Tri-service General Hospital in Taipei from January 2017 to December 2021.

RESULTS

The highest classification accuracies accomplished by deep learning machines trained on the unbalanced dataset for its training set, validation set, fivefold cross validation (CV), Kermany test set, and TSGH test set were 97.73%, 96.51%, 97.14%, 99.59%, and 81.03%, respectively. The highest classification accuracies accomplished by deep learning machines trained on the synthesis-balanced dataset for its training set, validation set, fivefold CV, Kermany test set, and TSGH test set were 98.60%, 98.41%, 98.52%, 99.38%, and 84.92%, respectively. In comparing the highest accuracies, deep learning machines trained on the synthesis-balanced dataset outperformed deep learning machines trained on the unbalanced dataset for the training set, validation set, fivefold CV, and TSGH test set.

CONCLUSIONS

Overall, deep learning machines on a synthesis-balanced dataset demonstrated to be advantageous over deep learning machines trained on an unbalanced dataset for the classification of retinal conditions.

Application of artificial intelligence-based dual-modality analysis combining fundus photography and optical coherence tomography in diabetic retinopathy screening in a community hospital

Background

To assess the feasibility and clinical utility of artificial intelligence (AI)-based screening for diabetic retinopathy (DR) and macular edema (ME) by combining fundus photos and optical coherence tomography (OCT) images in a community hospital.

Methods

Fundus photos and OCT images were taken for 600 diabetic patients in a community hospital. Ophthalmologists graded these fundus photos according to the International Clinical Diabetic Retinopathy (ICDR) Severity Scale as the ground truth. Two existing trained AI models were used to automatically classify the fundus images into DR grades according to ICDR, and to detect concomitant ME from OCT images, respectively. The criteria for referral were DR grades 2–4 and/or the presence of ME. The sensitivity and specificity of AI grading were evaluated. The number of referable DR cases confirmed by ophthalmologists and AI was calculated, respectively.

Results

DR was detected in 81 (13.5%) participants by ophthalmologists and in 94 (15.6%) by AI, and 45 (7.5%) and 53 (8.8%) participants were diagnosed with referable DR by ophthalmologists and by AI, respectively. The sensitivity, specificity and area under the curve (AUC) of AI for detecting DR were 91.67%, 96.92% and 0.944, respectively. For detecting referable DR, the sensitivity, specificity and AUC of AI were 97.78%, 98.38% and 0.981, respectively. ME was detected from OCT images in 49 (8.2%) participants by ophthalmologists and in 57 (9.5%) by AI, and the sensitivity, specificity and AUC of AI were 91.30%, 97.46% and 0.944, respectively. When combining fundus photos and OCT images, the number of referrals identified by ophthalmologists increased from 45 to 75 and from 53 to 85 by AI.

Conclusion

AI-based DR screening has high sensitivity and specificity and may feasibly improve the referral rate of community DR.

Optical Coherence Tomography Classification Systems for Diabetic Macular Edema and Their Associations With Visual Outcome and Treatment Responses - An Updated Review

ABSTRACT

Optical coherence tomography (OCT) is an invaluable imaging tool in detecting and assessing diabetic macular edema (DME). Over the past decade, there have been different proposed OCT-based classification systems for DME. In this review, we present an update of spectral-domain OCT (SDOCT)-based DME classifications over the past 5 years. In addition, we attempt to summarize the proposed OCT qualitative and quantitative parameters from different classification systems in relation to disease severity, risk of progression, and treatment outcome. Although some OCT-based measurements were found to have prognostic value on visual outcome, there has been a lack of consensus or guidelines on which parameters can be reliably used to predict treatment outcomes. We also summarize recent literatures on the prognostic value of these parameters including quantitative measures such as macular thickness or volume, central subfield thickness or foveal thickness, and qualitative features such as the morphology of the vitreoretinal interface, disorganization of retinal inner layers, ellipsoid zone disruption integrity, and hyperreflec-tive foci. In addition, we discuss that a framework to assess the validity of biomarkers for treatment outcome is essentially important in assessing the prognosis before deciding on treatment in DME. Finally, we echo with other experts on the demand for updating the current diabetic retinal disease classification.

Development and Validation of a Deep Learning Model to Predict the Occurrence and Severity of Retinopathy of Prematurity

IMPORTANCE

Retinopathy of prematurity (ROP) is the leading cause of childhood blindness worldwide. Prediction of ROP before onset holds great promise for reducing the risk of blindness.

OBJECTIVE

To develop and validate a deep learning (DL) system to predict the occurrence and severity of ROP before 45 weeks' postmenstrual age.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective prognostic study included 7033 retinal photographs of 725 infants in the training set and 763 retinal photographs of 90 infants in the external validation set, along with 46 characteristics for each infant. All images of both eyes from the same infant taken at the first screening were labeled according to the final diagnosis made between the first screening and 45 weeks' postmenstrual age. The DL system was developed using retinal photographs from the first ROP screening and clinical characteristics before or at the first screening in infants born between June 3, 2017, and August 28, 2019.

EXPOSURES

Two models were specifically designed for predictions of the occurrence (occurrence network [OC-Net]) and severity (severity network [SE-Net]) of ROP. Five-fold cross-validation was applied for internal validation.

MAIN OUTCOMES AND MEASURES

Area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity to evaluate the performance in ROP prediction.

RESULTS

This study included 815 infants (450 [55.2%] boys) with mean birth weight of 1.91 kg (95% CI, 1.87-1.95 kg) and mean gestational age of 33.1 weeks (95% CI, 32.9-33.3 weeks). In internal validation, mean AUC, accuracy, sensitivity, and specificity were 0.90 (95% CI, 0.88-0.92), 52.8% (95% CI, 49.2%-56.4%), 100% (95% CI, 97.4%-100%), and 37.8% (95% CI, 33.7%-42.1%), respectively, for OC-Net to predict ROP occurrence and 0.87 (95% CI, 0.82-0.91), 68.0% (95% CI, 61.2%-74.8%), 100% (95% CI, 93.2%-100%), and 46.6% (95% CI, 37.3%-56.0%), respectively, for SE-Net to predict severe ROP. In external validation, the AUC, accuracy, sensitivity, and specificity were 0.94, 33.3%, 100%, and 7.5%, respectively, for OC-Net, and 0.88, 56.0%, 100%, and 35.3%, respectively, for SE-Net.

CONCLUSIONS AND RELEVANCE

In this study, the DL system achieved promising accuracy in ROP prediction. This DL system is potentially useful in identifying infants with high risk of developing ROP.

SPECTRAL DOMAIN OPTICAL COHERENCE TOMOGRAPHY BIOMARKERS OF RETINAL HYPERPERMEABILITY AND CHOROIDAL INFLAMMATION AS PREDICTORS OF SHORT-TERM FUNCTIONAL AND ANATOMICAL OUTCOMES IN EYES WITH DIABETIC MACULAR EDEMA TREATED WITH INTRAVITREAL BEVACIZUMAB

PURPOSE

To assess spectral domain optical coherence tomography biomarkers of short-term outcomes in eyes with diabetic macular edema treated with intravitreal bevacizumab.

METHODS

In a prospective interventional case series, 66 eyes with diabetic macular edema underwent 3 monthly intravitreal bevacizumab injections. Best-corrected visual acuity measurement and spectral domain optical coherence tomography were performed at baseline and at 3 months. Multivariate regression analysis was performed to investigate the baseline spectral domain optical coherence tomography parameters as predictors of functional and anatomical outcomes.

RESULTS

Patients with diabetic nephropathy had greater subfoveal choroidal thickness (300.8 ± 35.54 vs. 253.0 ± 50.07 µm, P < 0.01) and were more likely to have subretinal fluid (r = 0.26, P = 0.03) at baseline. Multivariate analysis showed that the extent of external limiting membrane disruption (P = 0.03) and the extent of disorganization of retinal inner layers (P = 0.03) at baseline were predictors of best-corrected visual acuity at 3 months, whereas the extent of disorganization of retinal inner layers (P = 0.04) and duration of diabetes mellitus (P = 0.03) were predictors of central subfield thickness at 3 months.

CONCLUSION

External limiting membrane disruption and disorganization of retinal inner layers, as the spectral domain optical coherence tomography biomarkers of retinal hyperpermeability, can predict short-term outcomes in diabetic macular edema eyes treated with intravitreal bevacizumab.

Factors Associated With the Presence of Foveal Bulge in Eyes With Resolved Diabetic Macular Edema

Purpose: To evaluate factors associated with the presence of foveal bulge (FB) in resolved diabetic macular edema (DME) eyes.

Methods: A total of 165 eyes with complete integrity of ellipsoid zone (EZ) at the fovea and resolved DME were divided into two groups according to the presence of FB at 6 months after intravitreal injection of ranibizumab treatment. Best-corrected visual acuity (BCVA), central foveal thickness (CFT), outer nuclear layer (ONL) thickness, height of serous retinal detachment (SRD) and non-SRD, and inner segment (IS) and outer segment (OS) lengths of the two groups were measured and compared at baseline and each follow-up. The correlations between the presence of FB and pre- and post-treatment factors were determined by logistic regression analysis.

Results: At baseline, BCVA was significantly better, and CFT and incidence and height of SRD were significantly lower in the FB (+) group (all P < 0.05). At 6 months, FB was present in 65 (39.39%) eyes. Post-treatment BCVA was significantly better and OS length was significantly longer in the FB (+) group at 6 months (all P < 0.05). Multivariate analysis identified younger age, better BCVA, and lower CFT before treatment as significant predictors of the existence of FB at 6 months (all P < 0.05). At 6 months, better BCVA and longer OS length were significantly correlated with the existence of FB (all P < 0.05).

Conclusions: Factors associated with the presence of FB after the resolution of DME include younger age, better baseline BCVA and lower baseline CFT, and better post-treatment BCVA and longer post-treatment OS length.

Artificial intelligence for diabetic retinopathy

ABSTRACT

Diabetic retinopathy (DR) is an important cause of blindness globally, and its prevalence is increasing. Early detection and intervention can help change the outcomes of the disease. The rapid development of artificial intelligence (AI) in recent years has led to new possibilities for the screening and diagnosis of DR. An AI-based diagnostic system for the detection of DR has significant advantages, such as high efficiency, high accuracy, and lower demand for human resources. At the same time, there are shortcomings, such as the lack of standards for development and evaluation and the limited scope of application. This article demonstrates the current applications of AI in the field of DR, existing problems, and possible future development directions.

Artificial intelligence for improving sickle cell retinopathy diagnosis and management

Sickle cell retinopathy is often initially asymptomatic even in proliferative stages, but can progress to cause vision loss due to vitreous haemorrhages or tractional retinal detachments. Challenges with access and adherence to screening dilated fundus examinations, particularly in medically underserved areas where the burden of sickle cell disease is highest, highlight the need for novel approaches to screening for patients with vision-threatening sickle cell retinopathy. This article reviews the existing literature on and suggests future research directions for coupling artificial intelligence with multimodal retinal imaging to expand access to automated, accurate, imaging-based screening for sickle cell retinopathy. Given the variability in retinal specialist practice patterns with regards to monitoring and treatment of sickle cell retinopathy, we also discuss recent progress toward development of machine learning models that can quantitatively track disease progression over time. These artificial intelligence-based applications have great potential for informing evidence-based and resource-efficient clinical diagnosis and management of sickle cell retinopathy.

Machine learning-based prediction of anatomical outcome after idiopathic macular hole surgery

Background

To develop a machine learning (ML) model for the prediction of the idiopathic macular hole (IMH) status at 1 month after vitrectomy and internal limiting membrane peeling (VILMP) surgery.

Methods

A total of 288 IMH eyes from four ophthalmic centers were enrolled. All eyes underwent optical coherence tomography (OCT) examinations upon admission and one month after VILMP. First, 1,792 preoperative macular OCT parameters and 768 clinical variables of 256 eyes from two ophthalmic centers were used to train and internally validate ML models. Second, 224 preoperative macular OCT parameters and 96 clinical variables of 32 eyes from the other two centers were utilized for external validation. To fulfill the purpose of predicting postoperative IMH status (i.e., closed or open), five ML algorithms were trained and internally validated by the ten-fold cross-validation method, while the best-performing algorithm was further tested by an external validation set.

Results

In the internal validation, the mean area under the receiver operating characteristic curves (AUCs) of the five ML algorithms were 0.882-0.951. The AUC, accuracy, sensitivity, and specificity of the best-performing algorithm (i.e., random forest, RF) were 0.951, 0.892, 0.973, and 0.904, respectively. In the external validation, the AUC of RF was 0.940, with an accuracy of 0.875, a specificity of 0.875, and a sensitivity of 0.958.

Conclusions

Based on the preoperative OCT parameters and clinical variables, our ML model achieved remarkable accuracy in predicting IMH status after VILMP. Therefore, ML models may help optimize surgical planning for IMH patients in the future.