Diagnosis Of Diabetic Retinopathy Research Articles

A Review on Diabetic Retinopathy.

Diabetic retinopathy is a well-recognised microvascular complication of diabetes and is among the leading cause of blindness all over the world. Over the last decade, there have been advances in the diagnosis of diabetic retinopathy and diabetic macular edema. At the same time, newer therapies for the management of diabetic retinopathy have evolved. As a result of these advances, a decline in severe vision loss due to diabetes has been witnessed in some developing countries. However, there is a steady increase in the number of people affected with diabetes, and is expected to rise further in the coming years. Therefore, it is prudent to identify diabetic retinopathy, and timely intervention is needed to decrease the burden of severe vision loss. An effort has been made to review all the existing knowledge regarding diabetic retinopathy in this article and summarize the present treatment options for diabetic retinopathy.

Multimodal image fusion for the detection of diabetic retinopathy using optimized explainable AI-based Light GBM classifier

Diabetic Retinopathy (DR) is a widespread ocular condition and a significant contributor to global blindness. Timely identification and precise diagnosis of DR are essential for successfully managing and avoiding vision impairment. Visualizing and studying the intricate vascular network and other retinal structures has notable difficulties due to several factors that complicate the process. To overcome this, the proposed Dunnock-Scheduler optimization-based Light GBM (DkSO-Light GBM) is introduced for multimodal image fusion for DR detection. This approach can assist clinicians in making informed decisions, identifying essential features, and ensuring transparency in the automated DR diagnosis process. In this research, the Optical Coherence Tomography Angiography (OCTA) images undergo feature map generation using ResNet 101, and the DkSO algorithm is used by the AI-based Light Gradient Boosting Machine (GBM) to classify the normal and abnormal retina. The DkSO algorithm relies on the searching and scheduling characteristics, which enhance the model to identify DR more accurately by fusion of OCTA and fundus images. The experimental outcomes illustrate that the accuracy, sensitivity, specificity, precision, F1 score, balanced accuracy, and Mathews Correlation Coefficient (MCC) of the DkSO-Light GBM are 94.32 %, 94.94 %, 94.78 %, 94.78 %, 94.25 %, 94.86 %, and 91.77 % respectively, at a Training percentage (TP) of 90. In terms of k-fold 6, the metrics stand at 95.53 %, 94.72 %, 95.41 %, 94.16 %, 93.83 %, 95.07 %, and 92.00 %, respectively, signifying the superior efficiency of the DkSO-Light GBM model compared to other conventional techniques.

An Effective Heuristic Optimizer with Deep Learning-assisted Diabetic Retinopathy Diagnosis on Retinal Fundus Images

An Effective Heuristic Optimizer with Deep Learning-assisted Diabetic Retinopathy Diagnosis on Retinal Fundus Images

Diabetic Retinopathy-A Review.

Diabetic Retinopathy is a vascular microvascular disease also called diabetic eye disease caused by microangiopathy leading to progressive damage of the retina and blindness. The uncontrolled blood glycemic level or sugar level results in diabetic retinopathy. There are two stages of diabetic retinopathy: proliferative diabetic retinopathy and nonproliferative diabetic retinopathy. Symptoms of diabetic retinopathy often have no early warning signs, even muscular edema, which can cause rapid vision loss. Macular edema in which the blood vessels leak can also occur at any stage of diabetic retinopathy. Symptoms are darkened or distorted images and blurred vision that are not the same in both eyes. This review study primarily discusses the pathophysiology, genetics, and ALR, AGEs, VEGF, EPO, and eNOS involved in diabetic retinopathy. The longer a person has diabetes, the higher their risk of developing some ocular problems. During pregnancy, diabetic retinopathy may also be a problem for women with diabetes. NIH are recommends that all pregnant women with diabetes have an overall eye examination. Diagnosis of diabetic retinopathy is made during an eye examination that comprises ophthalmoscopy or fundus photography, and glow-in angiography for Fundus. Here, we present a review of the current insights into pathophysiology in diabetic retinopathy, as well as clinical treatments for diabetic retinopathy patients. Novel laboratory findings and related clinical trials are also analysed.

How Generalizable Are Findings from a Community-Based Prospective Cohort Study? Extending Estimates from the Adult Changes in Thought Study to Its Source Population

Background: The Adult Changes in Thought (ACT) study is a cohort of Kaiser Permanente Washington members ages 65+ that began in 1994. Objective: We wanted to know how well ACT participants represented all older adults in the region, and how well ACT findings on eye disease and its relationship with Alzheimer’s disease generalized to all older adults in the Seattle Metropolitan Region. Methods: We used participation weights derived from pooling ACT and Behavioral Risk Factor Surveillance System (BRFSS) data to estimate prevalences of common eye diseases and their associations with Alzheimer’s disease incidence. Cox proportional hazards models accounted for age, education, smoking, sex, and APOE genotype. Confidence intervals for weighted analyses were bootstrapped to account for error in estimating the weights. Results: ACT participants were fairly similar to older adults in the region. The largest differences were more self-reported current cholesterol medication use in BRFSS and higher proportions with low education in ACT. Incorporating the weights had little impact on prevalence estimates for age-related macular degeneration or glaucoma. Weighted estimates were slightly higher for diabetic retinopathy (weighted 5.7% (95% Confidence Interval 4.3, 7.1); unweighted 4.1% (3.6, 4.6)) and cataract history (weighted 51.8% (49.6, 54.3); unweighted 48.6% (47.3, 49.9)). The weighted hazard ratio for recent diabetic retinopathy diagnosis and Alzheimer’s disease was 1.84 (0.34, 4.29), versus 1.32 (0.87, 2.00) in unweighted ACT. Conclusions: Most, but not all, associations were similar after participation weighting. Even in community-based cohorts, extending inferences to broader populations may benefit from evaluation with participation weights.

Deep Transfer Learning-Based Automated Diabetic Retinopathy Detection Using Retinal Fundus Images in Remote Areas

Diabetic retinopathy (DR) significantly burdens ophthalmic healthcare due to its wide prevalence and high diagnostic costs. Especially in remote areas with limited medical access, undetected DR cases are on the rise. Our study introduces an advanced deep transfer learning-based system for real-time DR detection using fundus cameras to address this. This research aims to develop an efficient and timely assistance system for DR patients, empowering them to manage their health better. The proposed system leverages fundus imaging to collect retinal images, which are then transmitted to the processing unit for effective disease severity detection and classification. Comprehensive reports guide subsequent medical actions based on the identified stage. The proposed system achieves real-time DR detection by utilizing deep transfer learning algorithms, specifically VGGNet. The system’s performance is rigorously evaluated, comparing its classification accuracy to previous research outcomes. The experimental results demonstrate the robustness of the proposed system, achieving an impressive 97.6% classification accuracy during the detection phase, surpassing the performance of existing approaches. Implementing the automated system in remote areas has transformed healthcare dynamics, enabling early, cost-effective DR diagnosis for millions. The system also streamlines patient prioritization, facilitating timely interventions for early-stage DR cases.

DDLA: a double deep latent autoencoder for diabetic retinopathy diagnose based on continuous glucose sensors.

The current diagnosis of diabetic retinopathy is based on fundus images and clinical experience. However, considering the ineffectiveness and non-portability of medical devices, we aimed to develop a diagnostic model for diabetic retinopathy based on glucose series data from the wearable continuous glucose monitoring system. Therefore, this study developed a novel method, i.e., double deep latent autoencoder, for exploring glycemic variability influence from multi-day glucose data for diabetic retinopathy. Specifically, the model proposed in this research could encode continuous glucose sensor data with non-continuous and variable length via the integration of a data reorganization module and a novel encoding module with fragmented-missing-wise objective function. Additionally, the model implements a double deep autoencoder, which integrated convolutional neural network, long short-term memory, to jointly capturing the inter-day and intra-day glucose latent features from glucose series. The effectiveness of the proposed model is evaluated through a cross-validation method to clinical datasets of 765 type 2 diabetes patients. The proposed method achieves the highest accuracy value (0.89), precision value (0.88), and F1 score (0.73). The results suggest that our model can be used to remotely diagnose and screen for diabetic retinopathy by learning potential features of glucose series data collected by wearable continuous glucose monitoring systems.

Evaluation of systemic medications associated with diabetic retinopathy: a nested case–control study from the UK Biobank

AimsThis study aims to investigate the associations between commonly used systemic medications and diabetic retinopathy (DR).MethodsIndividuals with linked primary care prescription data from the UK Biobank were included. Cases were...

MediDRNet: Tackling category imbalance in diabetic retinopathy classification with dual-branch learning and prototypical contrastive learning

Background and Objective:The classification of diabetic retinopathy (DR) aims to utilize the implicit information in images for early diagnosis, to prevent and mitigate the further worsening of the condition. However, existing methods are often limited by the need to operate within large, annotated datasets to show significant advantages. Additionally, the number of samples for different categories within the dataset needs to be evenly distributed, because the characteristic of sample imbalance distribution can lead to an excessive focus on high-frequency disease categories, while neglecting the less common but equally important disease categories. Therefore, there is an urgent need to develop a new classification method that can effectively alleviate the issue of sample distribution imbalance, thereby enhancing the accuracy of diabetic retinopathy classification. Methods:In this work, we propose MediDRNet, a dual-branch network model based on prototypical contrastive learning. This model adopts prototype contrastive learning, creating prototypes for different levels of lesions, ensuring they represent the core features of each lesion level. It classifies by comparing the similarity between data points and their category prototypes. Our dual-branch network structure effectively resolves the issue of category imbalance and improves classification accuracy by emphasizing subtle differences in retinal lesions. Moreover, our approach combines a dual-branch network with specific lesion-level prototypes for core feature representation and incorporates the convolutional block attention module for enhanced lesion feature identification. Results:Our experiments using both the Kaggle and UWF classification datasets have demonstrated that MediDRNet exhibits exceptional performance compared to other advanced models in the industry, especially on the UWF DR classification dataset where it achieved state-of-the-art performance across all metrics. On the Kaggle DR classification dataset, it achieved the highest average classification accuracy (0.6327) and Macro-F1 score (0.6361). Particularly in the classification tasks for minority categories of diabetic retinopathy on the Kaggle dataset (Grades 1, 2, 3, and 4), the model reached high classification accuracies of 58.08%, 55.32%, 69.73%, and 90.21%, respectively. In the ablation study, the MediDRNet model proved to be more effective in feature extraction from diabetic retinal fundus images compared to other feature extraction methods. Conclusions:This study employed prototype contrastive learning and bidirectional branch learning strategies, successfully constructing a grading system for diabetic retinopathy lesions within imbalanced diabetic retinopathy datasets. Through a dual-branch network, the feature learning branch effectively facilitated a smooth transition of features from the grading network to the classification learning branch, accurately identifying minority sample categories. This method not only effectively resolved the issue of sample imbalance but also provided strong support for the precise grading and early diagnosis of diabetic retinopathy in clinical applications, showcasing exceptional performance in handling complex diabetic retinopathy datasets. Moreover, this research significantly improved the efficiency of prevention and management of disease progression in diabetic retinopathy patients within medical practice. We encourage the use and modification of our code, which is publicly accessible on GitHub: https://github.com/ReinforceLove/MediDRNet.

Cross-modality transfer learning with knowledge infusion for diabetic retinopathy grading.

Ultra-wide-field (UWF) fundus photography represents an emerging retinal imaging technique offering a broader field of view, thus enhancing its utility in screening and diagnosing various eye diseases, notably diabetic retinopathy (DR). However, the application of computer-aided diagnosis for DR using UWF images confronts two major challenges. The first challenge arises from the limited availability of labeled UWF data, making it daunting to train diagnostic models due to the high cost associated with manual annotation of medical images. Secondly, existing models' performance requires enhancement due to the absence of prior knowledge to guide the learning process. By leveraging extensively annotated datasets within the field, which encompass large-scale, high-quality color fundus image datasets annotated at either image-level or pixel-level, our objective is to transfer knowledge from these datasets to our target domain through unsupervised domain adaptation. Our approach presents a robust model for assessing the severity of diabetic retinopathy (DR) by leveraging unsupervised lesion-aware domain adaptation in ultra-wide-field (UWF) images. Furthermore, to harness the wealth of detailed annotations in publicly available color fundus image datasets, we integrate an adversarial lesion map generator. This generator supplements the grading model by incorporating auxiliary lesion information, drawing inspiration from the clinical methodology of evaluating DR severity by identifying and quantifying associated lesions. We conducted both quantitative and qualitative evaluations of our proposed method. In particular, among the six representative DR grading methods, our approach achieved an accuracy (ACC) of 68.18% and a precision (pre) of 67.43%. Additionally, we conducted extensive experiments in ablation studies to validate the effectiveness of each component of our proposed method. In conclusion, our method not only improves the accuracy of DR grading, but also enhances the interpretability of the results, providing clinicians with a reliable DR grading scheme.

Association between HDL cholesterol with diabetic retinopathy in diabetic patients: a cross-sectional retrospective study

ObjectiveDiabetic patients are often comorbid with dyslipidemia, however, the relationship between high-density lipoprotein cholesterol(HDL-C) and diabetic retinopathy (DR) in the adult diabetic population remains to be fully elucidated.The aim of this study is to evaluate the associations between HDL-C and DR in the United States adults with diabetes.MethodsA total of 1708 participants from the National Health and Nutrition Examination Survey (NHANES) 2005–2008 were enrolled in the present study. Fundus images of all study subjects were captured and evaluated using a digital camera and an ophthalmic digital imaging system, and the diagnosis of DR was made by the severity scale of the Early Treatment Diabetic Retinopathy Study (ETDRS).Roche Diagnostics were used to measure serum HDL-C concentration. The relationship of DR with HDL-C was investigated using multivariable logistic regression. The potential non-line correlation was explored with smooth curve fitting approach.ResultsThe fully-adjusted model showed that HDL-C positively correlated with DR(OR:1.69, 95%CI: 1.25–2.31).However, an inverted U-shaped association between them was observed by applying the smooth curve fitted method. The inflection point of HDL-C(1.99mmol/l) was calculated by utilizing the two-piecewise logistic regression model. In the subgroup analysis, the inverted U-shaped nonlinear correlation between HDL-C and DR was also found in female, Non-Hispanic White, and lower age groups.ConclusionOur study revealed an inverted U-shaped positive relationship between HDL-C and DR.The findings may provide us with a more comprehensive understanding of the association between HDL-C and DR.

Deep learning-driven automated quality assessment of ultra-widefield optical coherence tomography angiography images for diabetic retinopathy

AbstractImage quality assessment (IQA) of fundus images constitutes a foundational step in automated disease analysis. This process is pivotal in supporting the automation of screening, diagnosis, follow-up, and related academic research for diabetic retinopathy (DR). This study introduced a deep learning-based approach for IQA of ultra-widefield optical coherence tomography angiography (UW-OCTA) images of patients with DR. Given the novelty of ultra-widefield technology, its limited prevalence, the high costs associated with equipment and operational training, and concerns regarding ethics and patient privacy, UW-OCTA datasets are notably scarce. To address this, we initially pre-train a vision transformer (ViT) model on a dataset comprising 6 mm × 6 mm OCTA images, enabling the model to acquire a fundamental understanding of OCTA image characteristics and quality indicators. Subsequent fine-tuning on 12 mm × 12 mm UW-OCTA images aims to enhance accuracy in quality assessment. This transfer learning strategy leverages the generic features learned during pre-training and adjusts the model to evaluate UW-OCTA image quality effectively. Experimental results demonstrate that our proposed method achieves superior performance compared to ResNet18, ResNet34, and ResNet50, with an AUC of 0.9026 and a Kappa value of 0.7310. Additionally, ablation studies, including the omission of pre-training on 6 mm × 6 mm OCTA images and the substitution of the backbone network with the ViT base version, resulted in varying degrees of decline in AUC and Kappa values, confirming the efficacy of each module within our methodology.

Correction to: Advances in retinal microaneurysms detection, segmentation and datasets for the diagnosis of diabetic retinopathy: a systematic literature review

Correction to: Advances in retinal microaneurysms detection, segmentation and datasets for the diagnosis of diabetic retinopathy: a systematic literature review

Enhanced Diabetic Retinopathy Diagnosis: A Comparative Analysis of models with an Ensemble classifier and Deep Q Learning

- A novel method that combines the strengths of different classifiers such as Naive Bayes, Multi-Layer Perceptron (MLP), and Support Vector Machine (SVM) is introduced in this paper. This method tackles the urgent need for cutting-edge diagnostic methods in the field of ophthalmology, mainly for the identification of diabetic retinopathy (DR). The approach is ensemble-based. Classical methods in retinal analysis of images often fail as they are static and are unable to adjust to the unique details that each distinct image presents. This constraint results in less accurate and precise diagnostic results, highlighting the urgent need for more adaptable and dynamic methods. The suggested model differs significantly from previous methods. Through the use of an ensemble approach, it capitalizes on the distinct advantages of each classifier: the MLP process's sophisticated feature extraction skill, Naive Bayes' probabilistic analysis, and SVM's non-linear pattern recognition capacity. By combining these techniques, the inherent drawbacks of utilizing a single strategy are addressed, guaranteeing a more thorough examination of retinal samples and images. The core of this idea is the system using Deep Q Learning (DQL) for adaptive classifier selection. Using learned Q Values for various contexts, this reinforcement learning technique selects the best classifier adaptively for each unique retinal image, hence optimizing the ensemble. This approach not only advances diagnosis accuracy and precision but also guarantees ongoing learning and adaptation to keep up with changing data patterns and advances in imaging technology. Extensive experiments on the IDRiD & EyePACS Dataset show the effectiveness of this model with a 5.5% increase in overall accuracy with other performance metrics, the results show a significant improvement over the current method. They represent a significant advancement in the timely and accurate identification of diabetic retinopathy, which will ultimately benefit patients and lessen the strain on healthcare systems. Thus, this work represents a major step forward in patient care as well as a technological advance, opening the door to more efficient supervision and treatment of retinal illnesses.

Synchronous Diagnosis of Diabetic Retinopathy by a Handheld Retinal Camera, Artificial Intelligence, and Simultaneous Specialist Confirmation

PurposeDiabetic retinopathy (DR) is a leading cause of preventable blindness, particularly in underserved regions where access to ophthalmic care is limited. This study presents a proof of concept for utilizing a portable handheld retinal camera with an embedded artificial intelligence (AI) platform, complemented by a synchronous remote confirmation by retina specialists, for DR screening in an underserved rural area. DesignRetrospective cohort study Subjects1,115 individuals with diabetes. MethodsA retrospective analysis of a screening initiative conducted in four municipalities in Northeastern Brazil, targeting the diabetic population. A portable handheld retinal camera captured macula-centered and disc-centered images, which were analyzed by the AI system. Immediate push notifications were sent out to retina specialists upon the detection of significant abnormalities, enabling synchronous verification and confirmation, with on-site patient feedback within minutes. Referral criteria were established, and all referred patients underwent a complete ophthalmic work-up and subsequent treatment. Main outcome measuresProof-of-concept implementation success. ResultsOut of 2,052 invited individuals, 1,115 participated, with a mean age of 60.93 years and diabetes duration of 7.52 years, of whom 66.03% were women. The screening covered 2,222 eyes, revealing various retinal conditions. Referable eyes for DR were 11.84%, with an additional 13% for other conditions (diagnoses included various stages of DR, media opacity, nevus, drusen, enlarged cup-to-disc ratio, pigmentary changes, and other). AI performance for overall detection of referable cases (both DR and other conditions) was: sensitivity- 84.23% (95% CI- 82.63-85.84), specificity- 80.79% (95% CI- 79.05- 82.53). When we assessed whether AI matched any clinical diagnosis, be it referable or not, sensitivity was 85.67% (95% CI- 84.12-87.22), specificity was 98.86 (95% CI- 98.39-99.33), and AUC was 0.92 (95% CI- 0.91-0.94). ConclusionsThe integration of a portable device, AI analysis, and synchronous medical validation has the potential to play a crucial role in preventing blindness from DR, especially in socially unequal scenarios.

A multi-resolution convolutional attention network for efficient diabetic retinopathy classification

Diabetic retinopathy (DR) is a disorder commonly associated with individuals who have diabetes and is caused by excessive glucose levels in the blood, which induce retinal damage. Manual diagnosis of DR by medical experts is time-consuming; hence, a computer-aided diagnosis (CAD) solution for faster identification and improved efficiency is required. Diagnosis of DR is complex due to the prevalence of inter-class differences and small lesions, making it difficult for convolutional neural networks (CNNs) to identify discriminative regions effectively. The most critical challenge is locating complex patterns connected with visual differences and lesion locations. To tackle the aforementioned challenges, this work presents a novel multi-resolution convolutional attention network (MuR-CAN), which improves overall performance by emphasizing discriminative features. MuR-CAN includes a multi-dilation attention block (MDAB), which uses depth-wise convolution layers with various dilation rates to allow the model to prioritize relevant features and capture multi-scale spatial information. This method provides for simultaneously highlighting fine features and subtle visual patterns. Furthermore, the features extracted from the MuR-CAN are utilized to train the support vector machine (SVM) for DR classification. The integration of the SVM classifier with the proposed CNN architecture ensures improved accuracy. The experimental findings demonstrate that the proposed model outperforms the current state-of-the-art methods.

Increasing the reliability of diagnosis of diabetic retinopathy based on machine learning

Increasing the reliability of diagnosis of diabetic retinopathy based on machine learning

Telemedicine Screening for Eye Disorders in Federally Qualified Health Centers: Relationship to Vision-Targeted Health-Related Quality of Life.

Introduction: Federally Qualified Health Centers (FQHCs) play a crucial role as safety-net primary health care clinics in the United States, serving medically underserved areas and populations. However, eye services are rarely offered at FQHCs. We examined how telemedicine-generated ocular diagnoses impacted vision-targeted health-related quality of life at FQHCs in rural Alabama. Methods: We focused on patients who are at risk for glaucoma. Both visual function and retinal imaging were assessed. The telemedicine vision screening protocol performed by a remote ophthalmologist evaluated eyes for glaucoma, diabetic retinopathy, cataract, age-related macular degeneration, and a measurement of habitual visual acuity. The National Eye Institute Visual Function Questionnaire-9 (VFQ-9) was administered. Results: Using stepwise regression, the best-fitting model for predicting VFQ-9 scores incorporated visual acuity 20/40 or worse, a diabetic retinopathy diagnosis, and sociodemographic variables (gender, transportation, insurance type/status, and employment status). Conclusion: Vision-targeted, health-related quality of life in our FQHC settings was related to the visual acuity impairment and the diagnosis of diabetic retinopathy but was also influenced by a variety of sociodemographic factors.

Diabetic retinopathy prediction based on vision transformer and modified capsule network

Diabetic retinopathy is considered one of the most common diseases that can lead to blindness in the working age, and the chance of developing it increases as long as a person suffers from diabetes. Protecting the sight of the patient or decelerating the evolution of this disease depends on its early detection as well as identifying the exact levels of this pathology, which is done manually by ophthalmologists. This manual process is very consuming in terms of the time and experience of an expert ophthalmologist, which makes developing an automated method to aid in the diagnosis of diabetic retinopathy an essential and urgent need. In this paper, we aim to propose a new hybrid deep learning method based on a fine-tuning vision transformer and a modified capsule network for automatic diabetic retinopathy severity level prediction. The proposed approach consists of a new range of computer vision operations, including the power law transformation technique and the contrast-limiting adaptive histogram equalization technique in the preprocessing step. While the classification step builds up on a fine-tuning vision transformer, a modified capsule network, and a classification model combined with a classification model, The effectiveness of our approach was evaluated using four datasets, including the APTOS, Messidor-2, DDR, and EyePACS datasets, for the task of severity levels of diabetic retinopathy. We have attained excellent test accuracy scores on the four datasets, respectively: 88.18%, 87.78%, 80.36%, and 78.64%. Comparing our results with the state-of-the-art, we reached a better performance.

Diabetic Retinopathy Lesion Segmentation Method Based on Multi-Scale Attention and Lesion Perception

The early diagnosis of diabetic retinopathy (DR) can effectively prevent irreversible vision loss and assist ophthalmologists in providing timely and accurate treatment plans. However, the existing methods based on deep learning have a weak perception ability of different scale information in retinal fundus images, and the segmentation capability of subtle lesions is also insufficient. This paper aims to address these issues and proposes MLNet for DR lesion segmentation, which mainly consists of the Multi-Scale Attention Block (MSAB) and the Lesion Perception Block (LPB). The MSAB is designed to capture multi-scale lesion features in fundus images, while the LPB perceives subtle lesions in depth. In addition, a novel loss function with tailored lesion weight is designed to reduce the influence of imbalanced datasets on the algorithm. The performance comparison between MLNet and other state-of-the-art methods is carried out in the DDR dataset and DIARETDB1 dataset, and MLNet achieves the best results of 51.81% mAUPR, 49.85% mDice, and 37.19% mIoU in the DDR dataset, and 67.16% mAUPR and 61.82% mDice in the DIARETDB1 dataset. The generalization experiment of MLNet in the IDRiD dataset achieves 59.54% mAUPR, which is the best among other methods. The results show that MLNet has outstanding DR lesion segmentation ability.