Soft attention with Convolutional Neural Network for Grading Diabetic Retinopathy

Diabetic retinopathy (DR) is the most common complication of diabetes mellitus and one of the leading causes of preventable blindness in the adult working population. Visual loss can be prevented from the early stages of DR, when the treatments are effective. Therefore, early diagnosis is paramount. However, DR may be clinically asymptomatic until the advanced stage, when vision is already affected and treatment may become difficult. For this reason, diabetic patients should undergo regular eye examinations through screening programs. Traditionally, DR screening programs are run by trained specialists through visual inspection of the retinal images. However, this manual analysis is time consuming and expensive. With the increasing incidence of diabetes and the limited number of clinicians and sanitary resources, the early detection of DR becomes non-viable. For this reason, computed-aided diagnosis (CAD) systems are required to assist specialists for a fast, reliable diagnosis, allowing to reduce the workload and the associated costs. We hypothesize that the application of novel, automatic algorithms for fundus image analysis could contribute to the early diagnosis of DR. Consequently, the main objective of the present Doctoral Thesis is to study, design and develop novel methods based on the automatic analysis of fundus images to aid in the screening, diagnosis, and treatment of DR. In order to achieve the main goal, we built a private database and used five retinal public databases: DRIMDB, DIARETDB1, DRIVE, Messidor and Kaggle. The stages of fundus image processing covered in this Thesis are: retinal image quality assessment (RIQA), the location of the optic disc (OD) and the fovea, the segmentation of RLs and EXs, and the DR severity grading. RIQA was studied with two different approaches. The first approach was based on the combination of novel, global features. Results achieved 91.46% accuracy, 92.04% sensitivity, and 87.92% specificity using the private database. We developed a second approach aimed at RIQA based on deep learning. We achieved 95.29% accuracy with the private database and 99.48% accuracy with the DRIMDB database. The location of the OD and the fovea was performed using a combination of saliency maps. The proposed methods were evaluated over the private database and the public databases DRIVE, DIARETDB1 and Messidor. For the OD, we achieved 100% accuracy for all databases except Messidor (99.50%). As for the fovea location, we also reached 100% accuracy for all databases except Messidor (99.67%). The joint segmentation of RLs and EXs was accomplished by decomposing the fundus image into layers. Results were computed per pixel and per image. Using the private database, 88.34% per-image accuracy (ACCi) was reached for the RL detection and 95.41% ACCi for EX detection. An additional method was proposed for the segmentation of RLs based on superpixels. Evaluating this method with the private database, we obtained 84.45% ACCi. Results were validated using the DIARETDB1 database. Finally, we proposed a deep learning framework for the automatic DR severity grading. The method was based on a novel attention mechanism which performs a separate attention of the dark and the bright structures of the retina. The Kaggle DR detection dataset was used for development and validation. The International Clinical DR Scale was considered, which is made up of 5 DR severity levels. Classification results for all classes achieved 83.70% accuracy and a Quadratic Weighted Kappa of 0.78. The methods proposed in this Doctoral Thesis form a complete, automatic DR screening system, contributing to aid in the early detection of DR. In this way, diabetic patients could receive better attention for their ocular health avoiding vision loss. In addition, the workload of specialists could be relieved while healthcare costs are reduced.

Diabetic Retinopathy (DR) has become a leading cause of blindness among diabetic patients. Accurate and timely diagnosis of DR is critical to slowing disease progression. This research proposes a Hybrid Convolutional Neural Network (CNN)-based model, named Diabetic Retinopathy Detection Network (DRD-Net). The proposed DRD-Net designed to enhance diagnostic accuracy by addressing key challenges such as gradient vanishing and lesion scale variability in fundus images. Contrast-Limited Adaptive Histogram Equalization (CLAHE) was used to enhance contrast and highlight lesions in fundus images. To increase the diversity of training samples, the proposed framework employs geometric data augmentation techniques. DRD-Net incorporates the Swish activation function along with densely connected blocks to mitigate gradient vanishing and enhancing feature propagation within the network. Additionally, the model integrates two Inception blocks to facilitate multiscale feature extraction, which is essential for detecting small Regions of Interest (RoI) in fundus images. Experimental results demonstrate that DRD-Net achieves a precision of 84.4%, recall of 84.5%, F1-score of 84.1%, and accuracy of 85.1%, outperforming several state-of-the-art models on the IDRiD dataset. These results highlight DRD-Net’s potential as an effective solution for automated DR diagnosis, contributing to more efficient and accurate DR screening.

Diabetic-caused retinopathy is a disease that damages the blood vessels that are present in the part of retina of the eyes and might lead to permanent blindness in people who have diabetes. Ophthalmologists use the fundus images of the eyes, which are the patients identify this disease. However, manual detection of this abnormality is very time taking, cost-consuming, and has a higher probability of resulting in errors. Thus, to prevent this, a Deep Learning model is proposed to detect Diabetic caused retinopathy by using high-resolution fundus images which as a result decreases the chances of misdiagnosis. Initially, fundus images are pre-processed by using the Contrast limited adaptive histogram equalization (CLAHE) model and histogram-based segmentation model. In the next step, features from the pre-processed images are extracted. However, it becomes very difficult for the individual models to identify the behind features that are complicated and those can only classify Diabetic Retinopathy's various stages at low accuracy. Thus, to resolve this an ensemble of three deep Convolution Neural Network (CNN) models (Inceptionv3, Xception, and DenseNet121) is developed which will encode the rich features and improve the classification for different stages of DR like normal, mild, moderate, severe and PDR (Proliferative Diabetic Retinopathy). The results show that the proposed ensemble model detects all the Diabetic Retinopathy stages, not like the individual models and performs better comparatively. This improvement can thus reduce time, effort, and labor in healthcare services and will also increase the final decision accuracy.

Diabetic retinopathy (DR) is an eye disease which is caused by high blood sugar and high blood pressure and damage the blood vessels in the back (retina) of the eye. Diabetic retinopathy (DR) is not a reversible process and treatment only sustain vision. The number of ophthalmologists cannot meet the growing demands around the world. This study focuses on the automatic diagnosis of the disease through deep learning. Convolutional neural network (CNN) is more widely used as a deep learning method in medical image analysis. Multiclass image classification of images into non referable DR and referable DR has been done using proposed Convolutional neural network (CNN) model. For multiclass classification problem, the sensitivity, specificity and F2 score value for class 0 (no DR) are 81.75 %, 91.06 % and 81.80 % respectively. Whereas for class 1 (non-severe DR), sensitivity, specificity and F2 score values are 71.28 %, 78.52 % and 70.01 % respectively. Similarly for class 2 (sever DR), sensitivity, specificity and F2 score values are 73.03 %, 93.39 % and 75.08 % respectively. Bayesian optimization has been performed for tuning learning rate and gives optimal learning rate 0.000358 through the optimization process. The customized CNN is then trained using 0.000358 learning rate and then tested on test data The images in dataset have poor contrast and consists of impulse noises. Contrast limited adaptive histogram equalization (CLAHE) method is used to improve the contrast of the image followed by median filter to remove noise present in DR image.

The most common causes of blindness in adults worldwide, 2.6% of them are caused by diabetic retinopathy, which is a progressive disease caused by complications of diabetes mellitus. Early detection and prompt treatment help save eyesight. The artificial intelligence technology can provide objective and accurate screening results. Deep learning, especially the Convolutional Neural Network (CNN), which is part of artificial intelligence, has proven successful in solving image problems and is very well used in medical image analysis. CNN works well on large datasets, but it will affect network performance to overfitting in fewer datasets. So to solve the problem of small data, data augmentation techniques can be used. There are various kinds of data augmentation techniques. This study used the CNN method to classify diabetic retinopathy disease, compared several suitable data augmentation techniques for retinal fundus images, and used Contrast Limited Adaptive Histogram Equalization (CLAHE) for image enhancement. This study found that the augmented random zoom technique, together with CLAHE, provided the best accuracy of 98% with 96% sensitivity and 100% specificity.

Diabetic Mellitus is the most familiar disease around the globe. Long prevalence of diabetes causes several problems related to health. The most common issue is Diabetic Retinopathy (DR). Diabetic retinopathy is a situation in which the vessels inside the retina are vandalized, leaking harmful substances and fluids in the surrounding tissue resulting in hemorrhages, micro aneurysms in the eye and further into partial or complete vision loss. This disease if treated in the early stage can help to prevent vision loss, but since it takes time for diagnosis and there is a shortage of ophthalmologists' patients suffer vision loss even before diagnosis. Hence, early detection of DR may help in reducing the problem. Therefore, in this paper we investigate various approaches to understand the process of detecting Diabetic Retinopathy as accurately as possible and classifying them into different grades of treatable DR (NPDR) namely LO, L1 DR, L2 DR and Proliferate DR (PDR) using Deep Learning and Image Processing techniques also making some improvisations on the same to enhance the capability of other existing systems.

Diabetes is one of the most dangerous diseases that is facing by many people in the world. Diabetic Retinopathy (DR) is a retinal disease that is caused by the Diabetes Mellitus (DM). DM causes the lesions on the retinal layer thereby affecting the vision. If it is not detected at the initial stages, it might lead to complete blindness and also DR is an irreversible disease. Hence, early detection of DR is inevitable to avoid the vision loss of the patients. Expeditious detection can decrease the complications of the DR thereby vision of the patient is preserved. It is a laborious task for the ophthalmologists to diagnosis DR manually, since it takes a lot more time and cost-effective. Mis-diagnosis might happen if the ophthalmologists are not skillful and experienced in detecting DR from the fundus images. Over the past two decades, deep learning has shown a significant raise in the bio-medical image processing and its niche areas obtaining the best performance. The state-of-the-art Convolutional Neural Networks (CNN) models achieved notable performance in classifying DR, but the severity levels of DR are not analyzed. To address these challenges, we propose a CNN based model that is used to analyze the fundus oculi retinal images to locate the eyeball structure and observe the presence of DR. The proposed model's hyperparameters are regulated by the transfer learning techniques for mapping the label of the images. The dataset used for training and testing the model is taken from the Kaggle that contains the retinal images and its corresponding severity in a scale. Severity level of the images is classified into five different categories from a normal to DR presented eyeball. The accuracy of the proposed model is 94.92%, proving the confident detection and prediction of the DR from the retinal images.

This project proposes machine learning methods for detecting diabetic retinopathy by using CNN. The classification of retinal lesions from non-lesions using CNN classifiers is examined. Retinal retina images can be evaluated using machine learning-based medical image analysis. The image is pre-processed using the Contrast limited adaptive histogram equalization (CLAHE) filter. CLAHE is used to improve the visibility level of blurry images. The image is segmented using the FCM method to generate an appropriate threshold value for data segmentation. The method of feature extraction is implemented by RLC. The Gray Level Co – occurrence Matrix (GLCM) method is used for extracting statistical skin parameters in a given direction and distance. The use of convolutional neural network (CNN) algorithms has facilitated the early identification of diabetic retinopathy. This project provides excellent specificity and sensitivity for classifying images as with or without diabetic retinopathy. This project is implemented using MATLAB software.

ABSTRACTThe early detection of diabetic retinopathy (DR) illnesses improves diagnosis and lowers the risk of permanent blindness. As a result, screening for DR in fundus images is an important method used to diagnose diabetes and other eye diseases. However, detecting diseases manually requires a significant amount of time and work. Deep learning (DL) techniques have produced encouraging results in categorizing fundus images. Still, the multi‐class DR disease remains a difficult task. Thus, the proposed framework adopted a novel Criss‐Cross Attention‐Based Squeeze‐and‐Excitation Assisted Vision Transformer (CCA‐SE‐ViT) model to classify DR from fundus images. Initially, the defective region of the retina is segmented using a novel dilated depth‐wise separable convolutional U‐Net model (dDSC‐UNet). Then, using the segmented regions, the fundus images are classified into multiple classes as age‐related macular degeneration (AMD), DR, glaucoma, cataracts, myopia, hypertension, normal, other abnormalities, and DR cases are classified into no DR, mild, moderate, severe, and proliferative DR, respectively. The retinal fundus images are obtained from publicly available datasets like OIA‐ODIR and APTOS 2019. The proposed methodology for multi‐class categorization of retinal illnesses in the OIA‐ODIR dataset yielded 97.2% accuracy, 96.7% precision, 96.1% recall, 95.9% F1‐score, and 96.4% specificity. The APTOS dataset was used for multi‐class classification of DR illnesses, and the results were 99.68% accuracy, 99.08% precision, 99.31% recall, 99.19% F1‐score, and 99.26% specificity. The results demonstrated that the proposed method accurately identifies DR using retinal fundus images.

Multi-Level severity classification for diabetic retinopathy based on hybrid optimization enabled deep learning

There are countless number of people with diabetes around the world. diabetic retinopathy (DR), a major complication of diabetes, is a retinal disease that results in blindness if not treated in the earlier stages. Early detection of diabetic retinopathy in patients can help avoid further loss of eyesight. For early detection, it is important for diabetic patients to undergo medical tests after a frequent amount of times. Diagnosing DR using fundus images is a difficult and time-consuming task which requires professional expertise to find and identify if the DR traits are present in the eye. In our proposed system, we make use of artificial intelligence, convolutional neural networks precisely to make the process of detection and classification of diabetic retinopathy a lot simpler and fast. The proposed model makes use of AlexNet which is a convolutional neural network architecture which is trained based on fundus image database to accurately diagnose DR with minimum efforts. The database contains multiple cases of retinal hemorrhage, microaneurism, cotton wool spots, etc. The retinal image of the patient is fed to the system where initially the image is trimmed to match the input layer of AlexNet and then the neural network extracts features and the model gets trained in order to further classify the diabetic retinopathy patients. The last layer of the convolutional neural network is developed in such a way each segment of image is assigned a specific weight factor where higher the weight, more is the chances of getting severe DR.

Diabetic retinopathy, a condition resulting from prolonged high blood sugar levels that damage the retina, can cause vision impairment and, if untreated, lead to blindness. With advances in medical imaging and the availability of fundus image collections such as Madrid Messidor and DRIVE, computer-aided diagnosis (CAD) systems have become instrumental in identifying and categorizing cases. Machine learning, a branch of artificial intelligence, has demonstrated remarkable success in medical image processing, showing great potential for the early detection of diabetic retinopathy—a condition often challenging to diagnose in its early stages due to a lack of symptoms. This review examines prior studies leveraging machine learning algorithms, such as convolutional neural networks (CNNs), support vector machines (SVMs), and k-nearest neighbors (KNN), for diabetic retinopathy detection using fundus image datasets. It also explores existing challenges, including dataset variability, computational demands, and the generalizability of models across diverse populations. Highlighting methodologies, datasets, and performance metrics like accuracy, sensitivity, and specificity, this article aims to provide a cohesive understanding of the current landscape, delineate strengths and limitations, and suggest directions for future research.

Diabetic retinopathy is a kind of disorder which occurs due to high blood sugar level. This disorder affects retina in many ways. Blood vessels in the retina gets altered . Exudates are secreted, hemorrhages occur, swellings appear in the retina. Diabetic Retinopathy (DR) is the major cause of blindness. Automatic Recognition of DR lesions like Exudates, in digital fundus images can contribute to the diagnosis and screening of this disease. In this approach, an automatic and efficient method to detect the exudates is proposed. The real time retinal images are obtained from a nearby hospital. The retinal images are pre-processed via. Contrast Limited Adaptive Histogram Equalization (CLAHE). The preprocessed colour retinal images are segmented using K-Means Clustering technique. The segmented images establish a dataset of regions. To classify these segmented regions into Exudates and Non-Exudates, a set of features based on colour and texture are extracted. Classification is done using support Vector Machine This method appears promising as it can detect the very small areas of exudates. Keywords - Diabetic Retinopathy, Exudates, fundus image, k means clustering, SVM I. INTRODUCTION The retina is the inner and most important layer of the eye. It is composed of several important anatomical structures which can indicate various diseases. Cardiovascular disease such as stroke and myocardial infarcation can be identified from retinal blood vessels. Diabetic Retinopathy is the common retinal complication associated with diabetes. It is a major cause of blindness in both middle and advanced age groups. The International Diabetes Federation reports that over 50 million people in India have this disease and it is growing rapidly (IDF 2009a) (2). The estimated prevalence of diabetes for all age groups worldwide was 2.8% in 2000 and 4.4% in 2030 meaning that the total number of diabetes patients is forecasted to rise from 171 million in 2000 to 366 million in 2030 (3). Therefore regular screening is the most efficient way of reducing the vision loss. Diabetic Retinopathy is mainly caused by the changes in the blood vessels of the retina due to

Diabetic retinopathy (DR) is a leading contributor to vision impairment, particularly in areas with limited resources where access to specialized care is scarce. This study introduces an automated screening system for DR using attention- enhanced deep learning on retinal fundus images, specifically designed for these regions. The system leverages convolutional neural network (CNN) technology with integrated attention mechanisms to focus on critical features indicative of DR, such as microaneurysms and hemorrhages, improving detection accuracy and reliability. Varied retinal fundus images were used for training and validation, with data augmentation applied to enhance model robustness. The model was optimized for deployment on low-cost hardware, ensuring feasibility in resource-limited settings. Performance evaluation demonstrated high sensitivity and specificity, and attention maps provided interpretability for healthcare providers. This automated system has the potential to enhance early detection of diabetic retinopathy (DR) in underserved areas, facilitating timely intervention and reducing the risk of blindness. By making advanced diagnostic tools accessible, this approach promotes equitable healthcare and helps to prevent vision loss globally.

BackgroundWith the diabetes mellitus (DM) prevalence increasing annually, the human grading of retinal images to evaluate DR has posed a substantial burden worldwide. SmartEye is a recently developed fundus image processing and analysis system with lesion quantification function for DR screening. It is sensitive to the lesion area and can automatically identify the lesion position and size. We reported the diabetic retinopathy (DR) grading results of SmartEye versus ophthalmologists in analyzing images captured with non-mydriatic fundus cameras in community healthcare centers, as well as DR lesion quantitative analysis results on different disease stages.MethodsThis is a cross-sectional study. All the fundus images were collected from the Shanghai Diabetic Eye Study in Diabetics (SDES) program from Apr 2016 to Aug 2017. 19,904 fundus images were acquired from 6013 diabetic patients. The grading results of ophthalmologists and SmartEye are compared. Lesion quantification of several images at different DR stages is also presented.ResultsThe sensitivity for diagnosing no DR, mild NPDR (non-proliferative diabetic retinopathy), moderate NPDR, severe NPDR, PDR (proliferative diabetic retinopathy) are 86.19, 83.18, 88.64, 89.59, and 85.02%. The specificity are 63.07, 70.96, 64.16, 70.38, and 74.79%, respectively. The AUC are PDR, 0.80 (0.79, 0.81); severe NPDR, 0.80 (0.79, 0.80); moderate NPDR, 0.77 (0.76, 0.77); and mild NPDR, 0.78 (0.77, 0.79). Lesion quantification results showed that the total hemorrhage area, maximum hemorrhage area, total exudation area, and maximum exudation area increase with DR severity.ConclusionsSmartEye has a high diagnostic accuracy in DR screening program using non-mydriatic fundus cameras. SmartEye quantitative analysis may be an innovative and promising method of DR diagnosis and grading.