Diabetic Retinopathy Images Research Articles

Automatic Segmentation of Microaneurysm in Retinal Fundus Images

Diabetes interferes with the body’s ability to use and store sugar (glucose), which it can cause damage throughout the body. Generally, Microaneurysm (MA) is the first clinically observable lesions in the Diabetic Retinopathy (DR). Since interference of biological objects in eye fundus images are similar to MA, locate Microaneurysm lesions precisely is a challenging task for researches. This research employs U-Net Convolutional Neural Network to model saliency of objects in images, global and local context are both taken into account to provide a better initialization for the process. Feature extraction techniques are applied first to assign a local saliency value to each pixel by considering its local context from fundus images such as ORB, SURF, and MSER. The extracted feature vectors are applied for training the network. The sum of the weighted salient object regions produce the final saliency map, then implements U-Net to segment MA lesions.Our experiment has carried out using the publicly available Indian Diabetic Retinopathy Image Dataset (IDRiD), which has used in "Diabetic Retinopathy: Segmentation and Grading Challenge" workshop and our proposed method has given an outstanding accuracy of 98.78 %.

Gradient Descent Based Hyparameter Tuning of Xception Architecture for Diabetic Retinopathy Segmentation and Classification

Diabetic retinopathy (DR) is a widespread difficulty of diabetes and is considered as a main reason for vision loss in all over the globe. Several difficulties of DR can be avoided by controlling blood glucose level and timely medication. In real time, it is difficult to detect the DR and consumes more time in a manual way. This paper introduces a new Gradient Descent (GD) based Hyper parameter tuned Xception model called GD-Xception model to detect and classify DR images in an effective way. The GD-Xception model involves a series of subprocesses namely preprocessing, segmentation, feature extraction and classification. A set of extensive simulation takes place to ensure the effective outcome of the presented GD-Xception model. The presented model is tested using a DR dataset from Kaggle. The extensive experimental study clearly portrayed the superior outcome of the GD-Xception model with the maximum accuracy, sensitivity and specificity of 99.39%, 98.50% and 99.62% respectively.

Lesion Feature Detection of Retinopathy Images

In recent years, Diabetic Retinopathy (DR) is fast growing health problem in the society. Ophthalmologists use exudates, microaneurysms and hemorrhages features to diagnose the DR. Detection of these features plays important role in diagnosis and further treatment. This paper proposes automatic lesion detection and analysis methods so as to identify all three features without human intervention. The algorithm developed includes pre-processing as well as post processing, which is very important in DR images. Extraction of minute details in lesions requires highly enhanced images. Contrast enhancement and Morphological operations were applied on green channel DR images. Further, the segmentation is performed on enhanced DR images to capture required region of interest. The Kirsch algorithm has been applied to capture exudates candidate from DR image. Further Top-hat method on preprocessed image extracts microaneurysms and Thresholding technique has been applied for hemorrhage detection. The performance evaluation of algorithms is carried out through the publicly available database. The proposed algorithms depict better results in detection of three major parameters from digital retinopathy images with respect to toolkit evaluation adapted. This automatic detection method helps in reducing screening DR image time as well. The development of proposed algorithm leads to accurate DR detection by helping Ophthalmologists in the diagnosis process.

Distribution of peripheral lesions identified by mydriatic ultra-wide field fundus imaging in diabetic retinopathy.

To analyze the distribution of diabetic retinopathy (DR) lesions in an Indian population using ultra-wide field (UWF) fundus imaging. Seven hundred fifteen subjects (1406 eyes) with diabetic retinopathy in India were enrolled in this multicenter, prospective, observational study using UWF pseudocolor imaging with Optos Daytona Plus (Optos plc, Dunfermline, Scotland, UK). Images were transmitted to Doheny Image Reading Center, Los Angeles, CA, for grading. The ETDRS grid was overlaid on stereographic projections of UWF images, and images were graded independently by 2 masked graders. Lesion distribution was graded as predominantly central (PCL) or predominantly peripheral (PPL) according to previous criteria, considering both lesion number and area. An image was graded as PPL if > 50% of the lesion area was seen in at least one peripheral field as compared with the corresponding ETDRS field. Diabetic retinopathy severity was also assessed based on the International Classification of Diabetic Retinopathy (ICDR) grading scale. The main outcome measures were lesion distribution (PPL versus PCL): overall and within specific fields in eyes with various grades of DR. Lesion distribution was rated to be PPL in 37% of eyes and PCL in 63% of eyes (P < 0.003). The frequency of a PPL distribution varied significantly across all ICDR severity levels, with frequencies of mild non-proliferative DR (NPDR) (30.9%), moderate NPDR (40.3%), severe NPDR (38.5%) and PDR (34.9%), P = 0.005. When assessing which individual fields were rated to show a PPL distribution, the frequency was greatest in field 4 and least in field 7. For any grade of DR, temporal fields showed the greatest PPL frequency, followed in order by the superior, inferior, and nasal fields (P < 0.001). Only 3.5% of eyes showed PPL distribution in all five peripheral fields. One-third of the UWF images showed a PPL distribution in this cohort with the temporal quadrant having the widest distribution of PPL. As the PPL distribution varied significantly between various grades of DR, UWF imaging may prove to be important for screening of referral warranted retinopathy.

An Instinctive Application of Spatially Weighted Possibilistic Clustering Methods for the Detection of Lesions in Diabetic Retinopathy Images in Multi-dimensional Kernel Space

Diabetic Retinopathy is a result of microvascular changes linked with hyperglycemia. The pathology is characterized by swollen and weak blood capillaries resulted due to excess glucose absorption by endothelial walls of them. The swollen and thick capillaries may burst and exude protein and fat based particles into the retinal fundus, which get clotted and obstruct the vision flow into the eye. The clotted particles are called Exudates, whose volume and number will decide the severity of the condition. As a result new blood capillaries are triggered to nourish the retina. But they are weak and fragile and may eventually burst making the complication more critical. Eye diagnosis requires dilation of the pupa using mydriatic drops which tends to many temporary and prolonged side effects. Also the reliability of the diagnosis is dependent on the Ophthalmologists experience and require hectic labor and time. This paper presents an instinctive multi-dimensional kernel possibilistic c means clustering method with induced spatial constraint for the detection of retinal abnormalities. The proposed method helps the clinician by automating the lesion detection process and has proven its statistical significance compared to the existing state of art methods.

Diabetic retinopathy screening uptake after health education with or without retinal imaging within the facility in two AYUSH hospitals in Hyderabad, India: A nonrandomized pilot study

Purpose:In India, people with diabetes (PwDM) often seek care in the government-approved alternative medicine system, AYUSH (Ayurveda, Yoga and naturopathy, Unani, Siddha and Homeopathy). The purpose of this pilot study was to assess whether health education plus retinal imaging for diabetic retinopathy (DR) within an AYUSH hospital increased the uptake of screening for DR compared with health education and referral.Methods:The study was a nonrandomized pilot conducted in two AYUSH hospitals. Both hospitals received intervention on educating the AYUSH practitioners about DR screening and distributing health education materials to diabetic patients. In one hospital in addition to education, retinal imaging by a trained technician with remote grading by an ophthalmologist was provided, while in another hospital PwDM were referred to nearby eye hospitals for screening. The uptake of screening was assessed through registers and phone calls.Results:At baseline, only 10.7% of 178 PwDM were aware of DR and only 8% had undergone DR screening. After the intervention, in the hospital where screening was provided, all (100%) eligible patients (101) underwent digital imaging, whereas in the other hospital only 25% of 77 eligible patients underwent screening in eye hospitals (P < 0.001).Conclusion:AYUSH hospitals could provide a feasible and acceptable location for providing DR screening services. Further studies are required to assess scale-up of such intervention.

Automatic recognition and classification of diabetic retinopathy images by combining an attention mechanism and an efficient network

Automatic recognition and classification of diabetic retinopathy images by combining an attention mechanism and an efficient network

A Novel Texture Feature Extraction Using Curvelet Transform and Classification of Diabetic Retinopathy Images

A Novel Texture Feature Extraction Using Curvelet Transform and Classification of Diabetic Retinopathy Images

Classification among Microaneurysms, Exudates, and Lesion free Retinal Regions in the Eye Images using Transfer Learned CNNs

When pancreas fails to secrete sufficient insulin in the human body, the glucose level in blood either becomes too high or too low. This fluctuation in glucose level affects different body organs such as kidney, brain, and eye. When the complications start appearing in the eyes due to Diabetic Mellitus (DM), it is called Diabetic Retinopathy (DR). DR can be categorized in several classes based on the severity, it can be Microaneurysms (ME), Haemorrhages (HE), Hard and Soft Exudates (EX and SE). DR is a slow start process that starts with very mild symptoms, becomes moderate with the time and results in complete vision loss, if not detected on time. Early-stage detection may greatly bolster in vision loss. However, it is impassable to detect the symptoms of DR with naked eyes. Ophthalmologist harbor to the several approaches and algorithm which makes use of different Machine Learning (ML) methods and classifiers to overcome this disease. The burgeoning insistence of Convolutional Neural Network (CNN) and their advancement in extracting features from different fundus images captivate several researchers to strive on it. Transfer Learning (TL) techniques help to use pre-trained CNN on a dataset that has finite training data, especially that in under developing countries. In this work, we propose several CNN architecture along with distinct classifiers which segregate the different lesions (ME and EX) in DR images with very eye-catching accuracies.

A novel weighted compressive sensing using L1-magic recovery technique in medical image compression.

Recent technological advancement in computing technology, communication systems, and machine learning techniques provides opportunities to biomedical engineers to achieve the requirements of clinical practice. This requires storage and/or transmission of medical images with the conservation of the medical information over the communication channel. Accordingly, medical compression is necessary for efficient channel bandwidth utilization. To solve the trade-off between the compression ratio and the preservation of significant information, compressed sensing (CS) can be used. During image recovery in CS, an optimization algorithm is used, such as greedy pursuit, convex relaxation, and Bayesian framework. In the present work, a convex relaxation optimization called L1-magic is employed, where the objective function can be relaxed to the nearest convex norm, i.e., ℓ1-norm. In addition, the discrete cosine transform is used for recovery by transforming the image from time- to frequency-domain. To improve the medical image recovery, a weighted L1-magic is proposed using a threshold based on the image content, where high weight is given to the significant details in the image. Thus, the significant information in the image (values greater than the threshold) is multiplied by a weight factor according to the image characteristics for a successful recovery process. A comparative study of the proposed weighted L1-magic and orthogonal matching pursuit (OMP), one of the greedy algorithms, was conducted. Different metrics were measured, including the Structural Similarity Index Measure and Peak Signal-to-Noise Ratio (PSNR) to evaluate CS performance using the proposed weighted L1-magic as well as the weighted OMP and the principal component analysis (PCA) as a traditional compression method at different compression ratios (CR). The experimental results on diabetic retinopathy images dataset proved the superiority of the weighted L1-magic method, where for example as 0.4 CR, average PSNR is 19.37, 17.95, and 15.64 using the weighted L1-magic, weighted OMP, and PCA, respectively.

An Effective Approach to Classify Retina Images for Diabetic Retinopathy

An Effective Approach to Classify Retina Images for Diabetic Retinopathy

Identification of Diabetic Retinopathy from fundus images using CNNs

the Diabetic Retinopathy is the diabetes-mellitus to human vision that is the main cause of vision loss. The early stage detection of diabetic retinopathy is can play eminent role in the diabetes treatment. The fundus of retinal image is utilized to recognize the symptoms of diabetic retinopathy. Moreover, the above phenomena led us to propose this paper; here we propose segment based learning approach for identification of diabetic retinopathy. The segment based image level is required to obtain the identification of diabetic retinopathy images, the classifiers and features are equally learned from the data. Then, we adapt pre-trained CNN as the fine tune to achieve the segment level estimation of diabetic retinopathy. For identification of diabetic retinopathy, we achieve accuracy 96.97 and 98.46% at 20 and 30% and also achieve AUC (Area under Curve) 97.51 and 98.50 at 20 and 30% on the Kaggle dataset. Our proposed model outperforms much better than other models.

Automatic Detection of Diabetic Retinopathy in Retinal Fundus Photographs Based on Deep Learning Algorithm.

PurposeTo achieve automatic diabetic retinopathy (DR) detection in retinal fundus photographs through the use of a deep transfer learning approach using the Inception-v3 network.MethodsA total of 19,233 eye fundus color numerical images were retrospectively obtained from 5278 adult patients presenting for DR screening. The 8816 images passed image-quality review and were graded as no apparent DR (1374 images), mild nonproliferative DR (NPDR) (2152 images), moderate NPDR (2370 images), severe NPDR (1984 images), and proliferative DR (PDR) (936 images) by eight retinal experts according to the International Clinical Diabetic Retinopathy severity scale. After image preprocessing, 7935 DR images were selected from the above categories as a training dataset, while the rest of the images were used as validation dataset. We introduced a 10-fold cross-validation strategy to assess and optimize our model. We also selected the publicly independent Messidor-2 dataset to test the performance of our model. For discrimination between no referral (no apparent DR and mild NPDR) and referral (moderate NPDR, severe NPDR, and PDR), we also computed prediction accuracy, sensitivity, specificity, area under the receiver operating characteristic curve (AUC), and κ value.ResultsThe proposed approach achieved a high classification accuracy of 93.49% (95% confidence interval [CI], 93.13%–93.85%), with a 96.93% sensitivity (95% CI, 96.35%–97.51%) and a 93.45% specificity (95% CI, 93.12%–93.79%), while the AUC was up to 0.9905 (95% CI, 0.9887–0.9923) on the independent test dataset. The κ value of our best model was 0.919, while the three experts had κ values of 0.906, 0.931, and 0.914, independently.ConclusionsThis approach could automatically detect DR with excellent sensitivity, accuracy, and specificity and could aid in making a referral recommendation for further evaluation and treatment with high reliability.Translational RelevanceThis approach has great value in early DR screening using retinal fundus photographs.

An Instinctive Method for Lesion Detection in Diabetic Retinopathy Images using a Novel Spatial Possibilistic C means Clustering in Kernel Space

Diabetic Retinopathy (DR) is the prime cause of vision impediment which originates due micro vascular changes and hyperglycemia stimulated by Diabetes Mellitus (DM). Endothelial lining of the blood capillaries absorbs excess amount of glucose (glycoproteins), hence become thick but are fragile. The swollen capillaries may burst and leak water, proteins, and lipids and tends to fovea expansion. Further it triggers the revascularization to nourish retinal fundus. These new blood capillaries are weak and fragile and can further progress the state to chronic. Pupil dilation for fundus observation leads to many ill effects like head ache, brow pain, blurred vision and light sensitivity. Ophthalmologists cannot administer the pathos well if the symptoms are indolently addressed by the patient and therefore reliability during subjective diagnosis lags. This paper address a qualitative evaluation of novel possibilistic clustering methods with induced spatial constraint in kernel domain to detect the presence of exudates in non-dilated DR images. This methods are compiled in an N dimensional Kernel space which helps to easily segregate the non-linear data regions present in the lower dimensional input space. Also the inclusion of spatial information of a pixel neighborhood will improves the noise handling capability of the proposed methods by easily extricating the noisy pixels from the target lesions and hence improves the system’s accuracy in attaining reliable data. Statistical evaluation reveals that the proposed algorithms has attained better sensitivity and specificity compared to existing state-of-art works.

ANN and SVM to recognize Texture features for spontaneous Detection and Rating of Diabetic Retinopathy

The higher levels of blood glucose most often causes a metabolic disorder commonly called as Diabetes, scientifically as Diabetes Mellitus. A consequence of this is a major loss of vision and in long terms may eventually cause complete blindness. It initiates with swelling on blood vessels, formation of microaneurysms at the end of narrow capillaries. Haemorrhages due to rupture of small vessels and fluid leak causes exudates. The specialist examines it to diagnose and gives proper treatment. Fundus images are the fundamental tool for proper diagnosis of patients by medical experts. In this research work the fundus images are taken for processing, the neural network and support vector machine are trained for the proposed model. The features are extracted from the diabetic retinopathy image by using texture based algorithms such as Gabor, Local binary pattern and Gray level co-occurrence matrix for rating the level of diabetic retinopathy. The performance of all methods is calculated based on accuracy, precision, Recall and f-measure.

Recognition and Classification of Diabetic Retinopathy utilizing Digital Fundus Image with Hybrid Algorithms

Diabetic Retinopathy (Damage in Retina) is the most common threatening diabetic eye disease and cause leading vision loss and blindness. A patient with the diabetic disease needs to experience occasional screening of eye. To analysis, ophthalmologists may utilize fundus or retinal pictures of the patient gained from advanced fundus camera. However, if the symptoms are identified earlier and proper treatment is provided through regular screening and monitoring, the blindness or vision loss can be avoided. The present study is intended on developing an automatic system for the analysis of the retinal of fundus images by using image-processing techniques. So as to accelerate the procedure, the discovery of diabetic retinopathy image processing methods is utilized In this proposed study, the performance is evaluated using different segmentation algorithms and classifiers namely fuzzy c-means clustering, naïve Bayesian classifier, support vector machine to detect the diabetic retinopathy. The presentation of the strategy is assessed on the freely accessible retinal databases like DRIVE, STARE. The presentation of the retinal vessels on DRIVE database, sensitivity 100% and specificity 97.5% while for STARE database the sensitivity 99%, specificity 97%.The detection of accuracy can be defined with respect to expert physician hand-drawn and ground truths and the results are comparatively obtained and analyzed.

Impact of an educational intervention on eye gaze behaviour in retinal image interpretation by consultant and trainee ophthalmologists.

This study uses eye-tracking technology to assess the differences in gaze behaviours between ophthalmologists of different experience levels while interpreting retinal images of diabetic retinopathy. The differences in gaze behaviours before and after a teaching intervention which introduced a suggested search strategy is also investigated. A total of 9 trainees and 10 consultant ophthalmologists interpreted six retinal images. They were then shown a 5-min tutorial that demonstrated a search strategy. This was followed by six further retinal image interpretations. Participants completed questionnaires indicating clinical signs seen, appropriate retinopathy grade, and confidence. Eye movements were tracked during each interpretation.Overall, trainees compared to consultants demonstrated more uncertain and unstructured gaze behaviours. Trainee eye gaze metrics included: longer interpretation time, 36.5 s (SD = 6.2 vs. 31.4 s) (SD = 4.2) (p = 0.024), higher visit count, 17.38 visits (SD = 5.13) versus 12.18 visits(SD = 2.64) (p = 0.01), higher proportion of fixation, 57.0 per cent (SD = 5) versus 50.5 per cent (SD = 5) (p = 0.05) and shorter time to first fixation, 0.232 s (SD = 0.10) versus 0.821 s (SD = 0.77) (p = 0.001), respectively. The teaching intervention resulted in more focused gaze patterns in both groups. Pre-intervention and post-intervention mean proportion fixation on areas of interest were 38.6 per cent (SD = 6.8) and 51.8 per cent (SD = 13.9) for the trainee group, respectively, and 39.9 per cent (SD = 4.1) and 50.9 per cent (SD = 9.3) for the consultant group (p = 0.01).Consultants used more systematic and efficient approaches than trainees during interpretation. After the introduction of a suggested search strategy, trainees showed trends towards consultant eye gaze behaviours. Eye tracking gives an interesting insight into the thought processes of physicians carrying out complex tasks. The implication is that eye tracking may have future use in teaching and assessment. Its use in objectively assessing different teaching strategies could be a valuable tool for medical education.

IDRiD: Diabetic Retinopathy - Segmentation and Grading Challenge.

Diabetic Retinopathy (DR) is the most common cause of avoidable vision loss, predominantly affecting the working-age population across the globe. Screening for DR, coupled with timely consultation and treatment, is a globally trusted policy to avoid vision loss. However, implementation of DR screening programs is challenging due to the scarcity of medical professionals able to screen a growing global diabetic population at risk for DR. Computer-aided disease diagnosis in retinal image analysis could provide a sustainable approach for such large-scale screening effort. The recent scientific advances in computing capacity and machine learning approaches provide an avenue for biomedical scientists to reach this goal. Aiming to advance the state-of-the-art in automatic DR diagnosis, a grand challenge on "Diabetic Retinopathy - Segmentation and Grading" was organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI - 2018). In this paper, we report the set-up and results of this challenge that is primarily based on Indian Diabetic Retinopathy Image Dataset (IDRiD). There were three principal sub-challenges: lesion segmentation, disease severity grading, and localization of retinal landmarks and segmentation. These multiple tasks in this challenge allow to test the generalizability of algorithms, and this is what makes it different from existing ones. It received a positive response from the scientific community with 148 submissions from 495 registrations effectively entered in this challenge. This paper outlines the challenge, its organization, the dataset used, evaluation methods and results of top-performing participating solutions. The top-performing approaches utilized a blend of clinical information, data augmentation, and an ensemble of models. These findings have the potential to enable new developments in retinal image analysis and image-based DR screening in particular.

ES08.07 System Approach to Screening Management

ES08.07 System Approach to Screening Management

Retinal blood vessel segmentation from diabetic retinopathy images using tandem PCNN model and deep learning based SVM

Diabetic Retinopathy (DR) occurs due to Type-II diabetes. It causes damages to the retinal blood vessels and reason for visual impairment. The predicted center is around the probability of variation in the estimation of retinal veins, and the crisp enrolls vessel development inside the retina. To witness the changes segmentation of retinal blood vessels has to be made. A framework to upgrade the quality of the segmentation results over morbid retinal images is proposed. This framework utilizes Contrast Limited Adaptive Histogram Equalization (CLAHE) for eliminating the background from the source image and enhances the foreground blood vessel pixels, Tandem Pulse Coupled Neural Network (TPCNN) model is endorsed for automatic feature vectors generation, and Deep Learning Based Support Vector Machine (DLBSVM) is proposed for classification and extraction of blood vessels. The DLBSVM parameters are fine-tuned via Firefly algorithm. The STARE, DRIVE, HRF, REVIEW, and DRIONS fundus image datasets are deliberated to assess the recommended techniques. The results render that the proposed technologies improve the segmentation with 80.61% Sensitivity, 99.54% Specificity, and 99.49% Accuracy.