DIARETDB1 Dataset Research Articles

Detection of optic disc in human retinal images utilizing the Bitterling Fish Optimization (BFO) algorithm

Early detection and correct identification of the optic disc (OD) on scanned retinal images are significant for diagnosing and treating several ophthalmic conditions, including glaucoma and diabetic retinopathy. Conventional methods for detecting the OD often struggle with processing retinal images due to noise, changes in illumination, and complex overlapping images. This study presents the development of effective and accurate fixation of the optic disc using the Bitterling Fish Optimization (BFO) algorithm, which enhances the processes of OD imaging in speed and precision. The proposed method begins with image enhancement and noise suppression for preprocessing, followed by applying the BFO algorithm to locate and delineate the OD region. The performance evaluation of the algorithm was conducted within several public domain retinal images, including DRIVE, STARE, ORIGA, DRISHTI-GS, DiaRetDB0, and DiaRetDB1 datasets about some internal metrics: sensitivity (SE), specificity (SP), accuracy (ACC), DICE overlap coefficient, overlap and time of processing respectively. The technique based on BFO provided better results, with 99.33%, 99.94%, and 98.22% accuracy achieved for OD in DRIVE, DRISHTI-GS, and DiaRetDB 1, respectively. The approach also demonstrated high overlaps and good DICE results, with a DICE coefficient of 0.9501 for the DRISHTI-GS database. On average, the processing time per image was less than 2.5 s, proving the approach’s efficiency in computations. The BFO approach has demonstrated its effectiveness and scalability in detecting optic discs in retinal images in an automated manner. It showed impressive performance levels in terms of computation time and accuracy and was variation resistant irrespective of the quality of the image and the pathology present on it. This method holds significant potential for clinical use, providing a meaningful way of diagnosing and managing ocular disease at an early stage.

Assessment of retinal blood vessel segmentation using U-Net model: A deep learning approach

Segmentation of retinal blood vessels from fundus images is vital to assist ophthalmologists in diagnosing different eye diseases like Arteriosclerosis, Glaucoma, Diabetic Retinopathy, Hypertension, and Choroidal Neovascularization. Accurate detection and timely treatment of these eye diseases can prevent irreversible impairment of vision. In computer-aided automatic diagnosis, Preprocessing of fundus images is necessary to reduce the uneven retinal image illumination, as well as the undesirable effect arising from the contrast differences between the retinal blood vessels and the background. In this study, the Hessian-based Frangi vesselness filter is proposed to enhance vasculature in fundus images. This approach seeks to extract thin vessels to diminish the intensity disparity between thick and thin vessels. Its accuracy has never, however, been thoroughly evaluated. To verify the effectiveness of the suggested filter for boosting vessel-like structures in the fundus images, an experimental approach is presented in this paper. The ability of the filter to improve vessel-like structures in fundus images is validated, and an experimental technique for evaluating filter performance using the U-Net model is described. Five quantitative performances, namely Accuracy, Recall, Specificity, Precision, and F1 Score measures on the DRIVE, HRF, DIARETDB1, DIARETDB0, CHASEDB1, ORIGA, DRISHTI GS, DRIONS DB, FIRE, and FIOT datasets, were evaluated and quantified in these experiments to validate the efficacy of the proposed filter. The results demonstrated a noteworthy improvement by achieving an Accuracy of 99.79 %, 99.84 %, 99.64 %, 99.72 %, 99.57 %, 99.48 %, 99.77 %, 99.05 %, 99.93 %, and 99.40 % respectively. Empirical evidence supports the advantages of this approach.

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.

Haemorrhage diagnosis in colour fundus images using a fast-convolutional neural network based on a modified U-Net

ABSTRACT Retinal haemorrhage stands as an early indicator of diabetic retinopathy, necessitating accurate detection for timely diagnosis. Addressing this need, this study proposes an enhanced machine-based diagnostic test for diabetic retinopathy through an updated UNet framework, adept at scrutinizing fundus images for signs of retinal haemorrhages. The customized UNet underwent GPU training using the IDRiD database, validated against the publicly available DIARETDB1 and IDRiD datasets. Emphasizing the complexity of segmentation, the study employed preprocessing techniques, augmenting image quality and data integrity. Subsequently, the trained neural network showcased a remarkable performance boost, accurately identifying haemorrhage regions with 80% sensitivity, 99.6% specificity, and 98.6% accuracy. The experimental findings solidify the network’s reliability, showcasing potential to alleviate ophthalmologists’ workload significantly. Notably, achieving an Intersection over Union (IoU) of 76.61% and a Dice coefficient of 86.51% underscores the system’s competence. The study’s outcomes signify substantial enhancements in diagnosing critical diabetic retinal conditions, promising profound improvements in diagnostic accuracy and efficiency, thereby marking a significant advancement in automated retinal haemorrhage detection for diabetic retinopathy.

Predicting diabetic macular edema in retina fundus images based on optimized deep residual network techniques on medical internet of things

With the medical internet of things, many automated diagnostic models related to eye diseases are easier. The doctors could quickly contrast and compare retina fundus images. The retina image contains a lot of information in the image. The task of detecting diabetic macular edema from retinal images in the healthcare system is difficult because the details in these images are very small. This paper proposed the new model based on the medical internet of things for predicting diabetic macular edema in retina fundus images. The method called DMER (Diabetic Macular Edema in Retina fundus images) to detect diabetic macular edema in retina fundus images based on improving deep residual network being combined with feature pyramid network in the context of the medical internet of things. The DMER method includes the following stages: (i) ResNet101 improved combining with feature pyramid network is used to extract features of the image and obtain the map of these features; (ii) a region proposal network to look for potential anomalies; and (iii) the predicted bounding boxes against the true bounding box by the regression method to certify the capability of macular edema. The MESSIDOR and DIARETDB1 datasets are used for testing with evaluation criteria such as sensitivity, specificity, and accuracy. The accuracy of the DMER method is about 98.08% with MESSIDOR dataset and 98.92% with DIARETDB1 dataset. The results of the method DMER are better than those of the other methods up to the present time with the above datasets.

DETECTION DIABETIC RETINOPATHY WITH SUPERVISED LEARNING

Diabetic retinopathy is a common complication that occurs in people with diabetes mellitus. Diabetic retinopathy damage is characterized in the blood vessel system in the layer at the back of the eye, especially in tissues that respond to light. This research aims to detect diabetic retinopathy early by using SVM and Random forest. SVM is a classification technique that divides the input space into two classes. Random Forest is a supervised learning algorithm that utilizes a collection of decision trees trained using the bagging method. This research uses datasets from diaretdb1 and messidor to evaluate the performance of both methods. The diaretdb1 dataset consists of 178 data points with the diagnosis of Proliferative Diabetic Retinopathy and Non-Diabetic Retinopathy. In addition, the messidor dataset consists of 105 data points with the diagnosis of Diabetic Retinopathy and Non-Diabetic Retinopathy. Experimental results on the diaretdb1 dataset showed that SVM achieved 88% accuracy, while Random Forest achieved 91% accuracy. Similarly, on the messidor dataset, SVM achieved 80% accuracy, while Random Forest achieved 85% accuracy.

A Retinal Lesion Detection of Diabetic Retinopathy

Diabetic retinopathy (DR) is the most common cause of vision loss among people with diabetes and leading to the cause of blindness. Microaneurysms (MAs) are primary lesion of Diabetic retinopathy, so their detection can give time to prevent further vision loss. Hemorrhages (HMs) and exudates (EXs) are the other type of lesions occurs during Diabetic retinopathy. The Retinal lesion detection consists of retinal image preprocessing, Optic disc detection and removal, Blood vessel segmentation, Candidate Lesion detection, Feature extraction, Classification and post processing . Publicly available DiaretDB1 dataset is used for performance evaluation.

Optic Disc Detection on Retina Image using Extreme Learning Machine

Optic disk detection on retina image has become one of many initial steps in evaluation of Diabetic Macular Edema (DME) severity. As much as early the step is, the result of the step is extremely essential. This article discusses the optic disk detection on retina image based on the color histogram value. The detection is done by using color histogram value which is taken from window sliding process with the size of 50x50 pixels. First, the candidates of optic disc were detected using Extreme Learning Machine towards the histogram value. Then the optic disc was selected form the candidates of optic which has highest average intensity. 4 retina image datasets were employed in the evaluation, including Drions dataset, DRIVE dataset, DiaretDB1 dataset, and Messidor dataset. The result of evaluation then validated by medical expert. The model outcome reaches the accuracy as much as 85,39 % for DiaretDB1 dataset, 95% for DRIVE dataset, 98,18% for Drions and 99% for Messidor dataset.

Optic Disc Localization using Fuzzy C Mean and DB Scan Clustering - A Comparative Analysis

In the diagnosis and early detection of Glaucoma and diabetic retinopathy, when delicate vasculature grows in the retina, precise identification and localization of the border optic disc are highly significant. This research provides an automated method to localize and detect the optic disc. The proposed method uses the clustering methodology to locate the optic disc region. Fuzzy C Mean and Density Based Scan (DB Scan) clustering approach is evaluated on the publicly accessible DRIVE, diaretdb1, diaretdb0, and databases, which were created to aid comparative investigations on optic disc localization and detection in retinal images. With Diaretdb0 and Drive DB, the DB Scan clustering approach obtained an accuracy of 94.11% and 81.18%, respectively, which is better than the Fuzzy C mean, and it performs DB scan better for the DiaretDB1 dataset.

Exudate identification in retinal fundus images using precise textural verifications

One of the most salient diseases of retina is Diabetic Retinopathy (DR) which may lead to irreparable damages to eye vision in the advanced phases. A large number of the people infected with diabetes experience DR. The early identification of DR signs facilitates the treatment process and prevents from blindness. Hard Exudates (HE) are bright lesions appeared in retinal fundus images of DR patients. Thus, the detection of HEs is an important task preventing the progress of DR. However, the detection of HEs is a challenging process due to their different appearance features. In this paper, an automatic method for the identification of HEs with various sizes and shapes is proposed. The method works based on a pixel-wise approach. It considers several semi-circular regions around each pixel. For each semi-circular region, the intensity changes around several directions and non-necessarily equal radiuses are computed. All pixels for which several semi-circular regions include considerable intensity changes are considered as the pixels located in HEs. In order to reduce false positives, an optic disc localization method is proposed in the post-processing phase. The performance of the proposed method has been evaluated on DIARETDB0 and DIARETDB1 datasets. The experimental results confirm the improved performance of the suggested method in term of accuracy.

A Computationally Efficient Red-Lesion Extraction Method for Retinal Fundus Images

Retina is an important organ of the body, the diseases of which may lead to serious damages to human vision. Fundus retinal images are the common tools for the analysis of diabetic retinopathy (DR), which is an important retinal disease. Red-lesions are from important manifestations of DR in the fundus images. In this article, a novel method is suggested for the extraction of red-lesions from fundus images. This method can detect red-lesions without the need for prior segmentation of blood vessels or lesions. The new method works based on dividing the fundus image into square patches and finding the dark ones based on the percentage of dark pixels. After finding dark patches, it is necessary to discriminate the patches that belong to the blood vessel and red-lesion. The continuing structure of blood vessels is considered a discriminating factor for the mentioned purpose. To mathematically model the continuing structure, several states are considered for the way of locating dark patches in a neighborhood. The formation of the blood vessel in vertical, horizontal, and diagonal directions is modeled in the different states. Also, the conditions of the formation of red-lesion in each direction are declared. The performance of the proposed method is evaluated on several datasets. The simplicity of computations, high speed, and acceptable accuracy are significant advantages of this method. The proposed method is capable of providing 92% and 88% for sensitivity (SE) and specificity (SP) in the Diaretdb1 dataset. Also, it provides the values of 91% and 89% for SE and SP in the Diaretdb0 dataset. Furthermore, the SE and SP values for the FIRE dataset are equal to 90% and 92%, respectively.

CFLHCF: Simultaneous Detection of the Optic Disc and Exudates Using Color Features, Local Homogeneity and Contextual Features

Detecting Optic Disc (OD) and Exudates (EXs) in the fundus images has been challenging and demanding for the computer-aided diagnosis system. Existing algorithms for detecting OD and EXs are mainly based on traditional learning methods that heavily rely on enhanced OD and EXs features. Unlike traditional learning methods, a novel simultaneous detection of OD and EXs is presented. In this proposed novel Color Features, Local Homogeneity and Contextual Features (CFLHCF), the input original fundus image is preprocessed by color normalization, contrast enhancement, noise removal, and OD localization which the EXs is differentiated from its background information. Then, the preprocessed images are given as the input to Mathematical Morphology Binary Segmentation (MMBS) with Sobel Edge Detection (SED) technique, which detects the EXs from the given fundus images. An MMBS with a SED technique is implemented to boost a highly accurate segmented model for small EXs regions of EXs. The DiaretDB0, DiaretDB1, and STARE datasets are used to validate the proposed method. On the DiaretDB0 dataset, this technique archived an average sensitivity of 98.44% for EXs and a specificity of 98.72% for non‒EXs, which can potentially classify EXs even when the regions are trivial. With respect to sensitivity and specificity values, this method outperformed the previous state‒of‒the‒art methods by roughly 1.24% and 1.09% in the detection of EXs. Additionally, we show the EXs‒diagnosis in ~7 seconds per image.

Hybrid deep learning approaches for the detection of diabetic retinopathy using optimized wavelet based model

Diabetic retinopathy (DR) is a diabetic ocular manifestation that leads to loss of visual impairment and blindness in people worldwide. Detecting and diagnosing the DR remains a major question in this task. This developed work uses a robust hybrid binocular Siamese with a deep learning approach to classify the DR image. Initially, a pre-processing stage is introduced to remove unwanted noises. To perform this cross guided bilateral filter (CGBF) approach is emphasized. After the pre-processing, the feature extraction stage is presented to extract the features from the processed image. A wavelet-based Chimp optimization algorithm (WBCOA) is established for the extraction of features. After feature extraction, segmentation of optical disc (OD) and blood vessel (BV) is done via open closed watershed management (OCWSM). For the classification, binocular Siamese based AlexNet and GoogleNet with the SVM model are proposed in this work. The segmented OD and BV are input to the proposed hybrid DL network. Finally, the extracted images are fused and classified using the Support Vector Machine (SVM) model. The proposed method is implemented in Python and tested on DIARETDB0 (DB0) and DIARETDB1 (DB1) datasets. The proposed hybrid DL network attained 94% and 94.83% accuracy on DB0 and DB1 datasets, respectively. Also, the proposed model’s outcomes are compared with various existing approaches. The proposed method conducted statistical analysis for the DR image based on mean and standard deviation (SD) to obtain an efficient output. These outcomes prove that the proposed hybrid DL system is accurate for early DR detection and deliver effective treatment of diabetes.

Hemorrhage semantic segmentation in fundus images for the diagnosis of diabetic retinopathy by using a convolutional neural network

Because retinal hemorrhage is one of the earliest symptoms of diabetic retinopathy, its accurate identification is essential for early diagnosis. One of the major obstacles ophthalmologists face in making a quick and effective diagnosis is viewing too many images to manually identify lesions of different shapes and sizes. To this end, researchers are working to develop an automated method for screening for diabetic retinopathy. This paper presents a modified CNN UNet architecture for identifying retinal hemorrhages in fundus images. Using the graphics processing unit (GPU) and the IDRiD dataset, the proposed UNet was trained to segment and detect potential areas that may harbor retinal hemorrhages. The experiment was also tested using the IDRiD and DIARETDB1 datasets, both freely available on the Internet. We applied preprocessing to improve the image quality and increase the data, which play an important role in defining the complex features involved in the segmentation task. A significant improvement was then observed in the learning neural network that was able to effectively segment the bleeding and achieve sensitivity, specificity and accuracy of 80.49%, 99.68%, and 98.68%, respectively. The experimental results also yielded an IoU of 76.61% and a Dice value of 86.51%, showing that the predictions obtained by the network are effective and can significantly reduce the efforts of ophthalmologists. The results revealed a significant increase in the diagnostic performance of one of the most important retinal disorders caused by diabetes.

Retinal fundus image classification for diabetic retinopathy using SVM predictions.

Diabetic Retinopathy (DR) is one of the leading causes of blindness in all age groups. Inadequate blood supply to the retina, retinal vascular exudation, and intraocular hemorrhage cause DR. Despite recent advances in the diagnosis and treatment of DR, this complication remains a challenging task for physicians and patients. Hence, a comprehensive and automated technique for DR screening is necessary, which will give early detection of this disease. The proposed work focuses on 16 class classification method using Support Vector Machine (SVM) that predict abnormalities individually or in combination based on the selected class. Our proposed work comprises Gaussian mixture model (GMM), K-means, Maximum a Posteriori (MAP) algorithm, Principal Component Analysis (PCA), Grey level co-occurrence matrix (GLCM), and SVM for disease diagnosis using DR. The proposed method provides an accuracy of 77.3% on DIARETDB1 dataset. We expect this low computational cost will be helpful in the medicine and diagnosis of DR.

Diabetic retinopathy classification using a novel DAG network based on multi-feature of fundus images

Diabetic retinopathy is a kind of ophthalmic disease induced by diabetes, which is a chronic progressive disease affecting vision and even causing blindness. The features of diabetic retinopathy are of great significance to the effective diagnosis and prognosis of ophthalmologists. This paper proposes a novel DAG network model for the classification of diabetic retinopathy based on multi-feature fusion of fundus images. Firstly, under the advice of the doctor, three indicative features of diabetic retinopathy are extracted using different algorithms: retinal hemorrhagic plaque, fundus neovascularization, and retinal varices. Then three features are sent to a classification model based on a novel DAG network for realizing multi-feature fusion and feature learning. Finally, the optimized classification model is used to recognize and classify diabetic retinopathy. DIARETDB1 dataset and real hospital data from Dalian NO.3 People’s Hospital are used to evaluate the performance of the proposed method. For the DIARETDB1 dataset and real hospital data, the accuracy can reach 98.7% and 98.5%, respectively. The proposed model has high efficiency and practical application, which can assist doctors in the diagnosis and treatment of diabetic retinopathy.

An Ensemble Classifier Based on Individual Features for Detecting Microaneurysms in Diabetic Retinopathy

Individuals with diabetes are more likely to develop Diabetic Retinopathy (DR), a chronic ailment that can lead to blindness if left undiagnosed. Early-stage Diabetic Retinopathy (DR) is characterized by Microaneurysms (MA), which appear as tiny red lesions on the retina. This paper investigates a unique approach for the automated early identification of microaneurysms in eye images. A unique ensemble classifier technique is suggested in this work. Classifiers like SVM, KNN, Decision Tree, and Naïve Bayes are chosen in this study for building an ensemble model. After preprocessing the image , certain common image characteristics such as shape and intensity features were retrieved from the candidate. The mean absolute difference of each feature is computed. Based on mean ranges that would give improved classification results, an expert classifier is chosen and trained. The outputs of the classifiers are integrated for each of the distinct characteristics, and the number of categories that have been most frequently repeated is utilized to reach a final decision. The process has been comprehensively validated using two available open datasets, like e-ophtha and DIARETDB1. On the e-ophtha and DIARETDB1 datasets, the ensemble model achieved an AUC of 0.928 and 0.873, Sensitivity of 90.7% and 85%, Specificity of 90% and 91% respectively.

General deep learning model for detecting diabetic retinopathy

BackgroundDoctors can detect symptoms of diabetic retinopathy (DR) early by using retinal ophthalmoscopy, and they can improve diagnostic efficiency with the assistance of deep learning to select treatments and support personnel workflow. Conventionally, most deep learning methods for DR diagnosis categorize retinal ophthalmoscopy images into training and validation data sets according to the 80/20 rule, and they use the synthetic minority oversampling technique (SMOTE) in data processing (e.g., rotating, scaling, and translating training images) to increase the number of training samples. Oversampling training may lead to overfitting of the training model. Therefore, untrained or unverified images can yield erroneous predictions. Although the accuracy of prediction results is 90%–99%, this overfitting of training data may distort training module variables.ResultsThis study uses a 2-stage training method to solve the overfitting problem. In the training phase, to build the model, the Learning module 1 used to identify the DR and no-DR. The Learning module 2 on SMOTE synthetic datasets to identify the mild-NPDR, moderate NPDR, severe NPDR and proliferative DR classification. These two modules also used early stopping and data dividing methods to reduce overfitting by oversampling. In the test phase, we use the DIARETDB0, DIARETDB1, eOphtha, MESSIDOR, and DRIVE datasets to evaluate the performance of the training network. The prediction accuracy achieved to 85.38%, 84.27%, 85.75%, 86.73%, and 92.5%.ConclusionsBased on the experiment, a general deep learning model for detecting DR was developed, and it could be used with all DR databases. We provided a simple method of addressing the imbalance of DR databases, and this method can be used with other medical images.

Red-lesion extraction in retinal fundus images by directional intensity changes\u2019 analysis

Diabetic retinopathy (DR) is an important retinal disease threatening people with the long diabetic history. Blood leakage in retina leads to the formation of red lesions in retina the analysis of which is helpful in the determination of severity of disease. In this paper, a novel red-lesion extraction method is proposed. The new method firstly determines the boundary pixels of blood vessel and red lesions. Then, it determines the distinguishing features of boundary pixels of red-lesions to discriminate them from other boundary pixels. The main point utilized here is that a red lesion can be observed as significant intensity changes in almost all directions in the fundus image. This can be feasible through considering special neighborhood windows around the extracted boundary pixels. The performance of the proposed method has been evaluated for three different datasets including Diaretdb0, Diaretdb1 and Kaggle datasets. It is shown that the method is capable of providing the values of 0.87 and 0.88 for sensitivity and specificity of Diaretdb1, 0.89 and 0.9 for sensitivity and specificity of Diaretdb0, 0.82 and 0.9 for sensitivity and specificity of Kaggle. Also, the proposed method has a time-efficient performance in the red-lesion extraction process.

Detection of Diabetic Eye Disease from Retinal Images Using a Deep Learning Based CenterNet Model.

Diabetic retinopathy (DR) is an eye disease that alters the blood vessels of a person suffering from diabetes. Diabetic macular edema (DME) occurs when DR affects the macula, which causes fluid accumulation in the macula. Efficient screening systems require experts to manually analyze images to recognize diseases. However, due to the challenging nature of the screening method and lack of trained human resources, devising effective screening-oriented treatment is an expensive task. Automated systems are trying to cope with these challenges; however, these methods do not generalize well to multiple diseases and real-world scenarios. To solve the aforementioned issues, we propose a new method comprising two main steps. The first involves dataset preparation and feature extraction and the other relates to improving a custom deep learning based CenterNet model trained for eye disease classification. Initially, we generate annotations for suspected samples to locate the precise region of interest, while the other part of the proposed solution trains the Center Net model over annotated images. Specifically, we use DenseNet-100 as a feature extraction method on which the one-stage detector, CenterNet, is employed to localize and classify the disease lesions. We evaluated our method over challenging datasets, namely, APTOS-2019 and IDRiD, and attained average accuracy of 97.93% and 98.10%, respectively. We also performed cross-dataset validation with benchmark EYEPACS and Diaretdb1 datasets. Both qualitative and quantitative results demonstrate that our proposed approach outperforms state-of-the-art methods due to more effective localization power of CenterNet, as it can easily recognize small lesions and deal with over-fitted training data. Our proposed framework is proficient in correctly locating and classifying disease lesions. In comparison to existing DR and DME classification approaches, our method can extract representative key points from low-intensity and noisy images and accurately classify them. Hence our approach can play an important role in automated detection and recognition of DR and DME lesions.