DIARETDB1 Database Research Articles

Interval type-2 fuzzy approach for retinopathy detection in fundus images

In the manuscript, an automatic approach for analysis and detection of various stages of retinopathy defects in human eyes has been proposed. The approach consists of a robust preprocessing technique of the retina fundus image to mitigate the effects of noise and poor lightening in the image. To realize a compressive analysis of the defects, methods for extracting blood vessels and optic disc in the fundus image has also been developed. Adaptive Histogram Equalization (AHE), median filtering and Connected Component Analysis techniques were used in separating blood vessels and optic disc from each fundus image. The pre-processing utilizes canny edge detection and Morphological Closing on the fundus image. An interval type-2 fuzzy (IT2F) clustering is applied to segments an input image into four clusters. These four clusters from the fuzzy segmentation are further analyzed to extract various stages of retinopathy abnormalities (e.g., Hemorrhage, hard exudates etc.). The extracted blood vessels and optic disc are removed from the analysis to enhance the defects detection process. Experiments were conducted on DIARETDB1 database. The experimental results obtained are validated using the ground-truth images contained in DIARETDB1 database. Impressive results are recorded throughout the experiment. Hard-Exudates and Hemorrhage were detected from the fundus images and results from similarity indexes such as, accuracy (94.11%) sensitivity (93.03%) and specificity (98.45%) were recorded.

LW-CNN-based extraction with optimized encoder-decoder model for detection of diabetic retinopathy

<p>In the field of computer vision, automatic diabetic retinopathy (D.R.) screening is a well-established topic of study. It’s tough since the retinal vessels are hardly distinguishable from the backdrop in the fundus picture, and the structure is complicated. To learn data representations at numerous levels of abstraction, deep learning (DL) allows for the development of computational representations with several processing layers. Small, inconspicuous lesions generated by the disorder are hard to detect since they are tucked away beneath the eye’s structure. In this research, a lightweight convolutional neural network (LW-CNN) was used to extract structures from images of blood vessels, and different preprocessing methods were employed. The features are extracted, and then D.R. is classified using the suggested learning technique, which includes an encoder, dense branch. Effective categorization relies on the usage of multi-scale information collected from various nodes in the network. Grasshopper’s optimisation algorithm (GHOA) is used to fine-tune the recommended classifier’s hyper-parameters. The DIARETDB1 benchmark dataset is assessed using 80% training data and 20% testing data to get a diagnosis of the disease’s severity. The proposed model improved D.R. image classification with accuracy of 0.992 for DIARETDB1 database and 0.981 for APTOS 2019 blindness detection dataset. The state-of-the-art models for D.R. dataset images only achieved less accuracy and precision as compared with the proposed model.</p>

Detection and Classification of Hemorrhages in Retinal Images

Damage of the blood vessels in retina due to diabetes is called diabetic retinopathy (DR). Hemorrhages is the first clinically visible symptoms of DR. This paper presents a new technique to extract and classify the hemorrhages in fundus images. The normal objects such as blood vessels, fovea and optic disc inside retinal images are masked to distinguish them from hemorrhages. For masking blood vessels, thresholding that separates blood vessels and background intensity followed by a new filter to extract the border of vessels based on orientations of vessels are used. For masking optic disc, the image is divided into sub-images then the brightest window with maximum variance in intensity is selected. Then the candidate dark regions are extracted based on adaptive thresholding and top-hat morphological techniques. Features are extracted from each candidate region based on ophthalmologist selection such as color and size and pattern recognition techniques such as texture and wavelet features. Three different types of Support Vector Machine (SVM), Linear SVM, Quadratic SVM and Cubic SVM classifier are applied to classify the candidate dark regions as either hemorrhages or healthy. The efficacy of the proposed method is demonstrated using the standard benchmark DIARETDB1 database and by comparing the results with methods in silico. The performance of the method is measured based on average sensitivity, specificity, F-score and accuracy. Experimental results show the Linear SVM classifier gives better results than Cubic SVM and Quadratic SVM with respect to sensitivity and accuracy and with respect to specificity Quadratic SVM gives better result as compared to other SVMs.

A two-stage histogram equalization enhancement scheme for feature preservation in retinal fundus images

Early detection of ophthalmologic complications in the retinal fundus images is crucial and requires a quality image for accurate analysis and diagnosis. This often proves tricky due to the visual complexities of some retinal fundus images resulting in over-enhancement, non-uniform contrast, artificial boundaries, abrupt colour change, and other undesirable limitations in the performance of the existing image enhancement methods. In this paper, the uneven illumination and colour balance are globally adjusted using a histogram match. At the same time, the luminance component (V) of the HSV colour space is equalized with CLAHE for local contrast enhancement. Eight control parameter values are investigated to choose the most suitable values to restrict excessive enhancement and control the quality of the enhanced image. The performance of the proposed enhancement algorithm on the STARE, DRIVE, HRF, and DIARETDB1 databases shows that the method can prominently enhance colour retinal fundus images. The visual and quantitative comparisons of the proposed method with state-of-the-art techniques are remarkably outstanding. The statistical analysis test achieves a ρ-value far<0.05 justifying the statistical significance of the quantitative metrics and the high-quality performance of our enhancement framework.

Fractal analysis of retinal vasculature in relation with retinal diseases – an machine learning approach

Abstract Diabetic retinopathy (DR) is caused by diabetes mellitus. Vision loss occurs as a result of DR. The goal of this study was to use the DIARETDB-1, DIARETDB-0, STARE, MESSIDOR, E-ophtha-EX, and E-ophtha-MA databases to do Fourier fractal analysis and see how it is related to retinal illnesses. Following the extraction and inversion of colour channels, blood vessel augmentation was conducted. For the blood vessel enhanced image, the fractal dimension was determined. For DR patients and normal patients, measures such as standard deviation, mean, and significance were calculated. In the E-ophtha-EX database, significance was realized. In the DIARETDB-1, STARE, and DIARETDB-0 databases, the mean fractal value for normal patients is higher than for DR patients. The STARE database’s forecast of the association between fractal dimensions and various retinal disorders and the E-ophtha-EX database’s accomplishment of significance are the study’s main highlights. This study also improved the robustness of the blood vessel extraction there and increased the accuracy of its diagnosis.

Impact of activation functions and number of layers on detection of exudates using circular Hough transform and convolutional neural networks

Convolutional neural networks (CNN), which are used for object detection, are widely used in image recognition and segmentation fields due to their good performance. In this study, a method is proposed for the detection of exudate from DR lesions based on image processing and CNN model with different activation functions and layer numbers. Activation functions are one of the main reasons why the CNN model is complicated in hierarchical structure. CNN models have real artificial intelligence capabilities, especially thanks to non-linear activation functions. Although ReLU is the most commonly used activation function in CNN models, it has some shortcomings in practice. The most important shortcoming is that if the input value is negative, the learning process will slow down due to the inability to take the derivative. In order to solve this problem, the effect of the activation function in the CNN model on a real world problem and its image was investigated in this study. By applying circular Hough transform, one of the image processing methods, the OD region, which has a similar structure to the exuding regions, was determined and removed from the image, and then the detection of the exuding regions was carried out using the CNN model. Exudate detection was performed using ReLU, ELU, Leaky ReLU, Softplus and Swish activation functions in the CNN model and the results were compared. Experimental results in DiaretDB0 and DiaretDB1 databases show that the Swish activation function has a better performance than the others.

Fast and Robust Exudate Detection in Retinal Fundus Images Using Extreme Learning Machine Autoencoders and Modified KAZE Features.

Diabetic retinopathy(DR) is a health condition that affects the retinal blood vessels(BV) and arises in over half of people living with diabetes. Exudates(EX) are significant indications of DR. Early detection and treatment can prevent vision loss in many cases. EX detection is a challenging problem for ophthalmologists due to its different sizes and elevations as retinal fundus images frequently have irregular illumination and are poorly contrasting. Manual detection of EX is a time-consuming process to diagnose a mass number of diabetic patients. In the domain of signal processing, both SIFT (scale-invariant feature transform) and SURF (speed-up robust feature) methods are predominant in scale-invariant location retrieval and have shown a range of advantages. But, when extended to medical images with corresponding weak contrast between reference features and neighboring areas, these methods cannot differentiate significant features. Considering these, in this paper, a novel method is proposed based on modified KAZE features, which is an emerging technique to extract feature points and extreme learning machine autoencoders(ELMAE) for robust and fast localization of the EX in fundus images. The main stages of the proposed method are pre-processing, OD localization, dimensionality reduction using ELMAE, and EX localization. The proposed method is evaluated based on the freely accessible retinal database DIARETDB0, DIARETDB1, e-Ophtha, MESSIDOR, and local retinal database collected from Silchar Medical College and Hospital(SMCH). The sensitivity, specificity, and accuracy obtained by the proposed method are 96.5%, 96.4%, and 97%, respectively, with the processing time of 3.19 seconds per image. The results of this study are satisfactory with state-of-the-art methods. The results indicate that the approach taken can detect EX with less processing time and accurately from the fundus images.

ALGORITHM FOR THE IDENTIFICATION OF EXUDATES IN IMAGES OF THE HUMAN RETINA

Diabetic retinopathy is a complication of diabetes that affects the human eye. One of the very first signs of the disease is the appearance of a wound (exudate), which occurs due to the presence of changes in the blood vessels. Early detection of exudates will reduce the likelihood of developing blindness in patients with diabetic retinopathy. One of the problems in the process of detecting exudates is that the size of the object is quite small compared to the general image, so the task of detecting them is relevant at the present time. This article proposes an algorithm for identifying exudates in images of the human retina. This algorithm includes three main stages, namely: removal of the optic nerve head, determination of the location of the exudate and segmentation of the exudate. Optical disc removal is performed using the middle circle algorithm. In the step of detecting the location of the exudate, the image will be divided into smaller images and then classified into an exudate area and a non-exudate area based on the brightness threshold. Segmentation of the area of exudate is performed using the method of visibility. The study of the operation of the algorithm was carried out on a sample from the DIARETDB1 database by calculating the values of accuracy, sensitivity and specificity. The average accuracy of the proposed algorithm reached 99,33 %, the sensitivity and specificity of the proposed algorithm was 81,65 % and 99,42 %, respectively, which shows that the proposed method can segment exudate well.

Detection of Diabetic Retinopathy Using Discrete Wavelet Transform with Discrete Meyer in Retinal Images

One of the major complicated issues for extensive term diabetic aspirant is diabetic retinopathy (DR) which is an eye retinal syndrome, leads to blindness. The presence of exudates detects the disease, which can be prevented in the early stages by regular screening. Exudates can be automatically detected through inspecting digital retinal image. To detect the exudates for diagnosis the author proposed an algorithm called K-means Kernel support vector machine Radial basis function (KKR) approach, by the following main stages: extracting vessel and removal of optic disc followed by pre-processing, exudates detection and post processing. Wavelet dependent edge enhancement is used for dark portion separation of exudates in the retinal image by optically designed Wideband bandpass filter. Wavelet toolbox of MATLAB 2018a is used in this KKR algorithm. Statistical and structural texture features can be obtained using K-means segmentation process by integrating Local Binary Pattern (LBP) with Region Of Interest (ROI). Some features are selected and used Neural Network along with Radial Basis Function (RBF) to classify further. The KKR algorithm uses 80 fundus images from DIARETDB1 database and parameters are analyzed such as specificity, sensitivity and accuracy. The results obtained from proposed KKR algorithm have specificity of 81.57%, sensitivity of 87.56% and accuracy of 97.94% respectively.

Simple methods for the lesion detection and severity grading of diabetic retinopathy by image processing and transfer learning

Diabetic retinopathy (DR) has become one of the major causes of blindness. Due to the increased prevalence of diabetes worldwide, diabetic patients exhibit high probabilities of developing DR. There is a need to develop a labor-less computer-aided diagnosis system to support the clinical diagnosis. Here, we attempted to develop simple methods for severity grading and lesion detection from retinal fundus images. We developed a severity grading system for DR by transfer learning with a recent convolutional neural network called EfficientNet-B3 and the publicly available Kaggle Asia Pacific Tele-Ophthalmology Society (APTOS) 2019 training dataset, which includes artificial noise. After removing the blurred and duplicated images from the dataset using a numerical threshold, the trained model achieved specificity and sensitivity values ≳ 0.98 in the identification of DR retinas. For severity grading, the classification accuracy values of 0.84, 0.95, and 0.98 were recorded for the 1st, 2nd, and 3rd predicted labels, respectively. The utility of EfficientNets-B3 for the severity grading of DR as well as the detailed retinal areas referred were confirmed via visual explanation methods of convolutional neural networks. Lesion extraction was performed by applying an empirically defined threshold value to the enhanced retinal images. Although the extraction of blood vessels and detection of red lesions occurred simultaneously, the red and white lesions, including both soft and hard exudates, were clearly extracted. The detected lesion areas were further confirmed with ground truth using the DIARETDB1 database images with general accuracy. The simple and easily applicable methods proposed in this study will aid in the detection and severity grading of DR, which might help in the selection of appropriate treatment strategies for DR.

COMPUTER HYBRID SYSTEM OF HEMORRHAGE (HES) DETECTION USED FOR AIDED DIAGNOSIS OF DIABETIC RETINOPATHY

Diabetes cause’s metabolic and physiological abnormalities in the retina and the changes suggest a role for inflammation in the development of diabetic retinopathy. Abnormal blood vessels can form in the back of the eye of a person with diabetes. These new blood vessels are weaker and prone to breaking and causing hemorrhage (HEs). Diabetic retinopathy (DR) accounts for 31.5–54% of all cases of vitreous hemorrhage in adults in the world. Therefore, detection of HEs is still a challenging factor task for computer-aided diagnostics of DR. Many researchers have developed advanced algorithms of hemorrhages detection using fundus images. In this paper, a robust and computationally efficient approach for HEs with different shape and size detection and classification is presented. First, brightness correction and contrast enhancement are applied to fundus images. Second, candidate hemorrhages are extracted by using an unsupervised classification algorithm. Third, an approach based on mathematical morphology is carried out for vascular network and macula segmentation. Finally, a total of 13 HEs features are considered in this study and selected for classification. The proposed method is evaluated on 419 fundus images of DIARETDB0, DIARETDB1 and MESSIDOR databases. Experimental results show that overall average sensitivity, specificity, predictive value and accuracy for hemorrhage in lesion level are 98.90%, 99.66%, 97.63% and 99.56%, respectively. The results show that the proposed method outperforms other state-of-the-art methods in detection of hemorrhages. These results indicate that this new method may improve the performance of diagnosis of DR system.

Hemorrhage Detection Based on 3D CNN Deep Learning Framework and Feature Fusion for Evaluating Retinal Abnormality in Diabetic Patients.

Diabetic retinopathy (DR) is the main cause of blindness in diabetic patients. Early and accurate diagnosis can improve the analysis and prognosis of the disease. One of the earliest symptoms of DR are the hemorrhages in the retina. Therefore, we propose a new method for accurate hemorrhage detection from the retinal fundus images. First, the proposed method uses the modified contrast enhancement method to improve the edge details from the input retinal fundus images. In the second stage, a new convolutional neural network (CNN) architecture is proposed to detect hemorrhages. A modified pre-trained CNN model is used to extract features from the detected hemorrhages. In the third stage, all extracted feature vectors are fused using the convolutional sparse image decomposition method, and finally, the best features are selected by using the multi-logistic regression controlled entropy variance approach. The proposed method is evaluated on 1509 images from HRF, DRIVE, STARE, MESSIDOR, DIARETDB0, and DIARETDB1 databases and achieves the average accuracy of 97.71%, which is superior to the previous works. Moreover, the proposed hemorrhage detection system attains better performance, in terms of visual quality and quantitative analysis with high accuracy, in comparison with the state-of-the-art methods.

WITHDRAWN: An efficient detection of micro aneurysms from fundus images

Diabetic Retinopathy (DR) is used for elongated occurring Diabetes, and the foremost reason for visual damage is the discrepancies in the blood vessels of the retina. As the disease is asymptomatic, it is just being diagnosed through an ophthalmologist. Nevertheless, defensive and premature diagnosis is problematic owing to the absence of changes in terms of period and price. Throughout this phase, there is an injury on the retinal image such as Microaneurysms (MA), which is one of the initially noticed medical symptoms that specify the illness. Detection of MA in the early stage is one of the Challenging and an open issue in modern Literature. An Automatic System is introduced to assist the ophthalmologist for early detection. The retinal images need to pre-process prior to performing the segmentation using the clustering approach. The segmentation attained employing Enhanced Genetic K-Means (EGK-Means) Clustering algorithm from which the features are minimized through Support Vector Machines and, finally, the categorization of the retinal images is done through the usage of region descriptors employing Decision tree algorithm. The experimental results are carried out using DiaRetDB1 and Retinopathy Online Challenge (ROC) database, and it is observed that the specificity and sensitivity of the proposed methodology are higher when matched with the existing approaches.

Machine Learning for Diabetic Retinopathy Detection using Image Processing

The disorder of Diabetic Retinopathy (DR), a complication of Diabetes that may lead to blindness if not treated at an early stage, is diagnosed by evaluating the retina images of eye. However, the manual grading of images for identifying the seriousness of DR disease requires many resources and it also takes a lot of time. Automated systems give accurate results along with saving time. Ophthalmologists may find it useful in reducing their workload. Proposed work presents the method to correctly identify the lesions and classify DR images efficiently. Blood leaking out of veins form features such as exudates, microaneurysms and haemorrhages, on retina. Image processing techniques assist in DR detection. Median filtering is used on gray scale converted image to reduce noise. The features of the pre-processed images are extracted by textural feature analysis. Optic disc (OD) segmentation methodology is implemented for the removal of OD. Blood vessels are extracted using haar wavelet filters. KNN classifier is applied for classifying retinal image into diseased or healthy .The proposed algorithm is executed in MATLAB software and analyze results with regard to certain parameters such as accuracy, sensitivity, and specificity. The outcomes prove the superiority of the new method with sensitivity of 92.6%, specificity of 87.56% and accuracy of 95% on Diaretdb1 database.

Detection of Macula and Recognition of Aged-Related Macular Degeneration in Retinal Fundus Images

In aged people, the central vision is affected by Age-Related Macular Degeneration (AMD). From the digital retinal fundus images, AMD can be recognized because of the existence of Drusen, Choroidal Neovascularization (CNV), and Geographic Atrophy (GA). It is time-consuming and costly for the ophthalmologists to monitor fundus images. A monitoring system for automated digital fundus photography can reduce these problems. In this paper, we propose a new macula detection system based on contrast enhancement, top-hat transformation, and the modified Kirsch template method. Firstly, the retinal fundus image is processed through an image enhancement method so that the intensity distribution is improved for finer visualization. The contrast-enhanced image is further improved using the top-hat transformation function to make the intensities level differentiable between the macula and different sections of images. The retinal vessel is enhanced by employing the modified Kirsch's template method. It enhances the vasculature structures and suppresses the blob-like structures. Furthermore, the OTSU thresholding is used to segment out the dark regions and separate the vessel to extract the candidate regions. The dark region and the background estimated image are subtracted from the extracted blood vessels image to obtain the exact location of the macula. The proposed method applied on 1349 images of STARE, DRIVE, MESSIDOR, and DIARETDB1 databases and achieved the average sensitivity, specificity, accuracy, positive predicted value, F1 score, and area under curve of 97.79 %, 97.65 %, 97.60 %, 97.38 %, 97.57 %, and 96.97 %, respectively. Experimental results reveal that the proposed method attains better performance, in terms of visual quality and enriched quantitative analysis, in comparison with eminent state-of-the-art methods.

Optic disc and optic cup segmentation in retinal images

Automated segmentation of the optic disc (OD) and optic cup (OC) is important for retinal image analysis and retinal diabetic retinopathy systems. For OD segmentation, this paper presents a method done in three steps that combines variance and brightness features of the OD to localise leading to increased accuracy in detecting OD rather than using just one feature. As a first step, the image is divided into non-over lapping windows. Then the brightest window with maximum variance in intensity is selected. Subsequently the circular Hough transform (CHT) is applied to get the OD segmentation. OC segmentation is done in two steps: 1) blood vessels inside the OD are eliminated and; 2) restricted region growing performed to get the OC segmentation. The efficacy of the proposed method is demonstrated using the standard benchmark DRIVE and DIARETDB1 databases and by comparing the results of proposed method and some methods in silico.

Effective Fundus Image Decomposition for the Detection of Red Lesions and Hard Exudates to Aid in the Diagnosis of Diabetic Retinopathy.

Diabetic retinopathy (DR) is characterized by the presence of red lesions (RLs), such as microaneurysms and hemorrhages, and bright lesions, such as exudates (EXs). Early DR diagnosis is paramount to prevent serious sight damage. Computer-assisted diagnostic systems are based on the detection of those lesions through the analysis of fundus images. In this paper, a novel method is proposed for the automatic detection of RLs and EXs. As the main contribution, the fundus image was decomposed into various layers, including the lesion candidates, the reflective features of the retina, and the choroidal vasculature visible in tigroid retinas. We used a proprietary database containing 564 images, randomly divided into a training set and a test set, and the public database DiaretDB1 to verify the robustness of the algorithm. Lesion detection results were computed per pixel and per image. Using the proprietary database, 88.34% per-image accuracy (ACCi), 91.07% per-pixel positive predictive value (PPVp), and 85.25% per-pixel sensitivity (SEp) were reached for the detection of RLs. Using the public database, 90.16% ACCi, 96.26% PPV_p, and 84.79% SEp were obtained. As for the detection of EXs, 95.41% ACCi, 96.01% PPV_p, and 89.42% SE_p were reached with the proprietary database. Using the public database, 91.80% ACCi, 98.59% PPVp, and 91.65% SEp were obtained. The proposed method could be useful to aid in the diagnosis of DR, reducing the workload of specialists and improving the attention to diabetic patients.

An Integrated System for Detection Exudates and Severity Quantification for Diabetic Macular Edema

The prevalence of vision loss because of Diabetic macular edema (DME), which is a complication caused by diabetes mellitus, is increasing. Seven per cent of Indian diabetic population suffer from DME or damage to macula. The study proposes an integrated model to contribute to the prevention of loss of visual acuity. For detection of exudates and visualization of severity of DME image processing, machine learning and data analysis strategies are used. Experimentation on Indian population is done using Indian Diabetic Eye Diseases Dataset (IDEDD), acquired from metropolitan region of Mumbai and publicly available Indian Diabetic Retinopathy Image Dataset (IDRiD, Nanded) and a standard public database DIARETDB1- Standard Diabetic Retinopathy Database. To negate the effect of increase in the dimensionality and overfitting of the dataset parameters, a tree-based forward search algorithm for optimal feature subset identification is proposed. This provides global insight of effect of the statistical and spatial features on model behaviour. Machine learning algorithms of random forest and support vector machine are used for classification. The sensitivity and specificity of 97.1% and 100% is achieved for IDEDD data. Severity grading is done using image processing for segmenting the vessel mesh, optic disk, and macula from the abnormal retinal scans for detecting clinically significant macular edema by visualizing exudates. The distance between optic disc—fovea (OF) and the ratio of distance (OF) to that of disc diameter is formulated for localization of macula. Region-based and distance-based grading methods for quantification of the severity and visualization are implemented. The proposed algorithm enables the building of a predictive model for primordial and secondary prevention of DME with high classification accuracy and visualization of severity grading.

Automated segmentation of optic disc using statistical region merging and morphological operations.

Accurate Optic Disc (OD) segmentation is vital in designing systems that aid the diagnosis and evaluation of early phases of retinal diseases. However, in many images, the OD boundary is ambiguous, which makes the automated OD segmentation process very challenging. A method to segment OD based on statistical region merging and morphological operations is proposed in this paper. The proposed method is tested on standard databases MESSIDOR, DIARETDB1, DIARETDB0, and DRIONS-DB. The average overlap ratios are found to be 91.35% for DIARETDB1 images, 88.80% for DRIONS-DB images, 86.60% for DIARETDB0 images and 89.68% for MESSIDOR images, with average accuracies of 99.68%, 99.89%, 99.69%, and 99.93% respectively. A comparison with alternative methods showed that the proposed algorithm in OD segmentation is better than existing ones.

Microaneurysms detection in color fundus images using machine learning based on directional local contrast

BackgroundAs one of the majorcomplications of diabetes, diabetic retinopathy (DR) is a leadingcause of visual impairment and blindness due to delayed diagnosisand intervention. Microaneurysms appear as the earliest symptom ofDR. Accurate and reliable detection of microaneurysms in colorfundus images has great importance for DR screening.MethodsA microaneurysms' detection methodusing machine learning based on directional local contrast (DLC) isproposed for the early diagnosis of DR. First, blood vessels wereenhanced and segmented using improved enhancement function based onanalyzing eigenvalues of Hessian matrix. Next, with blood vesselsexcluded, microaneurysm candidate regions were obtained using shapecharacteristics and connected components analysis. After imagesegmented to patches, the features of each microaneurysm candidatepatch were extracted, and each candidate patch was classified intomicroaneurysm or non-microaneurysm. The main contributions of ourstudy are (1) making use of directional local contrast inmicroaneurysms' detection for the first time, which does make sensefor better microaneurysms' classification. (2) Applying threedifferent machine learning techniques for classification andcomparing their performance for microaneurysms' detection. Theproposed algorithm was trained and tested on e-ophtha MA database,and further tested on another independent DIARETDB1 database.Results of microaneurysms' detection on the two databases wereevaluated on lesion level and compared with existing algorithms.ResultsThe proposed method has achieved better performance compared with existing algorithms on accuracy and computation time. On e-ophtha MA and DIARETDB1 databases, the area under curve (AUC) of receiver operating characteristic (ROC) curve was 0.87 and 0.86, respectively. The free-response ROC (FROC) score on the two databases was 0.374 and 0.210, respectively. The computation time per image with resolution of 2544×1969, 1400×960 and 1500×1152 is 29 s, 3 s and 2.6 s, respectively.ConclusionsThe proposed methodusing machine learning based on directional local contrast of imagepatches can effectively detect microaneurysms in color fundus imagesand provide an effective scientific basis for early clinical DRdiagnosis.