Microaneurysms Research Articles

Automated Microaneurysms Detection and Classification using Multilevel Thresholding and Multilayer Perceptron

The purpose of this paper is to propose an automatic detection of microaneurysms (MAs) in the fundus retina images. In this work, E-optha database of 100 images were utilised to test the performance of the proposed method. The approach covers pre-processing, segmentation, post-processing, feature extraction and classification phases. In pre-processing, the images were filtered and the contrast enhanced. Then, the images were segmented using H-maxima and thresholding technique. Morphological operation was carried out to enhance the images before feature extraction and MAs candidate detection. The detected MAs candidates were classified using multilayer perceptron (MLP). After that, the detected MAs were classified into three classes including background (B), MAs and retinal blood vessels (RBVs). The performances of the classifiers were evaluated in terms of accuracy, sensitivity and specificity. The MLP classifier achieved a better performance than the support vector machine with the highest accuracy of 92.28% under condition 2. This study demonstrated a methodology for automatic detection of MAs using MLP. The proposed methodology successfully classify the MAs, B and RBVs and was reasonably fast to be implemented in real time.

Disparity of microaneurysm count between ultrawide field colour imaging and ultrawide field fluorescein angiography in eyes with diabetic retinopathy

AimsTo compare microaneurysm (MA) counts using ultrawide field colour images (UWF-CI) and ultrawide field fluorescein angiography (UWF-FA).MethodsRetrospective study including patients with type 1 or 2 diabetes mellitus receiving UWF-FA and...

Transfer Learning Based MA Detection TL-MAD

Diabetic Retinopathy (DR) is a microvascular complication of Diabetes that can lead to blindness if it is severe. Microaneurysm (MA) is the initial and main symptom of DR. In this paper, an automatic detection of DR from retinal fundus images of publicly available dataset has been proposed using transfer learning with pre-trained model VGG16 based on Convolutional Neural Network (CNN). Our method achieves improvement in accuracy for MA detection using retinal fundus images in prediction of Diabetic Retinopathy.

MultiColor imaging to detect different subtypes of retinal microaneurysms in diabetic retinopathy.

Retinal microaneurysms (MAs) are among the earliest signs of diabetic retinopathy (DR) and are typically detected by fluorescein angiography (FA). Confocal MultiColor is a noninvasive-imaging technique able to analyze different retinal features by capturing three simultaneous reflectance images. The main aim of the present study was to characterize morphological features of MAs by means of MultiColor images and to compare these with spectral domain optical coherence tomography (SD-OCT) and FA findings. A cross-sectional, observational study setting was chosen. Multimodal imaging included MultiColor, SD-OCT and FA images. We performed a qualitative analysis in order to assess the relationship between MultiColor and its green- and red-reflectance components, SD-OCT (hyperreflective, hyporeflective and mixed reflectivity) and FA findings. MAs detected on our MultiColor images were then categorized in accordance with a previously published histological classification. In our study FA images were used to detect 153 MAs in 30 eyes displaying DR. MultiColor was able to distinguish 122 MAs (80%). We identified green (16%), red (19%), and mixed (65%) MAs, corresponding to different reflectivity features detected by SD-OCT. MAs not visualized on MultiColor images corresponded to tiny hyperreflective lesions on SD-OCT. We compared our imaging findings with a histological MA classification reported in the literature. Our findings showed a strict relationship between MA subtypes and SD-OCT, suggesting that the composition of MAs (cells + endothelium + fibrosis) may influence the signal detected in MultiColor images. MultiColor appears to be a useful technique for investigating MA features in patients with DR.

OPTICAL COHERENCE TOMOGRAPHY ANGIOGRAPHY CHANGES AFTER SUBTHRESHOLD MICROPULSE YELLOW LASER IN DIABETIC MACULAR EDEMA.

To assess changes on optical coherence tomography (OCT) angiography in diabetic macular edema (DME) treated with subthreshold micropulse yellow laser (SMPL) over a period of 6 months. Thirty-five eyes (35 consecutive patients) with treatment-naive DME prospectively underwent (at baseline, 3 and 6 months) best-corrected visual acuity, swept-source OCT angiography/OCT, and fundus autofluorescence. Following parameters were evaluated on OCT angiography in the superficial capillary plexus (SCP) and deep capillary plexus (DCP): the area of foveal avascular zone, number of microaneurysms (MA), area of cysts, and presence of capillary network alterations. Microaneurysm change was also evaluated in 15 fellow eyes, not needing treatment over 6 months. Vessel and perfusion densities were evaluated in the SCP, DCP, and choriocapillaris, with image J. Retina thickness, number of hyperreflective retinal spots, and external limiting membrane integrity were evaluated on OCT. All measurements were performed by two masked graders, independently. All patients had diabetes mellitus Type 2 (mean age, 69.4 ± 10.9 years; duration of diabetes mellitus, 15.7 ± 8.7 years; and HbA1c 7.7 ± 1.2%). Mean best-corrected visual acuity at baseline was 69.7 ± 12.0 letters ETDRS, 72.7 ± 10.7 at 3 months (gain 3.1 ± 4.3, P = 0.0049) and 74.3 ± 9.5 at 6 months (gain 4.6 ± 7.2, P < 0.0001). Foveal avascular zone area decreased in the DCP at 6 months (P = 0.01). Area of cysts decreased in the SCP at 3 months and 6 months (P = 0.038; P = 0.049), and in the DCP at 6 months (P = 0.0071). Number of MA decreased at 6 months in the SCP (P = 0.0007) and at 3 months and 6 months in the DCP (P = 0.048; P < 0.0001) in treated eyes. No significant change in number of MA was found in nontreated eyes. There was no statistically significant change in any other OCT angiography/OCT parameter. Subthreshold micropulse yellow laser induces more pronounced changes in the DCP than in the SCP in DME. These changes occurred as early as 3 months after treatment. The evaluation of specific parameters in the DCP may help in determining treatment response.

Detection of Lesions for Diabetic Retinopathy By using Machine Learning Algorithms

In this paper existing writing for computer added diagnosis (CAD) based identification of lesions that might be connected in the early finding of Diabetic Retinopathy (DR) is talked about. The recognition of sores, for example, Microaneurysms (MA), Hemorrhages (HEM) and Exudates (EX) are incorporated in this paper. A range of methodologies starting from conventional morphology to deep learning techniques have been discussed. The different strategies like hand crafted feature extraction to automated CNN based component extraction, single lesion identification to multi sore recognition have been explored. The different stages in each methods beginning from the image preprocessing to classification stage are investigated. The exhibition of the proposed strategies are outlined by various performance measurement parameters and their used data sets are tabulated. Toward the end we examined the future headings.

Decrease in the number of microaneurysms in diabetic macular edema after anti-vascular endothelial growth factor therapy: implications for indocyanine green angiography-guided detection of refractory microaneurysms.

We evaluated changes in the numbers of microaneurysms (MAs) on fluorescein angiography (FA) and indocyanine green angiography (IA) in eyes with diabetic macular edema (DME) following intravitreal injection of anti-vascular endothelial growth factor (VEGF) agents. Twenty-one eyes of 16 patients with DME were included in this retrospective study. All patients received an initial loading dose of three monthly injections of anti-VEGF agents; thereafter, they received a pro re nata regimen for at least 12months of follow-up. FA and IA images were obtained before and at 6months after the initial injection. The median numbers of MAs significantly decreased from six (interquartile range [IQR] 3-7) MAs in early-phase FA, three (IQR 3-5) leaky MAs in late-phase FA, and two (IQR 1-4) MAs in late-phase IA at baseline to two (IQR 1-3) MAs in early-phase FA, one (IQR 0-2) leaky MA in late-phase FA, and one (IQR 0-2) MA in late-phase IA at 6months (P < 0.0001 for all). Only the median numbers of MAs in late-phase IA at baseline and at 6months were significantly higher in the recurrent DME group (13 eyes) than in the non-recurrent DME group (five eyes) (three [IQR 2-4] vs one [IQR 1-2], one [IQR 0.5-2] vs zero [P = 0.0185 and P = 0.009]). Intravitreal injection of anti-VEGF agents reduced the numbers of MAs in patients with DME. The numbers of MAs detected by late-phase IA might be useful predictors of DME recurrence.

A novel automated system of discriminating Microaneurysms in fundus images

Diabetic retinopathy, a chronic disease in diabetic patients leads to Vision loss, by disabling microvascular complications, if not recognized and cured at the earlier stage. This article explores a novel and reliable method for automatic early detection of Microaneurysms (MA) in fundus images. Microaneurysms characterized by small red spots on the retina, the red lesions are symptoms of early stage of DR. Development of an automated screening system would assist an ophthalmologist in diagnosing DR at an early stage. Hence, in this paper, a novel feature extraction technique using a Local Neighborhood Differential Coherence Pattern (LNDCP) is proposed. In this method, texture characteristics needed for classification by Feed Forward Neural Network (FFNN) is captured efficiently. The performance of the algorithm is validated using experiments on Retinopathy Online Challenge (ROC) public dataset and a single real-time dataset, AGAR300. Efficiency of the algorithm is benchmarked with state-of-art approaches and a Free-response Receiver Operating Characteristic (FROC) score of 0.481 and 0.442 have been achieved for ROC and AGAR300 respectively.

Improving Microaneurysm Detection from Non-dilated Diabetic Retinopathy Retinal Images using Feature Optimization

Diabetic retinopathy usually does not presents symptoms in an early stage until it gets to a severe stage. An early stage of diabetic retinopathy is associated with the presence of microaneurysms (MAs). The occurrence of blindness can be reduced significantly if MAs are detected. This paper presented an approach to improve automatic MAs detection using feature optimisation. Candidate MAs are detected using mathematic morphological techniques. Originally 20 features are presented. To verify the relevance of all original features, a feature optimisation process is performed. The optimal feature set is searched by a machine learning approach, like naive Bayes and support vector machine classifier. Hand-drawn ground-truth images from expert ophthalmologists are used to measure the performance evaluation. The results showed that the proposed optimal feature set could significantly improve MA detection.

Improving microaneurysm detection from non-dilated diabetic retinopathy retinal images using feature optimisation

Diabetic retinopathy usually does not presents symptoms in an early stage until it gets to a severe stage. An early stage of diabetic retinopathy is associated with the presence of microaneurysms (MAs). The occurrence of blindness can be reduced significantly if MAs are detected. This paper presented an approach to improve automatic MAs detection using feature optimisation. Candidate MAs are detected using mathematic morphological techniques. Originally 20 features are presented. To verify the relevance of all original features, a feature optimisation process is performed. The optimal feature set is searched by a machine learning approach, like naïve Bayes and support vector machine classifier. Hand-drawn ground-truth images from expert ophthalmologists are used to measure the performance evaluation. The results showed that the proposed optimal feature set could significantly improve MA detection.

A Fast Localization and Extraction of Microaneurysm for Early Detection of Diabetic Retionopathy

Diabetic Retinopathy (DR) is regarded as one of the leading causes of blindness globally. Microaneurysms (MAs) detection is essential to the computer aided diagnosis of DR at an early stage. However, the automatic detection of MAs is still a challenging problem as they are too tiny to be recognized and hard to be distinguished from other similar lesions. Therefore, we propose an efficient localization and extraction method for MAs, where the edge detection and Random Forest are utilized to enhance the accuracy of detection results. Finally, the proposed method is evaluated based on the public retinal image database MESSIDOR. Numerical results show that high accuracy and timely detection can be obtained with the proposed solution.

An Enhanced Residual U-Net for Microaneurysms and Exudates Segmentation in Fundus Images

Diabetic retinopathy (DR) is a leading cause of visual blindness. However if DR can be diagnosed and treated early, 90% of DR causing blindness can be prevented significantly. Microaneurysms (MAs) and exudates (EXs), as signs of DR, can be used for early DR diagnosis. However, MAs and EXs segmentation is a challenging task due to the low contrast of the lesions, the interference of noises, and the imbalance between the lesion areas and the background. In this paper, an enhanced residual U-Net (ERU-Net) for MAs and EXs segmentation is proposed. ERU-Net obtains three U-paths, which are composed by three upsampling paths together with one downsampling path. With such three U-paths structure, ERU-Net can enhance the corresponding features fusion and capture more details of fundus images. Also, a residual block is constructed in ERU-Net to extract more representative features. In the experiments, we evaluate the performance of ERU-Net for MAs and EXs segmentation on three public datasets, E-Ophtha, IDRiD, and DDR. The ERU-Net obtains the AUC values of 0.9956, 0.9962, 0.9801, 0.9866, 0.9679, 0.9609 for MAs and EXs segmentation on these three datasets, respectively, which are greater than that of the original U-Net. Compared with some traditional methods, convolutional neural networks and other recent U-Nets, ERU-Net also performs competitively. Besides, we have applied ERU-Net to segment optic disc (OD) on the DRISHTI-GS1 dataset, achieving the highest Jaccard index of 0.994 compared with the existing methods. The numerical results indicate that ERU-Net is a promising network for medical image segmentation.

Intelligent Automated Detection of Microaneurysms in Fundus Images Using Feature-Set Tuning

Due to the widespread of diabetes mellitus and its associated complications, a need for early detection of the leading symptoms in the masses is felt like never before. One of the earliest signs is the presence of microaneurysms (MAs) in the fundus images. This work presents a new technique for automatic detection of MAs in color fundus images. The proposed technique utilizes Genetic Programming (GP) and a set of 28 selected features from the preprocessed fundus images in order to evolve a mathematical expression. Through the binearization of the fitness scores, the optimal expression is evolved generation by generation through a stepwise enhancement process. The best expression is then used as a classifier for real world applications. Experimental results using three publically available datasets validate the usefulness of the proposed technique and its ability to outperform the state of the art contemporary approaches.

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.

Computer aided diabetic retinopathy detection based on ophthalmic photography: a systematic review and Meta-analysis.

To ensure the diagnostic value of computer aided techniques in diabetic retinopathy (DR) detection based on ophthalmic photography (OP). PubMed, EMBASE, Ei village, IEEE Xplore and Cochrane Library database were searched systematically for literatures about computer aided detection (CAD) in DR detection. The methodological quality of included studies was appraised by the Quality Assessment Tool for Diagnostic Accuracy Studies (QUADAS-2). Meta-DiSc was utilized and a random effects model was plotted to summarize data from those included studies. Summary receiver operating characteristic curves were selected to estimate the overall test performance. Subgroup analysis was used to identify the efficiency of CAD in detecting DR, exudates (EXs), microaneurysms (MAs) as well as hemorrhages (HMs), and neovascularizations (NVs). Publication bias was analyzed using STATA. Fourteen articles were finally included in this Meta-analysis after literature review. Pooled sensitivity and specificity were 90% (95%CI, 85%-94%) and 90% (95%CI, 80%-96%) respectively for CAD in DR detection. With regard to CAD in EXs detecting, pooled sensitivity, specificity were 89% (95%CI, 88%-90%) and 99% (95%CI, 99%-99%) respectively. In aspect of MAs and HMs detection, pooled sensitivity and specificity of CAD were 42% (95%CI, 41%-44%) and 93% (95%CI, 93%-93%) respectively. Besides, pooled sensitivity and specificity were 94% (95%CI, 89%-97%) and 87% (95%CI, 83%-90%) respectively for CAD in NVs detection. No potential publication bias was observed. CAD demonstrates overall high diagnostic accuracy for detecting DR and pathological lesions based on OP. Further prospective clinical trials are needed to prove such effect.

Classification of Diabetic Retinopathy Features using Bag of Feature Model

Deep learning (DL) as well as feature learning by unsupervised methods have made tremendous consideration in the past decades because of its great and dynamic capacity to change input data into high level depictions by means of various machine learning (ML) methods and approaches. Therefore these interests have also showed a fast and steady growth in the arena of medical image analysis, especially in Diabetic Retinopathy (DR) classification. On contradiction, manual interpretation involves excessive processing time, large amount of expertise and work. Sternness of the DR is analyzed relative to the existence of Microaneurysms (MAs), Exudates (EXs) and Hemorrhages(HEs). Spotting of DR in its early stage is crucial and important to avoid blindness. This paper proposes an algorithm to build an automated system to extract the above mentioned DR features which are the elemental and initial signs of diabetic retinopathy. Initial step in this algorithm is preprocessing of the original image. The next step in this features extraction algorithms is elimination of optic disc (OD) and blood vessels which have similar characteristic with these features. Blood vessels are segmented using Multi-Level Adaptive Thresholding. OD is segmented using morphological operations. Feature extraction and classification is achieved by using deep Bag of Feature (BoF) model which uses Speeded Up Robust Features Our method achieved 100% acuuracy in DRIVE database and over 90% accuracy for e-OPTHA database. Thus, the proposed methodology represents a track towards precise and highly automated DR diagnosis on a large substantial scale along with better sensitivity and specificity.

In vivo rotational three-dimensional OCTA analysis of microaneurysms in the human diabetic retina

The aim of this study was to explore whether rotational three-dimensional (3D) visualization of optical coherence tomography angiography (OCTA) volume data may yield valuable information regarding diabetic retinal microaneurysm (MA) characteristics. In this retrospective, observational study, we collected data from 20 patients (20 eyes) with diabetic retinopathy. Subjects were imaged with the SS-OCTA system (PLEX Elite 9000, Carl Zeiss Meditec Inc., Dublin, CA, USA). The OCTA volume data were processed with a volume projection removal algorithm and then exported to imageJ in order to obtain a 3D visualization of the analyzed MAs. The rotational three-dimensional OCTA images were qualitatively and quantitatively investigated. A total of 52 MAs were included in the analysis. On rotational 3D OCTA images, the number of vessels associated with each MA varied between 1 and 4, and most MAs (59.6%) were associated with 2 vessels. Moreover, in 20 MAs (38.4%) these vessels seem to originate from the SCP, while 26 MAs (50.0%) had associated vessels originating from the DVC, and 6 MAs had associated vessels arising from both the SCP and DVC (11.6%). Most MAs (31/52) had a ‘saccular’ shape. The number of retinal layers occupied by each MA ranged between 1 and 3 and the inner nuclear layer was the retinal layer most frequently occupied by MAs. In conclusion, this study used an algorithm to obtain rotational three-dimensional visualization of retinal MAs. The MAs’ architecture is complex and 3D visualization may clarify the true vascular origin of these lesions, which is often mistaken using en face OCTA images.

FILM: finding the location of microaneurysms on the retina.

Diabetes retinopathy (DR) is one of the leading cause of blindness among people suffering from diabetes. It is a lesion based disease which starts off as small red spots on the retina. These small red lesions are known as microaneurysms (MA). These microaneurysms gradually increase in size as the DR progresses, which eventually leads to blindness. Thus, DR can be prevented at a very early stage by eliminating the retinal microaneurysms. However, elimination of MA is a two step process. The first step requires detecting the presence of MA on the retina. The second step involves pinpointing the location of MA on the retina. Even though, these two steps are interdependent, there is no model available that can perform both steps simultaneously. Most of the models perform the first step successfully, while the second step is performed by opthamologists manually. Hence we have proposed an object detection model that integrates the two steps by detecting (first step) and pinpointing (second step) the MA on the retina simultaneously. This would help the opthamologists in directly finding the exact location of MA on the retina, thereby simplifying the process and eliminating any manual intervention.

Detection of Microaneurysms in Fundus Images Based on an Attention Mechanism.

Microaneurysms (MAs) are the earliest detectable diabetic retinopathy (DR) lesions. Thus, the ability to automatically detect MAs is critical for the early diagnosis of DR. However, achieving the accurate and reliable detection of MAs remains a significant challenge due to the size and complexity of retinal fundus images. Therefore, this paper presents a novel MA detection method based on a deep neural network with a multilayer attention mechanism for retinal fundus images. First, a series of equalization operations are performed to improve the quality of the fundus images. Then, based on the attention mechanism, multiple feature layers with obvious target features are fused to achieve preliminary MA detection. Finally, the spatial relationships between MAs and blood vessels are utilized to perform a secondary screening of the preliminary test results to obtain the final MA detection results. We evaluated the method on the IDRiD_VOC dataset, which was collected from the open IDRiD dataset. The results show that our method effectively improves the average accuracy and sensitivity of MA detection.

Microaneurysms Detection using Blob Analysis for Diabetic Retinopathy

Blob analysis is a mathematical method to find the region of interest (ROI) by focusing on the characteristics like brightness or colour. In this work, the process to segment Microaneurysms (MAs) involves two stages, which are pre-processing and segmentation. Pre-processing is a phase for noise removal and illumination correction. In this work, several methods were utilized namely Contrast Limited Adaptive Histogram Equalization (CLAHE), Normalization for contrast enhancement and median filter for noise removal. Then, continue with segmentation phase to segment the MAs from the image. In segmentation phase, several methods were used namely morphological opening, thresholding, Hessian Matrix 2D and Eigenvalue of Hessian Matrix. Finally all the resulting images were compared with the benchmark image to measure the accuracy and grading the stage of Diabetic Retinopathy (DR) by comparing the number of detected MAs. The segmentation accuracy of this project is 68% and 55% accuracy for stage grading