Detection Of Retinal Lesions Research Articles

Machine Teaching Allows for Rapid Development of Automated Systems for Retinal Lesion Detection From Small Image Datasets.

Machine teaching, a machine learning subfield, may allow for rapid development of artificial intelligence systems able to automatically identify emerging ocular biomarkers from small imaging datasets. We sought to use machine teaching to automatically identify retinal ischemic perivascular lesions (RIPLs) and subretinal drusenoid deposits (SDDs), two emerging ocular biomarkers of cardiovascular disease. IRB approval was obtained. Four small datasets of SD-OCT B-scans were used to train and test two distinct automated systems, one identifying RIPLs and the other identifying SDDs. An open-source interactive machine-learning software program, RootPainter, was used to perform annotation and training simultaneously over a 6-hour period. For SDDs at the B-scan level, test-set accuracy = 92%, sensitivity = 100%, specificity = 88%, positive predictive value (PPV) = 82%, and negative predictive value (NPV) = 100%. For RIPLs at the B-scan level, test-set accuracy = 90%, sensitivity = 60%, specificity = 93%, PPV = 50%, and NPV = 95%. Machine teaching demonstrates promise within ophthalmic imaging to rapidly allow for automated identification of novel biomarkers from small image datasets. [Ophthalmic Surg Lasers Imaging Retina 2024;55:XX-XX.].

Unraveling the complexity: deep learning for imbalanced retinal lesion detection and multi-disease identification

Unraveling the complexity: deep learning for imbalanced retinal lesion detection and multi-disease identification

Interest of Macular OCT in Severe Myopia: About 62 Cases

Myopic maculopathies are the leading cause of visual impairment. Clinically, they are characterized by a decrease in vision, most often at close range, sometimes associated with an impression of distorted vision. Numerous advances have recently been made in the clinical and therapeutic approach to this disease. Modern imaging examinations (optical coherence tomography, digital angiography) allow earlier detection of retinal lesions in myopic patients. A better understanding of the mechanisms leading to this complication may lead to new therapeutic perspectives.

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.

Intelligent Diagnosis of Multiple Peripheral Retinal Lesions in Ultra-widefield Fundus Images Based on Deep Learning.

Compared with traditional fundus examination techniques, ultra-widefield fundus (UWF) images provide 200° panoramic images of the retina, which allows better detection of peripheral retinal lesions. The advent of UWF provides effective solutions only for detection but still lacks efficient diagnostic capabilities. This study proposed a retinal lesion detection model to automatically locate and identify six relatively typical and high-incidence peripheral retinal lesions from UWF images which will enable early screening and rapid diagnosis. A total of 24,602 augmented ultra-widefield fundus images with labels corresponding to 6 peripheral retinal lesions and normal manifestation labelled by 5 ophthalmologists were included in this study. An object detection model named You Only Look Once X (YOLOX) was modified and trained to locate and classify the six peripheral retinal lesions including rhegmatogenous retinal detachment (RRD), retinal breaks (RB), white without pressure (WWOP), cystic retinal tuft (CRT), lattice degeneration (LD), and paving-stone degeneration (PSD). We applied coordinate attention block and generalized intersection over union (GIOU) loss to YOLOX and evaluated it for accuracy, sensitivity, specificity, precision, F1 score, and average precision (AP). This model was able to show the exact location and saliency map of the retinal lesions detected by the model thus contributing to efficient screening and diagnosis. The model reached an average accuracy of 96.64%, sensitivity of 87.97%, specificity of 98.04%, precision of 87.01%, F1 score of 87.39%, and mAP of 86.03% on test dataset 1 including 248 UWF images and reached an average accuracy of 95.04%, sensitivity of 83.90%, specificity of 96.70%, precision of 78.73%, F1 score of 81.96%, and mAP of 80.59% on external test dataset 2 including 586 UWF images, showing this system performs well in distinguishing the six peripheral retinal lesions. Focusing on peripheral retinal lesions, this work proposed a deep learning model, which automatically recognized multiple peripheral retinal lesions from UWF images and localized exact positions of lesions. Therefore, it has certain potential for early screening and intelligent diagnosis of peripheral retinal lesions.

DIABETIC RETINOPATHY LESION DETECTION FROM MULTISPECTRAL RETINAL IMAGES THROUGH NEURAL NETWORK

Diabetic Retinopathy (DR) is one of the fastest growing dysfunctions of human retina. Significant research has been conducted using RGB fundus imaging for automatic detection of retinal lesions affected by DR. However, due to only three imaging bands, the accuracy from RGB fundus images is unlikely to improve any further. In contrast to RGB imaging, multispectral imaging has the key advantage of multiple narrow wavelength bands that can be used as spectral features to improve the detection accuracy. Nevertheless, the inter and intra-retinal variation of color, contrast, and illumination is a challenge to process the multispectral images. In this study, a complete framework is proposed to develop and evaluate methods for automatic detection of DR lesions. A multispectral retinal image database, DIARETSPECDB1, is investigated in order to detect the most common DRs such as Microaneurysms (MA), Hard Exudates (HE) and Hemorrhages (HEM). The reflectance values of the spectral bands are used as features of a three-layer basic neural network (NN) to determine the baseline performance of multispectral data instead of any advanced model. According to the results, the model outperforms existing technique producing overall accuracy 94.5%, and the obtained specificity/sensitivity is 0.95/0.89, 0.97/0.89, and 0.88/0.84 for MAs, HEs and HEMs, respectively.

Retinal involvement of coronavirus disease (COVID-19): A systematic review

COVID-19 has wide-ranging manifestations which involve various organs. Although COVID-19 is known to affect mainly the respiratory system, many patients suffered from ocular manifestations following SARS-CoV-2 infection, especially conjunctivitis. However, there was some evidence of broader ocular involvement, which could involve the retinal layer. This review aims to analyze the possible explanation of retinal involvement in COVID-19 and whether comorbidities increase the risk of retinal involvement in COVID-19. We conducted a systematic literature search during September 2020 using the PUBMED database and other additional sources (e.g., Google Scholar). There was no year of publication nor language restriction. Six papers are used in this literature review: two studies reported retinal lesions in COVID-19 patients, one study detected SARS-CoV-2 RNA in retinal biopsies and three studies proposed possible mechanisms of retinal involvement in COVID-19. Ocular structures express ACE-2, such as the aqueous humor, pigmented epithelium, and retina. This is supported by the finding of SARS-CoV-2 nucleic acid in postmortem patients. Retinal involvement in COVID-19 patients includes hyperreflective retina lesions, subtle cotton wool spots, and microhemorrhage. However, these findings do not necessarily indicate the specificity of retinal involvement. Diabetes mellitus may be associated with disease severity in patients with COVID-19. Some evidence of retinal involvement in SARS-CoV-2 infection involves the detection of retinal lesions, the expression of ACE-2 in the retina, and the role of comorbidities.

Computerized retinal image analysis - a survey

The speedy development of digital imaging and computer vision has extended the potential of using these technologies in ophthalmology. Image processing systems are increasingly prominent in medical diagnostic systems and especially to modern ophthalmology. The retinal images give information about the health of the visual system. Retinal diseases, such as glaucoma, diabetic retinopathy, age-related macular degeneration, and many other diseases that can lead to blindness, manifest themselves in the retina. An automated system offers standardized large-scale screening at a lower cost, reduces human errors, and provides services to remote areas. Extensive research has been done since the last two decades in developing automated methods. Due to the fast evolution of new techniques, a comprehensive review is needed on such technique and algorithms present to date. This survey paper provides the reader a comprehensive review of the existing research in automated retinal image analysis. In this paper, automated computer aided methods used to diagnose retinal diseases have been reviewed. Several state-of-the art techniques and algorithms used to localize and segment features, such as optic disc and optic cup, macula and fovea, retinal blood vessels, detection of retinal lesions (microaneurysms, haemorrhages, exudates), are discussed and presented.

Computer Aided Diagnosis For Retinal Lesion Detection Based on Adaptive Boosting Classifier

Timely analysis and constant intensive care of patient’s pain since eye ailments must remained main worries in the computer-aided detection (CAD) ways and means. Intuiting sole or few precise styles of retinal wrongs has set an essential origination in computed-aid screen in the first few dates. Thus far, in line for to range of retinal wrongs and multilayered common useful productions, unplanned innovation of wounds with unrevealed and wide-ranging styles as retina leavings an exciting work. In this pattern, a weakly supervised practice, demanding just an order of common and not level retinal pictures underprivileged of vital to just luster their places and multiplicities, is helpful for this charge. Specifically, a eye copy is considered as a superposition of related, blood vessels and related noise (lesions embraced for unequal descriptions). Background is given as low-rank form followed by preprocessing steps, plus three-D plan, color regularization and blood vessels rejection. Related noise is eyed as stochastic changeable and reputable over Gaussian for classic imageries and mixture of Gaussian (MoG) for asymmetrical pictures, harmoniously. The coming run through codes together the related facts of fund us pictures and the related noise into on its own one and only ideal, and corporately enriches the ideal by typical and unequal pictures, which absolutely expose the low-rank subspace of the related and discriminate the wounds from related noise in unequal fundus pictures. Lastly, Ada boost classification process is castoff to increase sensing retinal injuries and to categorize the kinds of injury using a group of usual ones for exercise topographies. Investigational consequences prove that the future method is of well skills correctness and outdoes the preceding connected approaches.

Automated retinal lesion detection via image saliency analysis.

The detection of abnormalities such as lesions or leakage from retinal images is an important health informatics task for automated early diagnosis of diabetic and malarial retinopathy or other eye diseases, in order to prevent blindness and common systematic conditions. In this work, we propose a novel retinal lesion detection method by adapting the concepts of saliency. Retinal images are first segmented as superpixels, two new saliency feature representations: uniqueness and compactness, are then derived to represent the superpixels. The pixel level saliency is then estimated from these superpixel saliency values via a bilateral filter. These extracted saliency features form a matrix for low-rank analysis to achieve saliency detection. The precise contour of a lesion is finally extracted from the generated saliency map after removing confounding structures such as blood vessels, the optic disk, and the fovea. The main novelty of this method is that it is an effective tool for detecting different abnormalities at the pixel level from different modalities of retinal images, without the need to tune parameters. To evaluate its effectiveness, we have applied our method to seven public datasets of diabetic and malarial retinopathy with four different types of lesions: exudate, hemorrhage, microaneurysms, and leakage. The evaluation was undertaken at the pixel level, lesion level, or image level according to ground truth availability in these datasets. The experimental results show that the proposed method outperforms existing state-of-the-art ones in applicability, effectiveness, and accuracy.

A new model for retinal lesion detection of diabetic retinopathy using hierarchical self-organizing maps

Diabetes is a disease that impairs blood flow throughout the body. In this disease, the retinal blood vessels may leak and cause retinal swelling known as edema. The person’s sight might be affected if this swelling happens in the central vision area of retina, the macula. In this paper, we proposed a classification system, including a novel combination of Self-Organizing Maps (SOM) for detecting retinal lesions. The proposed system consists of a fast pre-processing step followed by lesion feature extraction and, finally, a detailed classification model. In the pre-processing stage, the system is divided into the three procedures of initial target lesion extraction, optical disk extraction, and eventually extracting retinal blood vessels from the retina. The second step is a combination of multiple features such as morphology, color, intensity, and moments. The classifier is a model of Hierarchical Self-Organizing Maps (HSOM), which aims to increase the accuracy and speed of classifying the lesions while considering the high amount of data in extracting the features. The overall accuracy and sensitivity of the proposed method according to the MESSIDOR database is 97.87% and 98.51%, respectively. The results show that the proposed model can detect and classify the Lesions in HDR images accurately.

Diabetic Retinopathy Detection Using Walter-Klein Contrast Enhancement

When sugar level (glucose) in the blood fails to regulate the insulin properly in human body, diabetic is occurred. The effect of diabetic on eye causes diabetic retinopathy. Diabetic retinopathy (DR) is a serious eye disease that originates from diabetes mellitus and is the most common cause of blindness in the developed countries. Therefore, much effort has been made to establish reliable computer aided screening systems based on color fund us images. Diabetic Retinopathy is one of a complicated diabetes which can cause blindness. It is a metabolic disordered patients perceive no symptoms until the disease is at late stage. So early detection and proper treatment has to be ensured. To serve this purpose, various automated systems have been designed. We propose an ensemble-based framework for retinal lesion detection. Unlike the well-known approach of considering the output of multiple classifiers, we propose a combination of of Retinal Lesion detectors, namely preprocessing methods and candidate extractors. The presence of micro aneurysms in the eye is one of the early signs of diabetic retinopathy. We analzye the input retinal images of the Diabetic patients and we can classify that the patient is affected by DR or not. If not affected, they are normal patient. If they are affected, further it classifies the different stages of diabetic retinopathy affected patients such as Mild, Moderate and Severe.

Automatic Detection of Retinal Lesions for Screening of Diabetic Retinopathy.

Diabetic retinopathy (DR) is characterized by the progressive deterioration of retina with the appearance of different types of lesions that include microaneurysms, hemorrhages, exudates, etc. Detection of these lesions plays a significant role for early diagnosis of DR. To this aim, this paper proposes a novel and automated lesion detection scheme, which consists of the four main steps: vessel extraction and optic disc removal, preprocessing, candidate lesion detection, and postprocessing. The optic disc and the blood vessels are suppressed first to facilitate further processing. Curvelet-based edge enhancement is done to separate out the dark lesions from the poorly illuminated retinal background, while the contrast between the bright lesions and the background is enhanced through an optimally designed wideband bandpass filter. The mutual information of the maximum matched filter response and the maximum Laplacian of Gaussian response are then jointly maximized. Differential evolution algorithm is used to determine the optimal values for the parameters of the fuzzy functions that determine the thresholds of segmenting the candidate regions. Morphology-based postprocessing is finally applied to exclude the falsely detected candidate pixels. Extensive simulations on different publicly available databases highlight an improved performance over the existing methods with an average accuracy of and robustness in detecting the various types of DR lesions irrespective of their intrinsic properties.

Red Lesion Detection Using Dynamic Shape Features for Diabetic Retinopathy Screening.

The development of an automatic telemedicine system for computer-aided screening and grading of diabetic retinopathy depends on reliable detection of retinal lesions in fundus images. In this paper, a novel method for automatic detection of both microaneurysms and hemorrhages in color fundus images is described and validated. The main contribution is a new set of shape features, called Dynamic Shape Features, that do not require precise segmentation of the regions to be classified. These features represent the evolution of the shape during image flooding and allow to discriminate between lesions and vessel segments. The method is validated per-lesion and per-image using six databases, four of which are publicly available. It proves to be robust with respect to variability in image resolution, quality and acquisition system. On the Retinopathy Online Challenge's database, the method achieves a FROC score of 0.420 which ranks it fourth. On the Messidor database, when detecting images with diabetic retinopathy, the proposed method achieves an area under the ROC curve of 0.899, comparable to the score of human experts, and it outperforms state-of-the-art approaches.

An algorithm combining two lesion‐detection methods of retinal microaneurysms for the reduction of human workload

PurposeReduction of workload in the detection of microaneurysms (MA) from retinal photographs is crucial for the diagnosis and screening of diabetic retinopathy. Automatic algorithms for the detection of retinal lesions can help reduce human intervention especially when the lesions are present in large numbers.MethodsTwo methods for lesion detection were combined in a single algorithm, one based on the analyses of the contrast between dark peak‐points and surrounding circular regions, and a second one based on the correlation between the intensity values in the photographs and a MA‐template. The two individual methods and the two methods combined were tested separately to compare their performance on retinal images from 26 high‐risk patients.ResultsBoth individual lesion‐detection methods missed clustered MAs. With the exclusion of grouped lesions, the two methods combined showed higher sensitivity and precision than the contrast and template methods alone, identifying 22% and 13% more lesions respectively.ConclusionsThe combination of the two methods can provide repeatable detection of unclustered MAs in photographs from high‐risk patients. Manual intervention is only required to select grouped MAs and to adjust the automatic results, considerably reducing human workload.

Computer aided quantification for retinal lesions in patients with moderate and severe non-proliferative diabetic retinopathy: a retrospective cohort study.

BackgroundDetection of retinal lesions like micro-aneurysms and exudates are important for the clinical diagnosis of diabetes retinopathy. The traditional subjective judgments by clinicians are dependent on their experience and can be subject to lack of consistency and therefore a quantification method is worthwhile.MethodsIn this study, 10 moderate non-proliferative diabetes retinopathy (NPDR) patients and 10 severe NPDR ones were retrospectively selected as a cohort. Mathematical morphological methods were used for automatic segmentation of lesions. For exudates detection, images were pre-processed with adaptive histogram equalization to enhance contrast, then binary images for area calculation were obtained by threshold classification. For micro-aneurysms detection, the images were pre-processed by top-hat and bottom-hat transformation, then Otsu method and Hough transform were used to classify micro-aneurysms. Post-processing morphological methods were used to preclude the false positive noise.ResultsAfter segmentation, the area of exuduates divided by optic disk area (exudates/disk ratio) and counts of microaneurysms were quantified and compared between the moderate and severe non-proliferative diabetic retinopathy groups, which had significant difference(P < 0.05).ConclusionsIn conclusion, morphological features of lesion might be an image marker for NPDR grading and computer aided quantification of retinal lesion could be a practical way for clinicians to better investigates diabetic retinopathy.

Detection and classification of retinal lesions for grading of diabetic retinopathy

Diabetic Retinopathy (DR) is an eye abnormality in which the human retina is affected due to an increasing amount of insulin in blood. The early detection and diagnosis of DR is vital to save the vision of diabetes patients. The early signs of DR which appear on the surface of the retina are microaneurysms, haemorrhages, and exudates. In this paper, we propose a system consisting of a novel hybrid classifier for the detection of retinal lesions. The proposed system consists of preprocessing, extraction of candidate lesions, feature set formulation, and classification. In preprocessing, the system eliminates background pixels and extracts the blood vessels and optic disc from the digital retinal image. The candidate lesion detection phase extracts, using filter banks, all regions which may possibly have any type of lesion. A feature set based on different descriptors, such as shape, intensity, and statistics, is formulated for each possible candidate region: this further helps in classifying that region. This paper presents an extension of the m-Mediods based modeling approach, and combines it with a Gaussian Mixture Model in an ensemble to form a hybrid classifier to improve the accuracy of the classification. The proposed system is assessed using standard fundus image databases with the help of performance parameters, such as, sensitivity, specificity, accuracy, and the Receiver Operating Characteristics curves for statistical analysis.

Comparison of image-assisted versus traditional fundus examination.

The purpose of this study was to evaluate the ability of image-assisted fundus examination to detect retinal lesions compared with traditional fundus examination. Subjects were imaged using a nonmydriatic ultrawide field scanning laser ophthalmoscope. After imaging, subjects underwent both a traditional and an image-assisted fundus examination, in random order. During the image-assisted method, ultrawide field scanning laser ophthalmoscopic images were reviewed in conjunction with a dilated fundus examination. Lesions detected by each method were assigned to one of three regions, ie, optic disc, posterior pole/macula, or mid-to-peripheral retina. Discrepancies between the image-assisted and the traditional examination methods were adjudicated by a retinal ophthalmologist. In total, 170 subjects (339 eyes) were recruited. Agreement between image-assisted and traditional fundus examination varied by lesion type and was excellent for staphyloma (kappa 0.76), fair for suspicious cupping (kappa 0.66), drusen in the posterior pole/macula and mid-to-peripheral retina (0.45, 0.41), retinal pigment epithelial changes in the posterior pole/macula (0.54), peripheral retinal degeneration (0.50), cobblestone (0.69), vitreoretinal interface abnormalities (0.40), and vitreous lesions (0.53). Agreement was poor for hemorrhage in the mid-to-peripheral retina (kappa 0.33), and nevi in the mid-to-peripheral retina (0.34). When the methods disagreed, the results indicated a statistically significant advantage for the image-assisted examination in detecting suspicious cupping (P = 0.04), drusen in the posterior pole/macula and mid-to-peripheral retina (P = 0.004, P < 0.001), retinal pigment epithelial changes in the posterior pole/macula (P = 0.04), nevi in the posterior pole/macula and mid-to-peripheral retina (P = 0.01, P = 0.007), peripheral retinal degeneration (P < 0.001), hemorrhage in the mid-to-peripheral retina (P = 0.01), and vitreous lesions (P < 0.001). Image-assisted fundus examination may enhance detection of retinal lesions compared with traditional fundus examination alone.

Wide-Field Fundus Imaging Using the Optos Optomap and a Disposable Eyelid Speculum

The EzSpec (a disposable eyelid speculum recently designed for intravitreal injections) is a useful device for removing eyelash artifacts and for potentially increasing the sensitivity of the Optos Optomap system, thereby improving the detection of retinal lesions.

Detection of retinal lesions in diabetic retinopathy: comparative evaluation of 7-field digital color photography versus red-free photography

Red-free light allows better detection of vascular lesions as this wavelength is absorbed by hemoglobin; however, the current gold standard for the detection and grading of diabetic retinopathy remains 7-field color fundus photography. The goal of this study was to compare the ability of 7-field fundus photography using red-free light to detect retinopathy lesions with corresponding images captured using standard 7-field color photography. Non-stereoscopic standard 7-field 30° digital color fundus photography and 7-field 30° digital red-free fundus photography were performed in 200 eyes of 103 patients with various grades of diabetic retinopathy ranging from mild to moderate non-proliferative diabetic retinopathy to proliferative diabetic retinopathy. The color images (n=1,400) were studied with corresponding red-free images (n=1,400) by one retina consultant (PV) and two senior residents training in retina. The various retinal lesions [microaneurysms, hemorrhages, hard exudates, soft exudates, intra-retinal microvascular anomalies (IRMA), neovascularization of the retina elsewhere (NVE), and neovascularization of the disc (NVD)] detected by all three observers in each of the photographs were noted followed by determination of agreement scores using κ values (range 0-1). Kappa coefficient was categorized as poor (≤0), slight (0.01-0.20), fair (0.2 -0.40), moderate (0.41-0.60), substantial (0.61-0.80), and almost perfect (0.81-1). The number of lesions detected by red-free images alone was higher for all observers and all abnormalities except hard exudates. Detection of IRMA was especially higher for all observers with red-free images. Between image pairs, there was substantial agreement for detection of hard exudates (average κ=0.62, range 0.60-0.65) and moderate agreement for detection of hemorrhages (average κ=0.52, range 0.45-0.58), soft exudates (average κ=0.51, range 0.42-0.61), NVE (average κ=0.47, range 0.39-0.53), and NVD (average κ=0.51, range 0.45-0.54). Fair agreement was noted for detection of microaneurysms (average κ=0.29, range 0.20-0.39) and IRMA (average κ=0.23, range 0.23-0.24). Inter-observer agreement with color images was substantial for hemorrhages (average κ=0.72), soft exudates (average κ=0.65), and NVD (average κ=0.65); moderate for microaneurysms (average κ=0.42), NVE (average κ=0.44), and hard exudates (average κ=0.59) and fair for IRMA (average κ=0.21). Inter-observer agreement with red-free images was substantial for hard exudates (average κ=0.63) and moderate for detection of hemorrhages (average κ=0.56), SE (average κ=0.60), IRMA (average κ=0.50), NVE (average κ=0.44), and NVD (average κ=0.45). Digital red-free photography has a higher level of detection ability for all retinal lesions of diabetic retinopathy. More advanced grades of retinopathy are likely to be detected earlier with red-free imaging because of its better ability to detect IRMA, NVE, and NVD. Red-free monochromatic imaging of the retina is a more effective and less costly alternative for detection of vision-threatening diabetic retinopathy.