The Role of Medical Image Modalities and AI in the Early Detection, Diagnosis and Grading of Retinal Diseases: A Survey

Diabetic retinopathy (DR), retinal vein occlusion (RVO), and age-related macular degeneration (AMD) pose significant global health challenges, often resulting in vision impairment and blindness. Automatic detection of these conditions is crucial, particularly in underserved rural areas with limited access to ophthalmic services. Despite remarkable advancements in artificial intelligence, especially convolutional neural networks (CNNs), their complexity can make interpretation difficult. In this study, we curated a dataset consisting of 15,089 color fundus photographs (CFPs) obtained from 8110 patients who underwent fundus fluorescein angiography (FFA) examination. The primary objective was to construct integrated models that merge CNNs with an attention mechanism. These models were designed for a hierarchical multilabel classification task, focusing on the detection of DR, RVO, AMD, and other fundus conditions. Furthermore, our approach extended to the detailed classification of DR, RVO, and AMD according to their respective subclasses. We employed a methodology that entails the translation of diagnostic information obtained from FFA results into CFPs. Our investigation focused on evaluating the models’ ability to achieve precise diagnoses solely based on CFPs. Remarkably, our models showcased improvements across diverse fundus conditions, with the ConvNeXt-base + attention model standing out for its exceptional performance. The ConvNeXt-base + attention model achieved remarkable metrics, including an area under the receiver operating characteristic curve (AUC) of 0.943, a referable F1 score of 0.870, and a Cohen’s kappa of 0.778 for DR detection. For RVO, it attained an AUC of 0.960, a referable F1 score of 0.854, and a Cohen’s kappa of 0.819. Furthermore, in AMD detection, the model achieved an AUC of 0.959, an F1 score of 0.727, and a Cohen’s kappa of 0.686. Impressively, the model demonstrated proficiency in subclassifying RVO and AMD, showcasing commendable sensitivity and specificity. Moreover, our models enhanced interpretability by visualizing attention weights on fundus images, aiding in the identification of disease findings. These outcomes underscore the substantial impact of our models in advancing the detection of DR, RVO, and AMD, offering the potential for improved patient outcomes and positively influencing the healthcare landscape.

The rapid advancement of computational infrastructure has led to unprecedented growth in machine learning, deep learning, and computer vision, fundamentally transforming the analysis of retinal images. By utilizing a wide array of visual cues extracted from retinal fundus images, sophisticated artificial intelligence models have been developed to diagnose various retinal disorders. This paper concentrates on the detection of Age-Related Macular Degeneration (AMD), a significant retinal condition, by offering an exhaustive examination of recent machine learning and deep learning methodologies. Additionally, it discusses potential obstacles and constraints associated with implementing this technology in the field of ophthalmology. Through a systematic review, this research aims to assess the efficacy of machine learning and deep learning techniques in discerning AMD from different modalities as they have shown promise in the field of AMD and retinal disorders diagnosis. Organized around prevalent datasets and imaging techniques, the paper initially outlines assessment criteria, image preprocessing methodologies, and learning frameworks before conducting a thorough investigation of diverse approaches for AMD detection. Drawing insights from the analysis of more than 30 selected studies, the conclusion underscores current research trajectories, major challenges, and future prospects in AMD diagnosis, providing a valuable resource for both scholars and practitioners in the domain.

Age-related macular ARM degeneration is an eye disease, that gradually degrades the macula, a part of the retina, which is responsible for central vision. It occurs in one of the two types, dry and wet age-related macular degeneration. The purpose of this paper is to diagnose the retinal disease age-related macular degeneration. An automated approach is proposed to help in the early detection of age-related macular degeneration using three models and their performances are compared. The amount of the disease spread in the retina can be identified by extracting the features of the retina. Detection of age-related macular degeneration disease has been done using probabilistic neural network PNN, Bayesian classification and support vector machine SVM and the two types of age-related macular degeneration are classified and diagnosed successfully. The results show that SVM achieves a higher performance measure than probabilistic neural network and Bayes classification.

The most common retinal diseases that are to be diagnosed are Diabetic Retinopathy (DR), Age-related Macular Degeneration (AMD) and Choroidal Neovascularization (CNV). For the people above 60 years of age, detection of these retinal diseases is an important task for treatment that reduces the risk of vision loss. Retinal fundus images play a significant role in the detection of DR, AMD and CNV disease diagnosis and treatment. The existing techniques for the detection of DR, AMD and CNV have not fulfilled with the classification accuracy of the retinal diseases effectively. This research work proposes an efficient classification framework for retinal fundus image recognition to overcome these drawbacks. Initially, the input image from the publicly available STARE database is preprocessed with the following three steps (a) Specular reflection removal and smoothing, (b) contrast enhancement and (c) retinal region expansion. With the preprocessed image, the features are extracted using Multi-Scale Discriminative Robust Local Binary Pattern (MS-DRLBP), based on RGB component selection, Gradient operation, and LBP descriptor. Finally, classification was done using hybrid Convolution Neural Network (CNN) and Radial Basis Function (RBF) model (CNN-RBF) which classifies the retinal fundus images into four classes such as DR, AMD, CNV and Normal (NR). Experimental results of the proposed method gives an accuracy of 97.22% compared with the existing other methodologies.

Diabetes mellitus is a complex metabolic disease that has become a global epidemic with more than 285 million cases worldwide. Major medical advances over the past decades have substantially improved its management, extending patients' survival. The latter is accompanied by an increased risk of developing chronic macro- and microvascular complications. Amongst them, diabetic retinopathy (DR) is the most common and frightening. Furthermore, during the past two decades, it has become the leading cause of visual loss. Irrespective of the type of diabetes, DR follows a well-known clinical and temporal course characterized by pericytes and neuronal cell loss, formation of acellular-occluded capillaries, occasional microaneurysms, increased leucostasis and thickening of the vascular basement membrane. These alterations progressively affect the integrity of retinal microvessels, leading to the breakdown of the blood-retinal barrier, widespread haemorrhage and neovascularization. Finally, tractional retinal detachment occurs leading to blindness. Nowadays, there is growing evidence that local inflammation and oxidative stress play pivotal roles in the pathogenesis of DR. Both processes have been associated with pericytes and neuronal degeneration observed early during DR progression. They may also be linked to sustained retinal vasculature damage that results in abnormal neovascularization. Currently, DR therapeutic options depend on highly invasive surgical procedures performed only at advanced stages of the disease, and which have proved to be ineffective to restore visual acuity. Therefore, the availability of less invasive and more effective strategies aimed to prevent or delay the onset of DR is highly desirable. Multipotent mesenchymal stromal cells, also referred to as mesenchymal stem cells (MSCs), are promising healing agents as they contribute to tissue regeneration by pleiotropic mechanisms, with no evidence of significant adverse events. Here, we revise the pathophysiology of DR to identify therapeutic targets for donor MSCs. Also, we discuss whether an MSC-based therapy could prevent or delay the onset of DR.

Purpose Studies regarding the relationship between age-related macular degeneration (AMD) and diabetic retinopathy (DR) conflict: while some support that AMD is protective against DR, others find the opposite. The mechanism by which AMD may protect against DR is unclear. We sought to assess the association between AMD and DR when controlling for glycemic control in patients with diabetes mellitus (DM) type II. Methods We identified 461 unique patients over 55 years old with a diagnosis of DM type II seen in our academic retina clinic in Stony Brook, New York between 12/31/2019 and 12/31/2020. Data were manually extracted and the population was split based on the presence of AMD diagnosis. Multivariate regression analyses were then performed comparing the prevalence of DR between groups while controlling for A1c. Secondary endpoints included demographic differences and smoking status. Results Among the 461 patients, 118 (25.6%) had a diagnosis of AMD. Compared to patients without AMD, patients with AMD were older (69 vs. 66; OR 1.05; p=0.005) and less likely to have DR (37.3% vs. 59.2%; OR 0.35; p<0.001). There was no difference in average A1c between groups. Conclusion This is the first reported study assessing the relationship between AMD and DR while controlling for A1c. In our population, diagnosis of AMD was associated with significantly reduced odds of having DR. As AMD and DR appear to be related even after holding A1c constant, researchers should use caution when using DR as a surrogate for diabetic control in the context of AMD.

ObjectivesThe aim of this work was to study the role of fundus fluorescein angiography and spectral domain optical coherence tomography (SDOCT) in the diagnosis of wet age-related macular degeneration (AMD).BackgroundAMD is the leading cause of severe visual impairment in elderly people. Various imaging methods are available for the diagnosis and the classification of AMD. Fluorescein angiography (FA) is the gold standard for the differential diagnosis of neovascular AMD and determination of lesion characteristics. SDOCT is used for the diagnosis and the follow-up of patients with neovascular AMD undergoing antivascular endothelial growth factor therapy, by providing high-resolution cross-sectional images of retinal pathology.Participants and methodsThis study included 30 patients divided into three groups: group Me, patients with dry AMD; group II, patients with wet AMD divided into classic choroidal neovascularization (CNV) and occult CNV; group III, patients with geographic atrophy. All participants underwent a full ophthalmologic evaluation, including dilated fundus examination by binocular indirect ophthalmoscope. Slit-lamp biomicroscopy was performed using the Goldman three mirror lens and Volk 78 lens; also, the best-corrected visual acuity was measured with the Landolt chart and converted to the logarithm of the minimum angle of resolution (logMAR) units, and fundus fluorescein angiography and then optical coherence tomography (OCT) were performed.ResultsUsing SDOCT imaging to delineate the lesion morphology, early AMD was detected by the presence of a normal foveal contour and minimal alteration in the macular area, classic CNV by the presence of a well-defined lesion with steep margins and a crater like configuration, and occult CNV by the presence of an ill-defined, flat lesion with a convex surface. FA-OCT overlay images provided a significant correlation between FA patterns and OCT features such as retinal pigment epithelium (RPE) complex changes. The sensitivity of cases with dry AMD in FA to cases with undulation in OCT was 100%, the specificity was 92%, the accuracy was 93%, the positive predictive value (PPV) was 75%, and the negative predictive value (NPV) was 100%. The sensitivity of cases with wet AMD in FA to cases with thickening in OCT was 96%, the specificity was 20%, the accuracy was 83%, the PPV was 86%, and the NPV was 50%. The sensitivity of cases with geographic atrophy of AMD in FA to cases with thinning in OCT was 100%, the specificity was 97%, the accuracy was 97%, the PPV was 50%, and the NPV was 100%.ConclusionSDOCT is highly sensitive for identifying AMD, CNV, and CNV activity, but cannot fully replace FA in the diagnosis of AMD.

Age-related Macular degeneration (AMD) is the third leading cause of incurable acute central vision loss. Optical coherence tomography (OCT) is a diagnostic process used for both AMD and diabetic macular edema (DME) detection. Spectral-domain OCT (SD-OCT), an improvement of traditional OCT, has revolutionized assessing AMD for its high acquiring rate, high efficiency, and resolution. To detect AMD from normal OCT scans many techniques have been adopted. Automatic detection of AMD has become popular recently. The use of a deep Convolutional Neural Network (CNN) has helped its cause vastly. Despite having achieved better performance, CNN models are often criticized for not giving any justification in decision-making. In this paper, we aim to visualize and critically analyze the decision of CNNs in context-based AMD detection. Multiple experiments were done using the DUKE OCT dataset, utilizing transfer learning in Resnet50 and Vgg16 model. After training the model for AMD detection, Gradient-weighted Class Activation Mapping (Grad-Cam) is used for feature visualization. With the feature mapped image, each layer mask was compared. We have found out that the Outer Nuclear layer to the Inner segment myeloid (ONL-ISM) has more predominance about 17.13% for normal and 6.64% for AMD in decision making.

Age-related Macular Degeneration (AMD) is the predominant cause of blindness in developed countries, specially in elderly people. Moreover, its prevalence is increasing due to the global population ageing. In this scenario, early detection is crucial to avert later vision impairment. Nonetheless, implementing large-scale screening programmes is usually not viable, since the population at-risk is large and the analysis must be performed by expert clinicians. Also, the diagnosis of AMD is considered to be particularly difficult, as it is characterized by many different lesions that, in many cases, resemble those of other macular diseases. To overcome these issues, several works have proposed automatic methods for the detection of AMD in retinography images, the most widely used modality for the screening of the disease. Nowadays, most of these works use Convolutional Neural Networks (CNNs) for the binary classification of images into AMD and non-AMD classes. In this work, we propose a novel approach based on CNNs that simultaneously performs AMD diagnosis and the classification of its potential lesions. This latter secondary task has not yet been addressed in this domain, and provides complementary useful information that improves the diagnosis performance and helps understanding the decision. A CNN model is trained using retinography images with image-level labels for both AMD and lesion presence, which are relatively easy to obtain. The experiments conducted in several public datasets show that the proposed approach improves the detection of AMD, while achieving satisfactory results in the identification of most lesions.

Cardiovascular Risk and Eye Health: A Cohort Study of the Pooled Cohort Equations and Ocular Disease Incidence.

Promising Artificial Intelligence-Machine Learning-Deep Learning Algorithms in Ophthalmology.

What Is the Issue?&#x0D; &#x0D; Wet age-related macular degeneration (AMD) is a serious health concern that causes reduced vision-related function, poor overall quality of life, and increased health care resource usage.&#x0D; Anti-vascular endothelial growth factor (VEGF) drugs are the first-line treatment for wet AMD, and they are injected into the eye. These drugs can slow disease progression but are expensive and require trained medical professionals to administer and monitor treatment response.&#x0D; The criteria for initial injection, maintenance, and discontinuation of anti-VEGF drugs in patients with wet AMD remain unclear.&#x0D; &#x0D; What Did We Do?&#x0D; &#x0D; To inform decisions about the appropriate use of anti-VEGF drugs to treat adults with wet AMD, CADTH sought to identify and summarize recommendations about criteria for the diagnosis of wet AMD and best practices for subsequent treatment with anti-VEGF drugs.&#x0D; An information specialist searched the peer-reviewed and grey literature for clinical practice guidelines published since 2018.&#x0D; &#x0D; What Did We Find?&#x0D; &#x0D; Three evidence-based guidelines recommend optical coherence tomography (OCT) to assess patients with suspected wet AMD. The National Institute for Health and Care Excellence (NICE) guidelines then strongly recommend fundus fluorescein angiography only when OCT does not exclude neovascular disease.&#x0D; The evidence-based guidelines recommend anti-VEGF drugs for wet or neovascular AMD in adults, particularly for patients with recent disease progression. One guideline suggests starting anti-VEGF treatment as soon as possible once the diagnosis of wet AMD is made.&#x0D; The NICE guideline suggests that there may not be any clinically significant distinctions in the effectiveness and safety of various anti-VEGF treatments, including aflibercept, bevacizumab, and ranibizumab, indicating a need for an individualized approach when selecting specific anti-VEGF drugs. However, 1 guideline is inconsistent with NICE, as it suggests choosing bevacizumab first.&#x0D; During the maintenance phase of using anti-VEGF drugs, it is important to closely monitor the patient's response and the disease activity and adjust the injection intervals accordingly.&#x0D; Guidelines recommend taking an individualized approach and emphasizing active patient involvement in switching and discontinuing anti-VEGF drugs for adults with wet AMD.&#x0D; Guidelines or guidance documents without a literature search or unclear methods provided generally consistent recommendations with the included evidence-based guidelines.&#x0D; We did not identify any evidence-based guidelines regarding the use of faricimab or brolucizumab. However, NICE published technology appraisals supporting the use of recently approved anti-VEGF drugs (faricimab and brolucizumab) as alternative options for patients who meet similar criteria for treatment with bevacizumab or aflibercept, and under agreed pricing conditions.&#x0D; &#x0D; What Does It Mean?&#x0D; &#x0D; All identified guidelines consistently recommend anti-VEGF drugs for adults with wet active AMD, with some providing guidance on the use of OCT and other ophthalmological assessments to identify criteria for starting and adjusting treatment.&#x0D; These guidelines recommend personalized and patient-centred approaches for selecting, switching, and discontinuing anti-VEGF medications. Cost is also a crucial factor in decision-making, especially for aflibercept, ranibizumab, and recently approved anti-VEGF drugs such as faricimab and brolucizumab.&#x0D; Further research and evidence-based guidelines are necessary to confirm whether newer drugs can be used in similar situations as older drugs.&#x0D;

Long-Term Metformin Use and Reduced Risk of Age-Related Macular Degeneration: A Large Database Study.

Retina Update: Entering an Era of Personalized Medicine in Ophthalmology.

The development of intravitreal drugs and intraocular implants