Enhancing the ophthalmic AI assessment with a fundus image quality classifier using local and global attention mechanisms.

Background: The assessment of image quality (IQA) plays a pivotal role in the realm of image-based computer-aided diagnosis techniques, with fundus imaging standing as the primary method for the screening and diagnosis of ophthalmic diseases. Conventional studies on fundus IQA tend to rely on simplistic datasets for evaluation, predominantly focusing on either local or global information, rather than a synthesis of both. Moreover, the interpretability of these studies often lacks compelling evidence. In order to address these issues, this study introduces the Local and Global Attention Aggregated Deep Neural Network (LGAANet), an innovative approach that integrates both local and global information for enhanced analysis. Methods: The LGAANet was developed and validated using a Multi-Source Heterogeneous Fundus (MSHF) database, encompassing a diverse collection of images. This dataset includes 802 color fundus photography (CFP) images (302 from portable cameras), and 500 ultrawide-field (UWF) images from 904 patients with diabetic retinopathy (DR) and glaucoma, as well as healthy individuals. The assessment of image quality was meticulously carried out by a trio of ophthalmologists, leveraging the human visual system as a benchmark. Furthermore, the model employs attention mechanisms and saliency maps to bolster its interpretability. Results: In testing with the CFP dataset, LGAANet demonstrated remarkable accuracy in three critical dimensions of image quality (illumination, clarity and contrast based on the characteristics of human visual system, and indicates the potential aspects to improve the image quality), recording scores of 0.947, 0.924, and 0.947, respectively. Similarly, when applied to the UWF dataset, the model achieved accuracies of 0.889, 0.913, and 0.923, respectively. These results underscore the efficacy of LGAANet in distinguishing between varying degrees of image quality with high precision. Conclusion: To our knowledge, LGAANet represents the inaugural algorithm trained on an MSHF dataset specifically for fundus IQA, marking a significant milestone in the advancement of computer-aided diagnosis in ophthalmology. This research significantly contributes to the field, offering a novel methodology for the assessment and interpretation of fundus images in the detection and diagnosis of ocular diseases.

Fine-Tuning SSL-Model to Enhance Detection of Cilioretinal Arteries on Colored Fundus Images.

Cilioretinal arteries are a common congenital anomaly of retinal blood supply. This paper presents a deep learning-based approach for the automated detection of a CRA from color fundus images. Leveraging the Vision Transformer architecture, a pre-trained model from RETFound was fine-tuned to transfer knowledge from a broader dataset to our specific task. An initial dataset of 85 was expanded to 170 images through data augmentation using self-supervised learning-driven techniques. To address the imbalance in the dataset and prevent overfitting, Focal Loss and Early Stopping were implemented. The model's performance was evaluated using a 70-30 split of the dataset for training and validation. The results showcase the potential of ophthalmic foundation models in enhancing detection of CRAs and reducing the effort required for labeling by retinal experts, as promising results could be achieved with only a small amount of training data through fine-tuning.

Hyperautofluorescent material inside areas of macular atrophy may reveal non-lipofuscin fluorophores in late stage AMD.

PURPOSE: To characterize fundus autofluorescence (FAF) in complete (cRORA) and incomplete retinal pigment epithelium and outer retinal atrophy (iRORA) by fluorescence lifetime imaging ophthalmology (FLIO). METHODS: Overall, 98 macular atrophy (MA) lesions in 42 eyes of 37 age-related macular degeneration (AMD) patients (mean age: 80.9 ± 5.8 years), 25 of them classified as iRORA and 73 as cRORA by OCT, were investigated by FLIO in a short (SSC: 498-560 nm) and a long wavelength channel (LSC: 560-720 nm). Differences of FAF lifetimes and peak emission wavelength (PEW) between atrophic lesions and intact retinal pigment epithelium (RPE) in the outer ring of the ETDRS grid were considered. RESULTS: FAF lifetimes in MA were longer and PEW were significantly (p < 0.001) shorter than in intact RPE by 112 ± 78 ps (SSC), 91 ± 64 ps (LSC), 27 ± 18 nm (PEW) in iRORA and by 227 ± 112 ps (SSC), 167 ± 81 ps (LSC), and 54 ± 17 nm (PEW) in cRORA. 37% of iRORA and 24% of cRORA were hyperautofluorescent in SSC. Persistent sub-RPE-BL material in MA was newly found as a hyperautofluorescent entity with lifetimes considerably longer than that of drusen and RPE. CONCLUSIONS: Despite RPE and, thus, lipofuscin are greatly absent in MA, considerable FAF, preferably at short wavelengths, was found in those lesions. Drusen, persistent sub-RPE-BL material, basal laminar deposits, persistent activated RPE, and sclera were identified as putative sources of this fluorescence. FLIO can help to characterize respective fluorophores.

The addition of pembrolizumab to neoadjuvant chemoradiotherapy did not increase the risk of developing postoperative anastomotic leakage for ESCC: an analysis from a prospective cohort.

BACKGROUND: To compare the difference of postoperative anastomotic leakage (AL) rate between neoadjuvant chemoradiotherapy (NCRT) with pembrolizumab and NCRT group, and investigate the risk factors of developing AL for locally advanced esophageal squamous cell cancer (ESCC). MATERIALS AND METHODS: The GF was contoured on the pretreatment planning computed tomography and dosimetric parameters were retrospectively calculated. Univariate and multivariate logistic regression analysis was performed to determine the independent risk predictors for the entire cohort. A nomogram risk prediction model for postoperative AL was established. RESULTS: A total of 160 ESCC patients were included for analysis. Of them, 112 were treated with NCRT with pembrolizumab and 44 patients with NCRT. Seventeen (10.6%) patients experienced postoperative AL with a rate of 10.7% (12/112) in NCRT with pembrolizumab and 11.4% (5/44) in NCRT group. For the entire cohort, mean, D50, Dmax, V5, V10 and V20 GF dose were statistically higher in those with AL (all p < 0.05). Multivariate logistic regression analysis indicated that tumor length (p = 0.012), volume of GF (p = 0.003) and mean dose of GF (p = 0.007) were independently predictors for postoperative AL. Using receiver operating characteristics analysis, the mean dose limit on the GF was defined as 14 Gy. CONCLUSION: Based on our prospective database, no significant difference of developing AL were observed between NCRT with pembrolizumab and NCRT group. We established an individualized nomograms based on mean GF dose combined with clinical indicators to predict AL in the early postoperative period.

Categories and Profiles of Uveitis in Vogt-Koyanagi-Harada Syndrome With Systemic Correlation: Inferences From a Tertiary Multispecialty Hospital.

Introduction Vogt-Koyanagi-Harada (VKH) syndrome is a granulomatous, autoimmune panuveitis, affecting the eyes, ears, skin, and meninges. It can cause choroiditis and can progress to the retina and optic disc causing visual loss. Imaging using fundus fluorescein angiography (FFA), indocyanine green angiography (ICGA), and enhanced depth imaging-ocular coherence tomography (EDI-OCT) is required for clinical evaluation and management. Steroids and immunosuppression are the treatment modalities used. Aim The aim of this study is to report the correlation and severity of uveitis in relation to systemic manifestations. Method A retrospective study including 100 patients with VKH syndrome was carried out. They were classified based on clinical manifestations and investigations such as FFA, ICGA, B-scan ultrasonography (USG), and ocular coherence tomography (OCT). Patients were characterized as complete, incomplete, and probable VKH syndrome. Laboratory investigations were performed, and statistical analysis was done. Results Probable VKH syndrome was found to be the most common form of presentation in our study population. Defective vision was the most common complaint among the patients. Extraocular manifestations included tinnitus, vertigo, alopecia, headache, fatigue, and vitiligo and were seen in 33% of the patients. Disc edema and serous retinal detachment were seen in 85% of the patients. Improvement was noted in 25% of the patients with the use of corticosteroids. Conclusion Response to treatment with systemic corticosteroids and immunosuppression in the acute phase of uveitis is better compared to chronic uveitis. The ophthalmologist is usually first consulted in VKH syndrome due to presenting ocular complaints. A multidisciplinary approach is key to providing holistic management.

American Society of Retina Specialists Artificial Intelligence Task Force Report.

Since the Artificial Intelligence Committee of the American Society of Retina Specialists developed the initial task force report in 2020, the artificial intelligence (AI) field has seen further adoption of US Food and Drug Administration-approved AI platforms and significant development of AI for various retinal conditions. With expansion of this technology comes further areas of challenges, including the data sources used in AI, the democracy of AI, commercialization, bias, and the need for provider education on the technology of AI. The overall focus of this committee report is to explore these recent issues as they relate to the continued development of AI and its integration into ophthalmology and retinal practice.

Diabetic retinopathy screening with confocal fundus camera and artificial intelligence - assisted grading.

PURPOSE: Screening for diabetic retinopathy (DR) by ophthalmologists is costly and labour-intensive. Artificial Intelligence (AI) for automated DR detection could be a clinically and economically alternative. We assessed the performance of a confocal fundus imaging system (DRSplus, Centervue SpA), coupled with an AI algorithm (RetCAD, Thirona B.V.) in a real-world setting. METHODS: 45° non-mydriatic retinal images from 506 patients with diabetes were graded both by an ophthalmologist and by the AI algorithm, according to the International Clinical Diabetic Retinopathy severity scale. Less than moderate retinopathy (DR scores 0, 1) was defined as non-referable, while more severe stages were defined as referable retinopathy. The gradings were then compared both at eye-level and patient-level. Key metrics included sensitivity, specificity all measured with a 95% Confidence Interval. RESULTS: The percentage of ungradable eyes according to the AI was 2.58%. The performances of the AI algorithm for detecting referable DR were 97.18% sensitivity, 93.73% specificity at eye-level and 98.70% sensitivity and 91.06% specificity at patient-level. CONCLUSIONS: DRSplus paired with RetCAD represents a reliable DR screening solution in a real-world setting. The high sensitivity of the system ensures that almost all patients requiring medical attention for DR are referred to an ophthalmologist for further evaluation.

Evaluating the potential of retinal photography in chronic kidney disease detection: a review.

Background: Chronic kidney disease (CKD) is a significant global health concern, emphasizing the necessity of early detection to facilitate prompt clinical intervention. Leveraging the unique ability of the retina to offer insights into systemic vascular health, it emerges as an interesting, non-invasive option for early CKD detection. Integrating this approach with existing invasive methods could provide a comprehensive understanding of patient health, enhancing diagnostic accuracy and treatment effectiveness. Objectives: The purpose of this review is to critically assess the potential of retinal imaging to serve as a diagnostic tool for CKD detection based on retinal vascular changes. The review tracks the evolution from conventional manual evaluations to the latest state-of-the-art in deep learning. Survey Methodology: A comprehensive examination of the literature was carried out, using targeted database searches and a three-step methodology for article evaluation: identification, screening, and inclusion based on Prisma guidelines. Priority was given to unique and new research concerning the detection of CKD with retinal imaging. A total of 70 publications from 457 that were initially discovered satisfied our inclusion criteria and were thus subjected to analysis. Out of the 70 studies included, 35 investigated the correlation between diabetic retinopathy and CKD, 23 centered on the detection of CKD via retinal imaging, and four attempted to automate the detection through the combination of artificial intelligence and retinal imaging. Results: Significant retinal features such as arteriolar narrowing, venular widening, specific retinopathy markers (like microaneurysms, hemorrhages, and exudates), and changes in arteriovenous ratio (AVR) have shown strong correlations with CKD progression. We also found that the combination of deep learning with retinal imaging for CKD detection could provide a very promising pathway. Accordingly, leveraging retinal imaging through this technique is expected to enhance the precision and prognostic capacity of the CKD detection system, offering a non-invasive diagnostic alternative that could transform patient care practices. Conclusion: In summary, retinal imaging holds high potential as a diagnostic tool for CKD because it is non-invasive, facilitates early detection through observable microvascular changes, offers predictive insights into renal health, and, when paired with deep learning algorithms, enhances the accuracy and effectiveness of CKD screening.

Evaluation of sensitivity and specificity of the slit-light method in the diagnosis of cyclotorsion.

PURPOSE: The purpose of this study was to evaluate the amount of sensitivity and specificity of the slit-light (SL) method in the diagnosis of ocular cyclotorsion. MATERIALS AND METHODS: One hundred and twenty eyes of 60 individuals (10-60 years old), with mean visual acuity of 0.08 ± 0.14 LogMAR, were divided into two groups (normal and torsion groups). Individuals without ocular motility disorder were selected as normal and patients with extraocular motility disorders and oblique muscle dysfunctions as the torsion group. The sensitivity and specificity of SL in the diagnosis of ocular torsion were measured by masked investigators and compared to fundus photography (FP). Inter- and intraobserver variability of these techniques was also determined. RESULTS: The amounts of sensitivity and specificity of SL, measured by the first examiner, were 60% and 92% for intorsion and 50% and 96% for extorsion assessment, respectively. These amounts were 53% and 95% for intorsion, and 54% and 97% for extorsion by the second examiner. The contingency coefficient between the two examiners was 68.6% for SL. This amount was 61% between FP and SL for the first examiner and 63% for the second. The contingency coefficient for the repeatability of SL was 72.2% for the first examiner and 75.7% for the second. This amount was 71.2% between the two examiners. CONCLUSION: SL can be considered a useful method for the diagnosis of cyclotorsion.

Cross-sectional Analysis of Outer Retinal Tubulation in Inherited Retinal Diseases: A Multicenter Study.

PURPOSE: This study aims to explore genetic variants that potentially lead to outer retinal tubulation (ORT), estimate the prevalence of ORT in these candidate genes, and investigate the clinical etiology of ORT in patients with inherited retinal diseases (IRDs), with respect to each gene. DESIGN: Retrospective cohort study. METHODS: A retrospective cross-sectional review was conducted on 565 patients with molecular diagnoses of IRD, confirming the presence of ORT as noted in each patient's respective spectral-domain optical coherence tomography (SD-OCT) imaging. Using SD-OCT imaging, the presence of ORT was analyzed in relation to specific genetic variants and phenotypic characteristics. Outcomes included the observed ORT frequencies across two gene-specific cohorts: non- retinal pigment epithelium (RPE)-specific genes, and RPE-specific genes; and to investigate the analogous characteristics caused by variants in these genes. RESULTS: Among the 565 patients included in this study, 104 exhibited ORT on SD-OCT. We observed ORT frequencies among the following genes from our patient cohort: 100% (23/23) forCHM, 100%(2/2) forPNPLA6, 100% (4/4) forRCBTB1, 100% formtDNA[100% (4/4) forMT-TL1and 100% (1/1) formtDNAdeletion], 100% (1/1) forOAT, 95.2% (20/21) forCYP4V2, 72.7% (8/11) forCHMfemale carriers, 66.7% (2/3) forC1QTNF5, 57.1% (8/14) forPROM1, 53.8% (7/13) forPRPH2, 42.9% (3/7) forCERKL, 28.6% (2/7) forCDHR1, 20% (1/5) forRPE65, 4% (18/445) forABCA4.In contrast, ORT was not observed in any patients with photoreceptor-specific gene variants, such asRHO(n=13),USH2A(n=118),EYS(n=70),PDE6B(n=10),PDE6A(n=4),and others. CONCLUSION: These results illustrate a compelling association between the presence of ORT and IRDs caused by variants in RPE-specific genes, as well as non-RPE-specific genes. In contrast, IRDs caused by photoreceptor-specific genes are typically not associated with ORT occurrence. Further analysis revealed that ORT tends to manifest in IRDs with milder intraretinal pigment migration (IPM), a finding that is typically associated with RPE-specific genes. These findings regarding ORT, genetic factors, atrophic patterns in the fundus, and IPM provide valuable insight into the complex etiology of IRDs. Future prospective studies are needed to further explore the association and underlying mechanisms of ORT in these contexts.

Predictive value of blood coagulation and routine blood indices for rebleeding after endoscopic treatment in hepatitis B-related cirrhotic patients with esophagogastric fundal varices: a logistic regression model analysis.

OBJECTIVE: To evaluate the predictive value of blood coagulation and routine blood indices for rebleeding after endoscopic treatment of ruptured esophagogastric fundal varices (EGVB) in cirrhotic patients with hepatitis B infection. METHODS: This retrospective analysis included 248 patients with hepatitis B-related cirrhosis and EGVB who received initial endoscopic treatment from October 2019 to March 2022 and were followed up for 12 months. Patients were divided into rebleeding and non-rebleeding groups. Laboratory indices were analyzed, and univariate and multivariate analyses identified predictors of rebleeding. The efficacy of a logistic regression model was evaluated using Receiver Operating Characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA), and a risk factor nomogram was constructed for assessing the predictive efficiency of those risk factors. RESULTS: Univariate analysis showed significant differences in portal vein diameters and lower Child-Pugh scores in the rebleeding group in contrast to those in the non-rebleeding group. Key laboratory markers such as platelet count (PLT), albumin (ALB), alanine aminotransferase (ALT), lymphocytes (LYM), and prognostic nutritional index (PNI) were lower, while prothrombin time (PT) and lactate levels (LN) were higher in the rebleeding group than those in the non-rebleeding group. Multivariate analysis identified portal vein diameter, PLT, ALT, PT, LYM, and PNI as significant predictors of rebleeding. The logistic model demonstrated high accuracy (AUC=0.986) and clinical value, validated by ROC curves, calibration curves (C-index =0.986), and DCA results. A risk factor predictive nomogram was successfully constructed. CONCLUSION: This study developed a logistic regression model with a nomogram for predicting EGVB-related rebleeding in patients with hepatitis B-related cirrhosis, achieving an AUC of 0.986, indicating high accuracy and significant clinical relevance.

Infrared live imaging and navigated laser for nematode photocoagulation in a child with diffuse unilateral subacute neuroretinitis (DUSN).

Purpose: To report the application of an infrared fundus imaging and navigated laser system to photocoagulate a nematode in diffuse unilateral subacute neuroretinitis (DUSN). Observations: A 14-year-old boy with DUSN was treated with systemic albendazole and corticosteroids. Laser photocoagulation of the visible nematode was performed using a navigated laser in live infrared fundus view (Navilas 577s, OD-OS GmbH, Berlin, Germany). While the localization of the nematode was difficult in regular fundoscopy due to the light-shy helminth, it could be well localized and targeted with the infrared live video mode and navigated laser system. No inflammatory flare-up was observed after the nematode was killed. Conclusions and Importance: Laser photocoagulation and systemic antihelminthic therapy are an established treatment for DUSN. Infrared imaging and navigated laser systems seem useful in targeting and killing mobile nematodes.

Ocular fundus changes and association with systemic conditions in systemic lupus erythematosus.

Systemic lupus erythematosus (SLE) is an autoimmune disease that affects multiple organs and systems. Ocular involvement is estimated to manifest in one-third of individuals with SLE, of which lupus retinopathy and choroidopathy represent the severe subtype accompanied by vision impairment. Advancements in multimodal ophthalmic imaging have allowed ophthalmologists to reveal subclinical microvascular and structural changes in fundus of patients with SLE without ocular manifestations. Both ocular manifestations and subclinical fundus damage have been shown to correlate with SLE disease activity and, in some patients, even precede other systemic injuries as the first presentation of SLE. Moreover, ocular fundus might serve as a window into the state of systemic vasculitis in patients with SLE. Given the similarities of the anatomy, physiological and pathological processes shared among ocular fundus, and other vital organ damage in SLE, such as kidney and brain, it is assumed that ocular fundus involvement has implications in the diagnosis and evaluation of other systemic impairments. Therefore, evaluating the fundus characteristics of patients with SLE not only contributes to the early diagnosis and intervention of potential vision damage, but also holds considerate significance for the evaluation of SLE vasculitis state and prediction of other systemic injuries.

Multi-step framework for glaucoma diagnosis in retinal fundus images using deep learning.

Glaucoma is one of the most common causes of blindness in the world. Screening glaucoma from retinal fundus images based on deep learning is a common method at present. In the diagnosis of glaucoma based on deep learning, the blood vessels within the optic disc interfere with the diagnosis, and there is also some pathological information outside the optic disc in fundus images. Therefore, integrating the original fundus image with the vessel-removed optic disc image can improve diagnostic efficiency. In this paper, we propose a novel multi-step framework named MSGC-CNN that can better diagnose glaucoma. In the framework, (1) we combine glaucoma pathological knowledge with deep learning model, fuse the features of original fundus image and optic disc region in which the interference of blood vessel is specifically removed by U-Net, and make glaucoma diagnosis based on the fused features. (2) Aiming at the characteristics of glaucoma fundus images, such as small amount of data, high resolution, and rich feature information, we design a new feature extraction network RA-ResNet and combined it with transfer learning. In order to verify our method, we conduct binary classification experiments on three public datasets, Drishti-GS, RIM-ONE-R3, and ACRIMA, with accuracy of 92.01%, 93.75%, and 97.87%. The results demonstrate a significant improvement over earlier results.

Using multi-modal imaging to refine the phenotype of PRPH2-associated retinal degeneration.

OBJECTIVE: To refine retinal PRPH2-associated retinal degeneration (PARD) phenotypes using multimodal imaging. DESIGN: Retrospective review of clinical records and multimodal imaging. SUBJECTS: Patients who visited the inherited retinal degeneration (IRD) clinic at two tertiary referral eye centers with molecularly confirmed IRD due to PRPH2 variants. METHODS: Retinal imaging was reviewed using ultra-widefield (UWF) pseudocolor, UWF fundus autofluorescence (FAF), and spectral-domain optical coherence tomography (SD-OCT). Phenotypes were identified in the macular or peripheral region. A combined phenotype was considered if any phenotypes were present in both macular and peripheral regions. Mixed phenotypes in the macula or peripheral retina were considered if there were two distinct phenotypes identified in the same eye. The presence or absence of atrophy in the macular or peripheral area was also noted. MAIN OUTCOME MEASURE: Grading of multimodal imaging by phenotype and atrophy. RESULTS: A total of 144 eyes of 72 patients were included in this study. The majority of the eyes had combined macular and peripheral phenotypes (89/14, 61.8%), while 44 (30.6%) eyes had isolated macular findings, and 11 (7.6%) had isolated peripheral findings. Twenty-five eyes were classified with mixed macular phenotypes while fundus flavimaculatus dystrophy type was the most common combined macular and peripheral phenotype (54/144, 37.5%): n = 10 with macular dystrophy and macular flavimaculatus dystrophy, and n = 15 with butterfly pattern dystrophy and macular flavimaculatus dystrophy. Nearly half of the eyes (71/144, 49.3%) were identified to have concomitant outer retinal atrophy. Fundus flavimaculatus type dystrophy was also associated with the highest proportion of concomitant atrophy (57/71, 80.3%). CONCLUSION: PARD demonstrates a wide array of phenotypes using multimodal imaging. We report that combinations of classically described phenotypes were often seen. Additionally, macular and peripheral atrophy were often associated with PARD phenotypes. Refinement of PARD phenotypes using newer multimodal imaging techniques will likely assist diagnosis and future clinical trials.

Unpaired fundus image enhancement based on constrained generative adversarial networks.

Fundus photography (FP) is a crucial technique for diagnosing the progression of ocular and systemic diseases in clinical studies, with wide applications in early clinical screening and diagnosis. However, due to the nonuniform illumination and imbalanced intensity caused by various reasons, the quality of fundus images is often severely weakened, brings challenges for automated screening, analysis, and diagnosis of diseases. To resolve this problem, we developed strongly constrained generative adversarial networks (SCGAN). The results demonstrate that the quality of various datasets were more significantly enhanced based on SCGAN, simultaneously more effectively retaining tissue and vascular information under various experimental conditions. Furthermore, the clinical effectiveness and robustness of this model were validated by showing its improved ability in vascular segmentation as well as disease diagnosis. Our study provides a new comprehensive approach for FP and also possesses the potential capacity to advance artificial intelligence-assisted ophthalmic examination.

Automated interpretation of retinal vein occlusion based on fundus fluorescein angiography images using deep learning: A retrospective, multi-center study.

Purpose: Fundus fluorescein angiography (FFA) is the gold standard for retinal vein occlusion (RVO) diagnosis. This study aims to develop a deep learning-based system to diagnose and classify RVO using FFA images, addressing the challenges of time-consuming and variable interpretations by ophthalmologists. Methods: 4028 FFA images of 467 eyes from 463 patients were collected and annotated. Three convolutional neural networks (CNN) models (ResNet50, VGG19, InceptionV3) were trained to generate the label of image quality, eye, location, phase, lesions, diagnosis, and macular involvement. The performance of the models was evaluated by accuracy, precision, recall, F-1 score, the area under the curve, confusion matrix, human-machine comparison, and Clinical validation on three external data sets. Results: The InceptionV3 model outperformed ResNet50 and VGG19 in labeling and interpreting FFA images for RVO diagnosis, achieving 77.63%-96.45% accuracy for basic information labels and 81.72%-96.45% for RVO-relevant labels. The comparison between the best CNN and ophthalmologists showed up to 19% accuracy improvement with the inceptionV3. Conclusion: This study developed a deep learning model capable of automatically multi-label and multi-classification of FFA images for RVO diagnosis. The proposed system is anticipated to serve as a new tool for diagnosing RVO in places short of medical resources.

Multi-degradation-adaptation network for fundus image enhancement with degradation representation learning.

Fundus image quality serves a crucial asset for medical diagnosis and applications. However, such images often suffer degradation during image acquisition where multiple types of degradation can occur in each image. Although recent deep learning based methods have shown promising results in image enhancement, they tend to focus on restoring one aspect of degradation and lack generalisability to multiple modes of degradation. We propose an adaptive image enhancement network that can simultaneously handle a mixture of different degradations. The main contribution of this work is to introduce our Multi-Degradation-Adaptive module which dynamically generates filters for different types of degradation. Moreover, we explore degradation representation learning and propose the degradation representation network and Multi-Degradation-Adaptive discriminator for our accompanying image enhancement network. Experimental results demonstrate that our method outperforms several existing state-of-the-art methods in fundus image enhancement. Code will be available at https://github.com/RuoyuGuo/MDA-Net.

Segmentation of retinal microaneurysms in fluorescein fundus angiography images by a novel three-step model.

Introduction: Microaneurysms serve as early signs of diabetic retinopathy, and their accurate detection is critical for effective treatment. Due to their low contrast and similarity to retinal vessels, distinguishing microaneurysms from background noise and retinal vessels in fluorescein fundus angiography (FFA) images poses a significant challenge. Methods: We present a model for automatic detection of microaneurysms. FFA images were pre-processed using Top-hat transformation, Gray-stretching, and Gaussian filter techniques to eliminate noise. The candidate microaneurysms were coarsely segmented using an improved matched filter algorithm. Real microaneurysms were segmented by a morphological strategy. To evaluate the segmentation performance, our proposed model was compared against other models, including Otsu's method, Region Growing, Global Threshold, Matched Filter, Fuzzy c-means, and K-means, using both self-constructed and publicly available datasets. Performance metrics such as accuracy, sensitivity, specificity, positive predictive value, and intersection-over-union were calculated. Results: The proposed model outperforms other models in terms of accuracy, sensitivity, specificity, positive predictive value, and intersection-over-union. The segmentation results obtained with our model closely align with benchmark standard. Our model demonstrates significant advantages for microaneurysm segmentation in FFA images and holds promise for clinical application in the diagnosis of diabetic retinopathy. Conclusion: The proposed model offers a robust and accurate approach to microaneurysm detection, outperforming existing methods and demonstrating potential for clinical application in the effective treatment of diabetic retinopathy.

Effect of propylene glycol mannate sulfate on non-proliferative diabetic retinopathy.

PURPOSE: To evaluate the effect of propylene glycol mannate sulfate (PGMS) on retinopathy in non-proliferative diabetic patients. METHODS: Eighty patients (111 eyes) with non-proliferative diabetic retinopathy were selected and retrospectively analyzed. Patients were divided into a control group (40 cases, 56 eyes) and an experimental group (40 cases, 55 eyes) using a random number table method. The control group continued had routine blood glucose management, while the experimental group received PGMS 100 mg additionally TID for 60 days. Changes in visual acuity, fundus conditions including hemorrhage points and exudation in each quadrant, and non-perfusion area were revealed through fundus angiography before and after the treatment period. RESULTS: After PGMS treatment, the experimental group demonstrated significant improvements compared to the control group in terms of eyesight improvement (P=0.002), the macular edema and macular retinal thickness (P=0.008). The total clinical efficacy rate of the experimental group was 67.86%, which was higher than 38.18% of the control group (P=0.032). Notably, there was a significant reduction in macular hemorrhage and hard extrusion. CONCLUSION: Oral administration of PGMS is an effective treatment for non-proliferative diabetic retinopathy.