Digital Color Fundus Photographs Research Articles

Rapid detection of diabetic retinopathy in retinal images: a new approach using transfer learning and synthetic minority over-sampling technique

The challenge of early detection of diabetic retinopathy (DR), a leading cause of vision loss in working-age individuals in developed nations, was addressed in this study. Current manual analysis of digital color fundus photographs by clinicians, although thorough, suffers from slow result turnaround, delaying necessary treatment. To expedite detection and improve treatment timeliness, a novel automated detection system for DR was developed. This system utilized convolutional neural networks. Visual geometry group 16-layer network (VGG16), a pre-trained deep learning model, for feature extraction from retinal images and the synthetic minority over-sampling technique (SMOTE) to handle class imbalance in the dataset. The system was designed to classify images into five categories: normal, mild DR, moderate DR, severe DR, and proliferative DR (PDR). Assessment of the system using the Kaggle diabetic retinopathy dataset resulted in a promising 93.94% accuracy during the training phase and 88.19% during validation. These results highlight the system's potential to enhance DR diagnosis speed and efficiency, leading to improved patient outcomes. The study concluded that automation and artificial intelligence (AI) could play a significant role in timely and efficient disease detection and management.

Retinal pigment epithelium-Bruch's membrane volume in grading of age-related macular degeneration.

To assess the agreement of optical coherence tomography (OCT) algorithm-based retinal pigment epithelium -Bruch's membrane complex volume (RBV) with fundus photograph-based age-related macular degeneration (AMD) grading. Digital color fundus photographs (CFPs) and spectral domain OCT images were acquired from 96 elderly subjects. CFPs were graded according to Age-Related Eye Disease Study (AREDS) classification. OCT image segmentation and RBV data calculation were done with Orion™ software. Univariate and multivariate analyses were performed to find out whether AMD lesion features associated with higher RBVs. RBV correlated with AMD grading (rs=0.338, P=0.001), the correlation was slightly stronger in early AMD (n=52; rs=0.432, P=0.001). RBV was higher in subjects with early AMD compared with those with no AMD lesions evident in fundus photographs (1.05±0.20 vs 0.96±0.13 mm3, P=0.023). In multivariate analysis higher RBVs were associated significantly with higher total drusen (β=0.388, P=0.027) and pigmentation areas (β=0.319, P=0.020) in fundus photographs, whereas depigmentation area (β=-0.295, P=0.015) associated with lower RBV. RBV correlate with AMD grading status, with a stronger association in patients with moderate, non-late AMD grades. This effect is driven mostly by lesions with drusen or pigmentation. Lesions with depigmentation tend to have lower values. RBV is more comprehensive measurement of the key area of AMD pathogenesis, compared to sole drusen volume analysis. RBV measurements are independent on grader variations and offer a possibility to quantify early and middle grade AMD lesions in a research setting, but may not substitute fundus photograph-based grading in the whole range of AMD spectrum.

Diabetic Retinopathy Diagnosis Categorization Using Deep Learning

Diabetic Retinopathy (DR) is an eye disease associated with chronic diabetes. DR is the leading cause of blindness among working aged adults around the world and estimated it may affect more than 93 million people. Progression to vision impairment can be slowed or controlled if DR is detected in time, however this can be difficult as the disease often shows few symptoms until it is too late to provide effective treatment. Currently, detecting DR is a time-consuming and manual process, which requires an ophthalmologist or trained clinician to examine and evaluate digital color fundus photographs of the retina, to identify DR by the presence of lesions associated with the vascular abnormalities caused by the disease. The automated method of DR screening will speed up the detection and decision-making process, which will help to control or manage DR progression. This paper presents an automated classification system, in which it analyzes fundus images with varying illumination and fields of view and generates a severity grade for diabetic retinopathy (DR) using machine learning models such as CNN, VGG-16 and VGG-19.This system achieves 80% sensitivity, 82% accuracy, 82% specificity, and 0.904 AUC for classifying images into 5 categories ranging from 0 to 4, where 0 is no DR and 4 is proliferative DR.

Differentiating Glaucomatous Optic Neuropathy From Non-glaucomatous Optic Neuropathies Using Deep Learning Algorithms

A deep learning framework to differentiate glaucomatous optic disc changes due to glaucomatous optic neuropathy (GON) from non-glaucomatous optic disc changes due to non-glaucomatous optic neuropathies (NGONs). Cross-sectional study. A deep-learning system was trained, validated, and externally tested to classify optic discs as normal, GON, or NGON, using 2183 digital color fundus photographs. A Single-Center data set of 1822 images (660 images of NGON, 676 images of GON, and 486 images of normal optic discs) was used for training and validation, whereas 361 photographs from 4 different data sets were used for external testing. Our algorithm removed the redundant information from the images using an optic disc segmentation (OD-SEG) network, after which we performed transfer learning with various pre-trained networks. Finally, we calculated sensitivity, specificity, F1-score, and precision to show the performance of the discrimination network in the validation and independent external data set. For classification, the algorithm with the best performance for the Single-Center data set was DenseNet121, with a sensitivity of 95.36%, precision of 95.35%, specificity of 92.19%, and F1 score of 95.40%. For the external validation data, the sensitivity and specificity of our network for differentiating GON from NGON were 85.53% and 89.02%, respectively. The glaucoma specialist who diagnosed those cases in masked fashion had a sensitivity of 71.05% and a specificity of 82.21%. The proposed algorithm for the differentiation of GON from NGON yields results that have a higher sensitivity than those of a glaucoma specialist, and its application for unseen data thus is extremely promising.

Development and Application of a Standardized Testset for an Artificial Intelligence Medical Device Intended for the Computer-Aided Diagnosis of Diabetic Retinopathy.

To explore a centralized approach to build test sets and assess the performance of an artificial intelligence medical device (AIMD) which is intended for computer-aided diagnosis of diabetic retinopathy (DR). A framework was proposed to conduct data collection, data curation, and annotation. Deidentified colour fundus photographs were collected from 11 partner hospitals with raw labels. Photographs with sensitive information or authenticity issues were excluded during vetting. A team of annotators was recruited through qualification examinations and trained. The annotation process included three steps: initial annotation, review, and arbitration. The annotated data then composed a standardized test set, which was further imported to algorithms under test (AUT) from different developers. The algorithm outputs were compared with the final annotation results (reference standard). The test set consists of 6327 digital colour fundus photographs. The final labels include 5 stages of DR and non-DR, as well as other ocular diseases and photographs with unacceptable quality. The Fleiss Kappa was 0.75 among the annotators. The Cohen's kappa between raw labels and final labels is 0.5. Using this test set, five AUTs were tested and compared quantitatively. The metrics include accuracy, sensitivity, and specificity. The AUTs showed inhomogeneous capabilities to classify different types of fundus photographs. This article demonstrated a workflow to build standardized test sets and conduct algorithm testing of the AIMD for computer-aided diagnosis of diabetic retinopathy. It may provide a reference to develop technical standards that promote product verification and quality control, improving the comparability of products.

Characterization of West African Crystalline Macular Dystrophy in the Ghanaian Population

West African crystalline maculopathy (WACM) is characterized by the presence of macular hyperrefractile crystal-like deposits. Although the underlying pathophysiology has not been elucidated, a few biologic drivers have been proposed. We analyzed a large WACM case series to gain a more robust understanding of its features and etiology. Prospective, cross-sectional cohort study. Participants with WACM were selected from the large cohort recruited in the Ghana Age-Related Macular Degeneration Study. Demographic and detailed medical histories, full ophthalmic examinations, digital color fundus photographs, and OCT images were obtained. All cases with WACM were evaluated by 3 retina experts. Crystal numbers, location, and distribution were determined. Associations between WACM and White age-related macular degeneration (AMD) risk variants were assessed using Firth's bias-reduced logistic regression, including age and sex as covariates. Phenotypic features of, and genetic associations with, WACM. West African crystalline maculopathy was identified in 106 eyes of 53 participants: 22 were bilateral and 24 were unilateral. Grading for AMD was not possible in 1 eye in 7 participants with WACM; therefore, laterality was not assessed in these subjects. Thirty-eight participants were women and were 14 men; sex was unrecorded for 1 participant. The mean age was 68.4 years (range, 45-101 years). Typical WACM crystals were demonstrated on OCT, which were more easily identified at high contrast and predominantly located at the inner limiting membrane. In eyes with copathology, crystals localized deeper in the inner retina, with wider retinal distribution over copathology lesions. There was no association with age or sex. A significant association was observed between the complement factor H (CFH) 402H risk variant and WACM. This study confirms the localization of crystals adjacent to the inner limiting membrane and distribution over lesions in eyes with copathology. The evaluation of OCT images under high contrast allows improved identification. West African crystalline maculopathy may be associated with the CFH-CFHR5 AMD risk locus identified among Whites; however, it is also possible that the combination of crystals and the CFH 402H allele increases the risk for developing late AMD. Further analyses using larger sample sizes are warranted to identify causalities between genotype and WACM phenotype.

Comparison between optical coherence tomography angiography and fundus fluorescein angiography for early detection of diabetic retinopathy

Purpose This work studied the ability of optical coherence tomography angiography (OCTA) to detect and analyze microvasculature changes in the retina in early nonproliferative diabetic retinopathy (NPDR) compared with fluorescein angiography (FA) regarding the posterior pole. The study compared two groups of patients with NPDR, a group with diabetic macular edema (DME) and another group without DME, with a group of health individuals as the control. The study compared foveal avascular zone (FAZ) size, capillary perfusion (CP), and vessel density in the superficial capillary plexus (SCP) in both the control and the two groups of patients with NPDR with and without DME. The number of microaneurysms (MAs) was also analyzed in the SCP and deep capillary plexus of OCTA and compared with FA. Patients and methods The study included 40 eyes of patients with NPDR and 20 eyes of volunteers as control. FA, OCT, and OCTA were done for all 40 eyes of NPDR, and only OCTA was done for the control group. OCTA parameters were as follows: ‘vessel density (mm/mm2), CP (%) (in SCP), and FAZ area (mm2) in SCP of macular area.' OCTA was performed using both 3×3 mm and 6 × 6 mm scans of the OCTA and with production of an OCT macular map. Digital color fundus photography and FA photographs were analyzed in relation to NPDR with evaluation of;Deg;MA;Deg;s, FAZ appearance, and late dye leakage. For;Deg;MA;Deg; counting, a 3×3 mm section centered on the fovea was manually delimited on FA images. Results MA count comparison between FA and OCTA (SCP and deep capillary plexus) showed highly a statistically significant differences between them, with P value less than 0.001 in all cases. DMI appeared early in all eyes with NPDR, presented with vessel density, CP, and the FAZ area in the SCP in both groups of patients with NPDR (in group A without DME and in group B with DME), with no significant statistical difference between the two groups of NPDR. Group A had a mean±SD vessel density of 9.95±2.805 mm/mm2, mean±SD of CP 18.97±4.754%, and mean±SD FAZ size of 0.59±0.228 mm2, whereas in group B, mean±SD vessel density was 10.19±3.152 mm/mm2, mean±SD CP was 20.14±6.131%, and mean±SD FAZ size was 0.63±0.270 mm2. When comparing groups A and B of patients with NPDR with the control group (group C), the control group showed a mean±SD vessel density of 13.23±2.454 mm/mm2, mean±SD of CP 27.89±5.627%, and mean±SD FAZ size of 0.36±0.097 mm2. A highly significant difference was found between patients with NPDR (both groups A and B) and healthy participants (group C). Conclusions MAs are more prominent in FA than in OCTA. Superficial vessel density and CP of macular region in NPDR eyes with and without DME are both decreased, in correlation to control group. The FAZ region of patients with NPDR is also enlarged when compared with healthy participants. CP and vessel density cannot be evaluated by FA, which is an advantage of OCTA over FA, but by using FA, we can evaluate the periphery, which cannot be evaluated by OCTA up till now.

Screening for Diabetic Retinopathy Using an Automated Diagnostic System Based on Deep Learning: Diagnostic Accuracy Assessment

Purpose: To evaluate the diagnostic accuracy of a diagnostic system software for the automated screening of diabetic retinopathy (DR) on digital colour fundus photographs, the 2019 Convolutional Neural Network (CNN) model with Inception-V3. Methods: In this cross-sectional study, 295 fundus images were analysed by the CNN model and compared to a panel of ophthalmologists. Images were obtained from a dataset acquired within a screening programme. Diagnostic accuracy measures and respective 95% CI were calculated. Results: The sensitivity and specificity of the CNN model in diagnosing referable DR was 81% (95% CI 66–90%) and 97% (95% CI 95–99%), respectively. Positive predictive value was 86% (95% CI 72–94%) and negative predictive value 96% (95% CI 93–98%). The positive likelihood ratio was 33 (95% CI 15–75) and the negative was 0.20 (95% CI 0.11–0.35). Its clinical impact is demonstrated by the change observed in the pre-test probability of referable DR (assuming a prevalence of 16%) to a post-test probability for a positive test result of 86% and for a negative test result of 4%. Conclusion: A CNN model negative test result safely excludes DR, and its use may significantly reduce the burden of ophthalmologists at reading centres.

Artificial Intelligence in Ophthalmology: Evolutions in Asia.

Artificial intelligence (AI) has been studied in ophthalmology since availability of digital information in ophthalmic care. The significant turning point was availability of commercial digital color fundus photography in the late 1990s, which caused digital screening for diabetic retinopathy (DR) to take off. Automated Retinal Disease Assessment software was then developed using machine learning to detect abnormal lesions in fundus to screen DR. The use of this version of AI had not been generalized because the specificity at 45% was not high enough, although the sensitivity reached 90%. The recent breakthrough in machine learning is the invent of deep learning, which accelerates its performance to be on par with experts. The first 2 breakthrough studies on deep learning for screening DR were conducted in Asia. The first represented collaboration of datasets between Asia and the United States for algorithms development, whereas the second represented algorithms developed in Asia but validated in different populations across the world. Both found accuracy for detecting referable DR of >95%. Diversity and variety are unique strengths of Asia for AI studies. There are many more studies of AI ongoing in Asia not only as prospective deployments in DR but in glaucoma, age-related macular degeneration, cataract, and systemic disease, such as Alzheimer's disease. Some Asian countries have laid out plans for digital health care system using AI as one of the puzzle pieces for solving blindness. More studies on AI and digital health are expected to come from Asia in this new decade.

Leucine-rich α-2-glycoprotein predicts proliferative diabetic retinopathy in type 2 diabetes

AimWe aim to examine the association of plasma leucine-rich-α-2-glycoprotein 1 (LRG1) with diabetic retinopathy (DR) in type 2 diabetes. MethodsAt baseline visit, plasma LRG1 levels were assessed using ELISA. Central arterial stiffness was estimated by carotid-femoral pulse wave velocity (PWV). At follow-up visit (median = 3.2 years), digital color fundus photographs were assessed for DR. DR severity was categorized into non-proliferative DR (NPDR) and proliferative DR (PDR). ResultsDR was diagnosed in 396 (32.8%) of 1206 patients. DR has higher LRG1 than non-DR (19.5 ± 11.3vs.16.9 ± 8.9 μg/ml, p ≪ 0.001). After adjustment, LRG1 was not associated with DR (OR = 1.2, [95% CI, 0.96–1.30], p = 0.16). LRG1 was higher in PDR (n = 107) than NPDR (n = 270) (23.2 ± 15.4vs.18.1 ± 8.9 μg/ml, n = 270, p ≪ 0.001). After adjustment, with 1-SD increase in LRG1, the relative risk of NPDR and PDR was 0.99 ([0.83–1.18], p = 0.91) and 1.42 ([95% CI, 1.14–1.76], p = 0.002) (p-trend = 0.01), respectively. We didn't observe significant improvement in AUC after adding LRG1 into the model. Baseline PWV mediated 12.0% of the association between LRG1 and PDR (p = 0.03). ConclusionBaseline plasma LRG1 is associated with PDR, suggesting it maybe a promising biomarker for prediction for advanced proliferative stages of DR. The mediation result indicates the potential benefit of ameliorating central arterial stiffness to prevent PDR.

Association of central arterial stiffness with the presence and severity of diabetic retinopathy in Asians with type 2 diabetes.

Arterial stiffness has been associated with diabetic retinopathy; however, the information is limited in Asians. We aim to examine the association of central arterial stiffness with the presence and severity of diabetic retinopathy in type 2 diabetes mellitus patients in Singapore. Arterial stiffness was estimated by carotid-femoral pulse wave velocity and augmentation index using applanation tonometry method. Digital colour fundus photographs from 1,203 patients were assessed for diabetic retinopathy. Diabetic retinopathy severity was categorized into non-proliferative diabetic retinopathy and proliferative diabetic retinopathy. Logistic regression model was used to evaluate the associations of diabetic retinopathy with pulse wave velocity and augmentation index. Diabetic retinopathy was diagnosed in 391 (32.5%) patients, including 271 non-proliferative diabetic retinopathy and 108 proliferative diabetic retinopathy. Diabetic retinopathy have higher pulse wave velocity (11.2 ± 3.3 vs 9.5 ± 2.6 m/s, p < 0.001) and augmentation index (28.4 ± 9.4 vs 26.1 ± 10.6%, p < 0.001) than non-diabetic retinopathy. After multivariable adjustment, pulse wave velocity [odds ratio = 1.11 (95% confidence interval = 1.05-1.17), p < 0.001] and augmentation index [odds ratio = 1.03 (95% confidence interval = 1.01-1.04), p = 0.009] was associated with diabetic retinopathy. In severity analyses, pulse wave velocity was associated with non-proliferative diabetic retinopathy [odds ratio = 1.10 (95% confidence interval = 1.03-1.17), p = 0.002] and proliferative diabetic retinopathy [odds ratio = 1.15 (95% confidence interval = 1.06-1.25), p = 0.001] (p-trend < 0.001). Augmentation index showed significant associations with non-proliferative diabetic retinopathy [odds ratio = 1.02 (95% confidence interval = 1.01-1.04), p = 0.008], but not with proliferative diabetic retinopathy [odds ratio = 1.01 (95% confidence interval = 0.98-1.04), p = 0.36] (p-trend = 0.03). Central arterial stiffness was associated with the presence and severity of diabetic retinopathy in type 2 diabetes mellitus patients, suggesting its etiologic implication in diabetic retinopathy.

FAMILIAL CENTRAL SEROUS CHORIORETINOPATHY.

To assess ophthalmologic characteristics in patients and unaffected individuals in families with multiple members affected by central serous chorioretinopathy (CSC), both at presentation and long-term follow-up. In 103 subjects from 23 families with at least 2 affected patients with CSC per family, prospective extensive ophthalmologic examination was performed, including best-corrected visual acuity, indirect ophthalmoscopy, digital color fundus photography, optical coherence tomography, fundus autofluorescence, and fluorescein angiography imaging. From these, 24 individuals from 6 families had undergone extensive ophthalmologic examination in either 1994 or 1995 and were followed up in this study. Subretinal fluid accumulation on optical coherence tomography and/or "hot spots" of leakage on fluorescein angiography indicative of CSC were detected in 45 of 103 phenotyped subjects (44%). Findings suggestive of CSC, but without the presence of subretinal fluid on optical coherence tomography and/or "hot spots" of leakage on fluorescein angiography, were observed in an additional 27 family members (26%). In 4 of 17 previously nonaffected subjects (24%) from the 24 individuals that were followed up after more than 20 years, we found more severe abnormalities. Extensive ophthalmologic phenotyping resulted in the detection of (suggestive) CSC in 52% of family members of patients with CSC. Genetic factors may play an important role in these specific CSC cases. Moreover, during follow-up, progressive disease can occur in a noteworthy number of patients.

Metaheuristic Techniques for Detection of Optic Disc in Retinal Fundus Images

Eye diseases like glaucoma and diabetic retinopathy are known to be the thieves of eye-sight that are responsible for causing the vision loss worldwide. Automatic detection of such diseases with the help of the digital color fundus photography helps in early diagnosis and treatment. From the fundus images, optic disc is required to be analyzed to diagnose the disease. In this paper, a technique has been proposed for locating optic disc through metaheuristic techniques namely Ant Colony Optimization algorithm, Bacterial Foraging Optimization, Firefly algorithm, Cuckoo Search algorithm and Krill Herd algorithm. A comparison has been made amongst all of them and also with existing disc detection techniques. The bacterial foraging algorithm has shown the best results as it has obtained 99.55% accuracy with DiaRetDB1 database, 100% accuracy with HEI-MED database, 100% with DRIVE database and 98% with STARE database.

The association of serum creatinine and estimated glomerular filtration rate variability with diabetic retinopathy in Asians with type 2 diabetes: A nested case-control study.

Fluctuation of kidney function may signify intra-glomerular microvascular hemodynamic instability. We aim to examine the association of long-term serum creatinine and estimated glomerular filtration rate variability with diabetic retinopathy. We included type 2 diabetes mellitus patients who attended the Diabetes Centre in 2011-2014 and were followed up (median = 3.2 years). Digital colour fundus photographs were assessed for diabetic retinopathy at follow-up. Diabetic retinopathy severity was categorized into non-proliferative diabetic retinopathy and proliferative diabetic retinopathy. We conducted a nested case-control study involving 177 diabetic retinopathy (118 non-proliferative diabetic retinopathy, 50 proliferative diabetic retinopathy) and 327 age- and gender-matched non-diabetic retinopathy. Serum creatinine measured before follow-up visit was obtained (⩾3 readings/patient). Variability was calculated as intra-individual standard deviation/√ n/( n - 1). Diabetic retinopathy have higher adjusted-serum creatinine-standard deviation than non-diabetic retinopathy [9.1 (4.9-21.6) vs 5.4 (3.4-10.1) µM, p < 0.001]. After multivariable adjustment, adjusted-serum creatinine-standard deviation was associated with diabetic retinopathy [odds ratio = 1.47, 95% confidence interval (1.02-2.10), p = 0.04]. The area under the curve increased significantly after adding adjusted-serum creatinine-standard deviation [0.70 (0.65-0.75) vs 0.72 (0.68-0.77), p < 0.03]. Proliferative diabetic retinopathy have higher adjusted-serum creatinine-standard deviation than non-proliferative diabetic retinopathy [15.5 (6.6-39.7) vs 7.47 (4.52-17.8) µM, p < 0.001]. After adjustment, adjusted-serum creatinine-standard deviation remained associated with non-proliferative diabetic retinopathy [1.48 (1.04-2.12), p = 0.03] and proliferative diabetic retinopathy [2.43 (1.34-4.39), p = 0.003; p-trend = 0.002]. Similar findings were observed for estimated glomerular filtration rate variability. Serum creatinine and estimated glomerular filtration rate variability is associated with the presence and severity of diabetic retinopathy independent of intra-individual means. This may inform novel therapeutic strategies aiming to achieve stable renal function in type 2 diabetes mellitus.

Outer Retina and Choroidal Thickness in Intermediate Age-Related Macular Degeneration: Reticular Pseudodrusen Findings

Purpose: To evaluate outer retina and choroidal thickness in subjects with intermediate age-related macular degeneration (iAMD) and to describe associations with the presence of reticular pseudodrusen (RPD). Methods: This was a retrospective, cross-sectional analysis of 157 consecutive eyes (specifically: 62 eyes classified as having RPD and 95 eyes with drusen ≥125 µm). Only cases with digital color fundus photographs, red-free, and infrared, obtained and graded according to the Age-Related Eye Disease Study to define iAMD, were used for this study. Outer retina and choroidal thickness were manually segmented and quantified at 12 locations in the horizontal meridian. Results: RPD appeared to be associated with thinning of the outer layers even after adjustment for gender and age. The presence of RPD in iAMD decreased with increase of choroidal thickness (total odds ratio [OR] 0.991, 95% confidence interval [CI] 0.985-0.996; nasal OR 0.992, 95% CI 0.986-0.997), with increased thickness of the myoid zone of the photoreceptors (total OR 0.812, 95% CI 0.688-0.958; nasal OR 0.863, 95% CI 0.755-0.987) and with increased thickness of the outer segment of the photoreceptors (total OR 0.850, 95% CI 0.731-0.989; nasal OR 0.857, 95% CI 0.736-0.989). Conclusions: The greatest differences between eyes with and without RPD are found at the level of the choroidal thickness and at the level of the photoreceptors.

Semi-automated quantification of hard exudates in colour fundus photographs diagnosed with diabetic retinopathy

BackgroundHard exudates (HEs) are the classical sign of diabetic retinopathy (DR) which is one of the leading causes of blindness, especially in developing countries. Accordingly, disease screening involves examining HEs qualitatively using fundus camera. However, for monitoring the treatment response, quantification of HEs becomes crucial and hence clinicians now seek to measure the area of HEs in the digital colour fundus (CF) photographs. Against this backdrop, we proposed an algorithm to quantify HEs using CF images and compare with previously reported technique using ImageJ.MethodsCF photographs of 30 eyes (20 patients) with diabetic macular edema were obtained. A robust semi-automated algorithm was developed to quantify area covered by HEs. In particular, the proposed algorithm, a two pronged methodology, involved performing top-hat filtering, second order statistical filtering, and thresholding of the colour fundus images. Subsequently, two masked observers performed HEs measurements using previously reported ImageJ-based protocol and compared with those obtained through proposed method. Intra and inter-observer grading was performed for determining percentage area of HEs identified by the individual algorithm.ResultsOf the 30 subjects, 21 were males and 9 were females with a mean age of the 50.25 ± 7.80 years (range 33–66 years). The correlation between the two measurements of semi-automated and ImageJ were 0.99 and 0.99 respectively. Previously reported method detected only 0–30% of the HEs area in 9 images, 30–60% in 12 images and 60–90% in remaining images, and more than 90% in none. In contrast, proposed method, detected 60–90% of the HEs area in 13 images and 90–100% in remaining 17 images.ConclusionProposed method semi-automated algorithm achieved acceptable accuracy, qualitatively and quantitatively, on a heterogeneous dataset. Further, quantitative analysis performed based on intra- and inter-observer grading showed that proposed methodology detects HEs more accurately than previously reported ImageJ-based technique. In particular, we proposed algorithm detect faint HEs also as opposed to the earlier method.

Direct medical cost associated with diabetic retinopathy severity in type 2 diabetes in Singapore

Diabetic retinopathy (DR) is a leading cause of vision-loss globally among type 2 diabetes (T2DM) patients. Information on the economic burden of DR in Singapore is limited. We aim to identify the total annual direct medical costs of DR at different stages, and to examine factors influencing the costs. Four hundreds and seventy T2DM patients who attended the Diabetes Centre in a secondary hospital in Singapore in 2011–2014 were included. Digital color fundus photographs were assessed for DR in a masked fashion. Retinopathy severity was further categorized into non-proliferative DR (NPDR), including mild, moderate and severe NPDR, and proliferative DR (PDR). Medical costs were assessed using hospital administrative data. DR was diagnosed in 172 (39.5%) patients, including 51 mild, 62 moderate and 18 severe NPDR, and 41 PDR. The median cost in DR [2012.0 (1111.2–4192.3)] was significantly higher than that in non-DR patients [1158.1 (724.1–1838.9)] (p<0.001). The corresponding costs for mild, moderate, severe NPDR and PDR were [1167.1 (895.4–2012.0)], [2212.0 (1215.5–3825.5)], [2717.5 (1444.0–6310.7)], and [3594.8.1 (1978.4–8427.7)], respectively. After adjustment, the corresponding cost ratios for mild, moderate, severe NPDR, and PDR relative to non-DR were 1.1 (p = 0.827), 1.8 (p = 0.003), 2.0 (p = 0.031) and 2.3 (p<0.001), respectively. The other factors affecting the total cost include smoking (ratio = 1.7, p = 0.019), neuropathy (ratio = 1.9, p = 0.001) and chronic kidney disease (CKD) (ratio = 1.4, p = 0.019). The presence and severity of DR was associated with increased direct medical costs in T2DM. Our results suggest that preventing progression of DR may reduce the economic burden of DR.

Automated Brightness and Contrast Adjustment of Color Fundus Photographs for the Grading of Age-Related Macular Degeneration

PurposeThe purpose of this study was to develop an algorithm to automatically standardize the brightness, contrast, and color balance of digital color fundus photographs used to grade AMD and to validate this algorithm by determining the effects of the standardization on image quality and disease grading. MethodsSeven-field color photographs of patients (>50 years) with any stage of AMD and a control group were acquired at two study sites, with either the Topcon TRC-50DX or Zeiss FF-450 Plus cameras. Field 2 photographs were analyzed. Pixel brightness values in the red, green, and blue (RGB) color channels were adjusted in custom-built software to make the mean brightness and contrast of the images equal to optimal values determined by the Age-Related Eye Disease Study (AREDS) 2 group.ResultsColor photographs of 370 eyes were analyzed. We found a wide range of brightness and contrast values in the images at baseline, even for those taken with the same camera. After processing, image brightness variability (brightest image–dimmest image in a color channel) was reduced 69-fold, 62-fold, and 96-fold for the RGB channels. Contrast variability was reduced 6-fold, 8-fold, and 13-fold, respectively, after adjustment. Of the 23% images considered nongradable before adjustment, only 5.7% remained nongradable.ConclusionsThis automated software enables rapid and accurate standardization of color photographs for AMD grading.Translational RelevanceThis work offers the potential to be the future of assessing and grading AMD from photos for clinical research and teleimaging.

A Comprehensive Study of Retinal Vessel Classification Methods in Fundus Images

Nowadays, it is obvious that there is a relationship between changes in the retinal vessel structure and diseases such as diabetic, hypertension, stroke, and the other cardiovascular diseases in adults as well as retinopathy of prematurity in infants. Retinal fundus images provide non-invasive visualization of the retinal vessel structure. Applying image processing techniques in the study of digital color fundus photographs and analyzing their vasculature is a reliable approach for early diagnosis of the aforementioned diseases. Reduction in the arteriolar-venular ratio of retina is one of the primary signs of hypertension, diabetic, and cardiovascular diseases which can be calculated by analyzing the fundus images. To achieve a precise measuring of this parameter and meaningful diagnostic results, accurate classification of arteries and veins is necessary. Classification of vessels in fundus images faces with some challenges that make it difficult. In this paper, a comprehensive study of the proposed methods for classification of arteries and veins in fundus images is presented. Considering that these methods are evaluated on different datasets and use different evaluation criteria, it is not possible to conduct a fair comparison of their performance. Therefore, we evaluate the classification methods from modeling perspective. This analysis reveals that most of the proposed approaches have focused on statistics, and geometric models in spatial domain and transform domain models have received less attention. This could suggest the possibility of using transform models, especially data adaptive ones, for modeling of the fundus images in future classification approaches.

Retinal vessels caliber changes after strabismus surgery: results of 6mo follow-up.

To evaluate the effect of strabismus surgery on retinal vessels calibers with digital color fundus photographs. Two hundred consecutive strabismus patients underwent surgery, and 118 patients (female/male, 55/63) who finished 6-month follow-up were finally included in this study. Optic disc-centered digital color fundus photographs of both eyes of all patients were taken prior to surgery and 6mo post surgery. The retinal vascular caliber of 116 operated eyes were measured using the computer program IVAN. The operated eyes were divided into 3 groups according to the surgical methods, recession of one muscle, one muscle recession and one muscle folding, one muscle resection and one muscle recession. The effect of number of altered muscles on retinal vessels was analyzed using statistic software SPSS 16.0. The mean age was 12.4±8.6y. Averaged central retinal artery equivalent (CRAE) of all patients was 120.31±23.02 µm preoperatively, and 122.87±15.93 µm six months after surgery. Averaged central retinal vein equivalent (CRVE) was 171.11±31.73 µm preoperatively and 175.02±21.00 µm postoperatively. There was no significant difference of averaged CRAE (P=0.22) or CRVE (P=0.19) before and after operation. Averaged arteriole to venule ratio (AVR) was 0.71±0.07 before surgery and 0.70±0.07 after surgery. Comparison of preoperative and postoperative retinal vessels calibers among different surgical groups did not show significant differences. Also, there was no advantage of rectus muscle folding to muscle resection. Up to 6mo after strabismus surgery, the retinal vascular calibers were not altered. No more than two muscles in one surgery are safe for retinal perfusion.