Automated classification of choroidal neovascularization, diabetic macular edema, and drusen from retinal OCT images using vision transformers: a comparative study.

Classifying retinal diseases is a complex problem because the early problematic areas of retinal disorders are quite small and conservative. In recent years, Transformer architectures have been successfully applied to solve various retinal related health problems. Age-related macular degeneration (AMD) and diabetic macular edema (DME), two prevalent retinal diseases, can cause partial or total blindness. Diseases therefore require an early and accurate detection. In this study, we proposed Vision Transformer (ViT), Tokens-To-Token Vision Transformer (T2T-ViT) and Mobile Vision Transformer (Mobile-ViT) algorithms to detect choroidal neovascularization (CNV), drusen, and diabetic macular edema (DME), and normal using optical coherence tomography (OCT) images. The predictive accuracies of ViT, T2T-ViT and Mobile-ViT achieved on the dataset for the classification of OCT images are 95.14%, 96.07% and 99.17% respectively. Experimental results obtained from ViT approaches showed that Mobile-ViT have superior performance with regard to classification accuracy in comparison with the others. Overall, it has been observed that ViT architectures have the capacity to classify with high accuracy in the diagnosis of retinal diseases.

MediDRNet: Tackling category imbalance in diabetic retinopathy classification with dual-branch learning and prototypical contrastive learning.

BACKGROUND AND OBJECTIVE: The classification of diabetic retinopathy (DR) aims to utilize the implicit information in images for early diagnosis, to prevent and mitigate the further worsening of the condition. However, existing methods are often limited by the need to operate within large, annotated datasets to show significant advantages. Additionally, the number of samples for different categories within the dataset needs to be evenly distributed, because the characteristic of sample imbalance distribution can lead to an excessive focus on high-frequency disease categories, while neglecting the less common but equally important disease categories. Therefore, there is an urgent need to develop a new classification method that can effectively alleviate the issue of sample distribution imbalance, thereby enhancing the accuracy of diabetic retinopathy classification. METHODS: In this work, we propose MediDRNet, a dual-branch network model based on prototypical contrastive learning. This model adopts prototype contrastive learning, creating prototypes for different levels of lesions, ensuring they represent the core features of each lesion level. It classifies by comparing the similarity between data points and their category prototypes. Our dual-branch network structure effectively resolves the issue of category imbalance and improves classification accuracy by emphasizing subtle differences in retinal lesions. Moreover, our approach combines a dual-branch network with specific lesion-level prototypes for core feature representation and incorporates the convolutional block attention module for enhanced lesion feature identification. RESULTS: Our experiments using both the Kaggle and UWF classification datasets have demonstrated that MediDRNet exhibits exceptional performance compared to other advanced models in the industry, especially on the UWF DR classification dataset where it achieved state-of-the-art performance across all metrics. On the Kaggle DR classification dataset, it achieved the highest average classification accuracy (0.6327) and Macro-F1 score (0.6361). Particularly in the classification tasks for minority categories of diabetic retinopathy on the Kaggle dataset (Grades 1, 2, 3, and 4), the model reached high classification accuracies of 58.08%, 55.32%, 69.73%, and 90.21%, respectively. In the ablation study, the MediDRNet model proved to be more effective in feature extraction from diabetic retinal fundus images compared to other feature extraction methods. CONCLUSIONS: This study employed prototype contrastive learning and bidirectional branch learning strategies, successfully constructing a grading system for diabetic retinopathy lesions within imbalanced diabetic retinopathy datasets. Through a dual-branch network, the feature learning branch effectively facilitated a smooth transition of features from the grading network to the classification learning branch, accurately identifying minority sample categories. This method not only effectively resolved the issue of sample imbalance but also provided strong support for the precise grading and early diagnosis of diabetic retinopathy in clinical applications, showcasing exceptional performance in handling complex diabetic retinopathy datasets. Moreover, this research significantly improved the efficiency of prevention and management of disease progression in diabetic retinopathy patients within medical practice. We encourage the use and modification of our code, which is publicly accessible on GitHub: https://github.com/ReinforceLove/MediDRNet.

Self-supervised category selective attention classifier network for diabetic macular edema classification.

AIMS: This study aims to develop an advanced model for the classification of Diabetic Macular Edema (DME) using deep learning techniques. Specifically, the objective is to introduce a novel architecture, SSCSAC-Net, that leverages self-supervised learning and category-selective attention mechanisms to improve the precision of DME classification. METHODS: The proposed SSCSAC-Net integrates self-supervised learning to effectively utilize unlabeled data for learning robust features related to DME. Additionally, it incorporates a category-specific attention mechanism and a domain-specific layer into the ResNet-152 base architecture. The model is trained using an ensemble of unsupervised and supervised learning techniques. Benchmark datasets are utilized for testing the model's performance, ensuring its robustness and generalizability across different data distributions. RESULTS: Evaluation of the SSCSAC-Net on multiple datasets demonstrates its superior performance compared to existing techniques. The model achieves high accuracy, precision, and recall rates, with an accuracy of 98.7%, precision of 98.6%, and recall of 98.8%. Furthermore, the incorporation of self-supervised learning reduces the dependency on extensive labeled data, making the solution more scalable and cost-effective. CONCLUSIONS: The proposed SSCSAC-Net represents a significant advancement in automated DME classification. By effectively using self-supervised learning and attention mechanisms, the model offers improved accuracy in identifying DME-related features within retinal images. Its robustness and generalizability across different datasets highlight its potential for clinical applications, providing a valuable tool for clinicians in diagnosing DME effectively.

An enumerative pre-processing approach for retinopathy severity grading using an interpretable classifier: a comparative study.

BACKGROUND: Diabetic retinopathy (DR) is a serious eye complication that results in permanent vision damage. As the number of patients suffering from DR increases, so does the delay in treatment for DR diagnosis. To bridge this gap, an efficient DR screening system that assists clinicians is required. Although many artificial intelligence (AI) screening systems have been deployed in recent years, accuracy remains a metric that can be improved. METHODS: An enumerative pre-processing approach is implemented in the deep learning model to attain better accuracies for DR severity grading. The proposed approach is compared with various pre-trained models, and the necessary performance metrics were tabulated. This paper also presents the comparative analysis of various optimization algorithms that are utilized in the deep network model, and the results were outlined. RESULTS: The experimental results are carried out on the MESSIDOR dataset to assess the performance. The experimental results show that an enumerative pipeline combination K1-K2-K3-DFNN-LOA shows better results when compared with other combinations. When compared with various optimization algorithms and pre-trained models, the proposed model has better performance with maximum accuracy, precision, recall, F1 score, and macro-averaged metric of 97.60%, 94.60%, 98.40%, 94.60%, and 0.97, respectively. CONCLUSION: This study focussed on developing and implementing a DR screening system on color fundus photographs. This artificial intelligence-based system offers the possibility to enhance the efficacy and approachability of DR diagnosis.

DEVELOPMENT AND VALIDATION OF AN EXPLAINABLE ARTIFICIAL INTELLIGENCE FRAMEWORK FOR MACULAR DISEASE DIAGNOSIS BASED ON OPTICAL COHERENCE TOMOGRAPHY IMAGES.

PURPOSE: To develop and validate an artificial intelligence framework for identifying multiple retinal lesions at image level and performing an explainable macular disease diagnosis at eye level in optical coherence tomography images. METHODS: A total of 26,815 optical coherence tomography images were collected from 865 eyes, and 9 retinal lesions and 3 macular diseases were labeled by ophthalmologists, including diabetic macular edema and dry/wet age-related macular degeneration. We applied deep learning to classify retinal lesions at image level and random forests to achieve an explainable disease diagnosis at eye level. The performance of the integrated two-stage framework was evaluated and compared with human experts. RESULTS: On testing data set of 2,480 optical coherence tomography images from 80 eyes, the deep learning model achieved an average area under curve of 0.978 (95% confidence interval, 0.971-0.983) for lesion classification. In addition, random forests performed accurate disease diagnosis with a 0% error rate, which achieved the same accuracy as one of the human experts and was better than the other three experts. It also revealed that the detection of specific lesions in the center of macular region had more contribution to macular disease diagnosis. CONCLUSION: The integrated method achieved high accuracy and interpretability in retinal lesion classification and macular disease diagnosis in optical coherence tomography images and could have the potential to facilitate the clinical diagnosis.

Weakly Supervised Sensitive Heatmap framework to classify and localize diabetic retinopathy lesions.

Vision loss happens due to diabetic retinopathy (DR) in severe stages. Thus, an automatic detection method applied to diagnose DR in an earlier phase may help medical doctors to make better decisions. DR is considered one of the main risks, leading to blindness. Computer-Aided Diagnosis systems play an essential role in detecting features in fundus images. Fundus images may include blood vessels, exudates, micro-aneurysm, hemorrhages, and neovascularization. In this paper, our model combines automatic detection for the diabetic retinopathy classification with localization methods depending on weakly-supervised learning. The model has four stages; in stage one, various preprocessing techniques are applied to smooth the data set. In stage two, the network had gotten deeply to the optic disk segment for eliminating any exudate's false prediction because the exudates had the same color pixel as the optic disk. In stage three, the network is fed through training data to classify each label. Finally, the layers of the convolution neural network are re-edited, and used to localize the impact of DR on the patient's eye. The framework tackles the matching technique between two essential concepts where the classification problem depends on the supervised learning method. While the localization problem was obtained by the weakly supervised method. An additional layer known as weakly supervised sensitive heat map (WSSH) was added to detect the ROI of the lesion at a test accuracy of 98.65%, while comparing with Class Activation Map that involved weakly supervised technology achieved 0.954. The main purpose is to learn a representation that collect the central localization of discriminative features in a retina image. CNN-WSSH model is able to highlight decisive features in a single forward pass for getting the best detection of lesions.

CD146/Soluble CD146 Pathway Is a Novel Biomarker of Angiogenesis and Inflammation in Proliferative Diabetic Retinopathy.

Purpose: Inflammation, angiogenesis and fibrosis are pathological hallmarks of proliferative diabetic retinopathy (PDR). The CD146/sCD146 pathway displays proinflammatory and proangiogenic properties. We investigated the role of this pathway in the pathophysiology of PDR. Methods: Vitreous samples from 41 PDR and 27 nondiabetic patients, epiretinal fibrovascular membranes from 18 PDR patients, rat retinas, human retinal microvascular endothelial cells (HRMECs) and human retinal Müller glial cells were studied by ELISA, Western blot analysis, immunohistochemistry and immunofluorescence microscopy analysis. Blood-retinal barrier breakdown was assessed with fluorescein isothiocyanate-conjugated dextran. Results: sCD146 and VEGF levels were significantly higher in vitreous samples from PDR patients than in nondiabetic patients. In epiretinal membranes, immunohistochemical analysis revealed CD146 expression in leukocytes, vascular endothelial cells and myofibroblasts. Significant positive correlations were detected between numbers of blood vessels expressing CD31, reflecting angiogenic activity of PDR, and numbers of blood vessels and stromal cells expressing CD146. Western blot analysis showed significant increase of CD146 in diabetic rat retinas. sCD146 induced upregulation of phospho-ERK1/2, NF-κB , VEGF and MMP-9 in Müller cells. The hypoxia mimetic agent cobalt chloride, VEGF and TNF-α induced upregulation of sCD146 in HRMECs. The MMP inhibitor ONO-4817 attenuated TNF-α-induced upregulation of sCD146 in HRMECs. Intravitreal administration of sCD146 in normal rats significantly increased retinal vascular permeability and induced significant upregulation of phospho-ERK1/2, intercellular adhesion molecule-1 and VEGF in the retina. sCD146 induced migration of HRMECs. Conclusions: These results suggest that the CD146/sCD146 pathway is involved in the initiation and progression of PDR.

An Effective Method for Detecting and Classifying Diabetic Retinopathy Lesions Based on Deep Learning.

Diabetic retinopathy occurs as a result of the harmful effects of diabetes on the eyes. Diabetic retinopathy is also a disease that should be diagnosed early. If not treated early, vision loss may occur. It is estimated that one third of more than half a million diabetic patients will have diabetic retinopathy by the 22nd century. Many effective methods have been proposed for disease detection with deep learning. In this study, unlike other studies, a deep learning-based method has been proposed in which diabetic retinopathy lesions are detected automatically and independently of datasets, and the detected lesions are classified. In the first stage of the proposed method, a data pool is created by collecting diabetic retinopathy data from different datasets. With Faster RCNN, lesions are detected, and the region of interests are marked. The images obtained in the second stage are classified using the transfer learning and attention mechanism. The method tested in Kaggle and MESSIDOR datasets reached 99.1% and 100% ACC and 99.9% and 100% AUC, respectively. When the obtained results are compared with other results in the literature, it is seen that more successful results are obtained.

En face image-based classification of diabetic macular edema using swept source optical coherence tomography.

This retrospective study was performed to classify diabetic macular edema (DME) based on the localization and area of the fluid and to investigate the relationship of the classification with visual acuity (VA). The fluid was visualized using en face optical coherence tomography (OCT) images constructed using swept-source OCT. A total of 128 eyes with DME were included. The retina was segmented into: Segment 1, mainly comprising the inner nuclear layer and outer plexiform layer, including Henle's fiber layer; and Segment 2, mainly comprising the outer nuclear layer. DME was classified as: foveal cystoid space at Segment 1 and no fluid at Segment 2 (n = 24), parafoveal cystoid space at Segment 1 and no fluid at Segment 2 (n = 25), parafoveal cystoid space at Segment 1 and diffuse fluid at Segment 2 (n = 16), diffuse fluid at both segments (n = 37), and diffuse fluid at both segments with subretinal fluid (n = 26). Eyes with diffuse fluid at Segment 2 showed significantly poorer VA, higher ellipsoid zone disruption rates, and greater central subfield thickness than did those without fluid at Segment 2 (P < 0.001 for all). These results indicate the importance of the localization and area of the fluid for VA in DME.

Universal adversarial attacks on deep neural networks for medical image classification.

BACKGROUND: Deep neural networks (DNNs) are widely investigated in medical image classification to achieve automated support for clinical diagnosis. It is necessary to evaluate the robustness of medical DNN tasks against adversarial attacks, as high-stake decision-making will be made based on the diagnosis. Several previous studies have considered simple adversarial attacks. However, the vulnerability of DNNs to more realistic and higher risk attacks, such as universal adversarial perturbation (UAP), which is a single perturbation that can induce DNN failure in most classification tasks has not been evaluated yet. METHODS: We focus on three representative DNN-based medical image classification tasks (i.e., skin cancer, referable diabetic retinopathy, and pneumonia classifications) and investigate their vulnerability to the seven model architectures of UAPs. RESULTS: We demonstrate that DNNs are vulnerable to both nontargeted UAPs, which cause a task failure resulting in an input being assigned an incorrect class, and to targeted UAPs, which cause the DNN to classify an input into a specific class. The almost imperceptible UAPs achieved > 80% success rates for nontargeted and targeted attacks. The vulnerability to UAPs depended very little on the model architecture. Moreover, we discovered that adversarial retraining, which is known to be an effective method for adversarial defenses, increased DNNs' robustness against UAPs in only very few cases. CONCLUSION: Unlike previous assumptions, the results indicate that DNN-based clinical diagnosis is easier to deceive because of adversarial attacks. Adversaries can cause failed diagnoses at lower costs (e.g., without consideration of data distribution); moreover, they can affect the diagnosis. The effects of adversarial defenses may not be limited. Our findings emphasize that more careful consideration is required in developing DNNs for medical imaging and their practical applications.

Associations between Vascular Endothelial Growth Factor Gene Polymorphisms and Different Types of Diabetic Retinopathy Susceptibility: A Systematic Review and Meta-Analysis.

BACKGROUND: Vascular endothelial growth factor (VEGF) gene polymorphisms have been shown to be associated with the risk of diabetic retinopathy (DR), but the results were inconsistent. The aim of this study was to systematically assess the associations between VEGF gene polymorphisms and different types of DR (nonproliferative DR and proliferative DR). METHODS: Electronic databases PubMed, Embase, Web of Science, CNKI, and WANFANG DATA were searched for articles on the associations between VEGF gene polymorphisms and different types of DR up to November 6, 2019. Pooled odds ratios (ORs) and 95% confidence intervals (CIs) were calculated, and subgroup analyses were conducted by ethnicity. Sensitivity analysis was conducted to assess the stability of the results. Publication bias was assessed by using the Egger regression asymmetry test and visualization of funnel plots. A systematic review was conducted for polymorphisms with a high degree of heterogeneity (I 2 > 75%) or studied in only one study. RESULTS: A total of 13 and 18 studies analyzed the associations between VEGF SNPs and nonproliferative DR (NPDR) as well as proliferative DR (PDR), respectively. There were significant associations between rs2010963 and NPDR in Asian (dominant model: OR = 1.29, 95%CI = 1.04 - 1.60); and rs2010963 is associated with PDR in total population (dominant model: OR = 1.20, 95%CI = 1.03 - 1.41), either Asian (recessive model: OR = 1.57, 95%CI = 1.04 - 2.35) or Caucasian (recessive model: OR = 1.83, 95%CI = 1.28 - 2.63). Rs833061 is associated with PDR in Asian (recessive model: OR = 1.58, 95%CI = 1.11 - 2.26). Rs699947 is associated with NPDR in the total population (dominant model: OR = 2.04, 95%CI = 1.30 - 3.21) and associated with PDR in Asian (dominant model: OR = 1.72, 95%CI = 1.05 - 2.84). CONCLUSIONS: Rs2010963, rs833061, and rs699947 are associated with NPDR or PDR, which may be involved in the occurrence and development of DR.

Widefield optical coherence tomography angiography for early detection and objective evaluation of proliferative diabetic retinopathy.

PURPOSE: To evaluate the utility of widefield optical coherence tomography angiography (WF-OCTA) compared with clinical examination in grading diabetic retinopathy in patients diagnosed clinically with proliferative diabetic retinopathy (PDR) or severe non-proliferative diabetic retinopathy (NPDR). DESIGN: This retrospective observational case series included patients diagnosed clinically with PDR or severe NPDR. Patients underwent standard clinical examination and WF-OCTA imaging (PLEX Elite 9000, Carl Zeiss Meditec AG) using 12×12 montage scans between August 2018 and January 2019. Two trained graders identified neovascularisation at the disc (NVD) and neovascularisation elsewhere (NVE) on WF-OCTA which were compared with the clinical examination, and to ultra-widefield fluorescein angiography (UWFA) when available. RESULTS: Seventy-nine eyes of 46 patients were evaluated. Of those, 57 eyes were diagnosed clinically with PDR, and 22 with severe NPDR. NVD was detected on OCTA-B scan as preretinal hyperreflective material (PRHM) in 39 eyes (100%) with evident flow signals in 79.5% compared with 51.3% detected clinically. We further classified NVD on OCTA into four subtypes and found that subtypes 1 and 2 could not be seen on clinical examination alone. WF-OCTA detected NVE in 81% of the cases compared with 55.7% detected clinically. Using WF-OCTA resulted in a higher percentage of PDR grading (88.6%) than on clinical examination (72.2%). When available, UWFA confirmed the WF-OCTA diagnosis in the majority of cases. CONCLUSION: This study demonstrates that WF-OCTA has a higher detection rate of PDR than clinical examination. This suggests that this modality could be used non-invasively for the purpose of early detection and characterisation of neovascularisation.

Diagnostic accuracy of diabetic retinopathy grading by an artificial intelligence-enabled algorithm compared with a human standard for wide-field true-colour confocal scanning and standard digital retinal images.

BACKGROUND: Photographic diabetic retinopathy screening requires labour-intensive grading of retinal images by humans. Automated retinal image analysis software (ARIAS) could provide an alternative to human grading. We compare the performance of an ARIAS using true-colour, wide-field confocal scanning images and standard fundus images in the English National Diabetic Eye Screening Programme (NDESP) against human grading. METHODS: Cross-sectional study with consecutive recruitment of patients attending annual diabetic eye screening. Imaging with mydriasis was performed (two-field protocol) with the EIDON platform (CenterVue, Padua, Italy) and standard NDESP cameras. Human grading was carried out according to NDESP protocol. Images were processed by EyeArt V.2.1.0 (Eyenuk Inc, Woodland Hills, California). The reference standard for analysis was the human grade of standard NDESP images. RESULTS: We included 1257 patients. Sensitivity estimates for retinopathy grades were: EIDON images; 92.27% (95% CI: 88.43% to 94.69%) for any retinopathy, 99% (95% CI: 95.35% to 100%) for vision-threatening retinopathy and 100% (95% CI: 61% to 100%) for proliferative retinopathy. For NDESP images: 92.26% (95% CI: 88.37% to 94.69%) for any retinopathy, 100% (95% CI: 99.53% to 100%) for vision-threatening retinopathy and 100% (95% CI: 61% to 100%) for proliferative retinopathy. One case of vision-threatening retinopathy (R1M1) was missed by the EyeArt when analysing the EIDON images, but identified by the human graders. The EyeArt identified all cases of vision-threatening retinopathy in the standard images. CONCLUSION: EyeArt identified diabetic retinopathy in EIDON images with similar sensitivity to standard images in a large-scale screening programme, exceeding the sensitivity threshold recommended for a screening test. Further work to optimise the identification of 'no retinopathy' and to understand the differential lesion detection in the two imaging systems would enhance the use of these two innovative technologies in a diabetic retinopathy screening setting.

Comparison Between Graders in Detection of Diabetic Neovascularization With Swept Source Optical Coherence Tomography Angiography and Fluorescein Angiography.

PURPOSE: We compared the ability of ophthalmologists to identify neovascularization (NV) in patients with proliferative diabetic retinopathy using swept-source optical coherence tomography angiography (SS-OCTA) and fluorescein angiography (FA). DESIGN: Retrospective study comparing diagnostic instruments. METHODS: Eyes with proliferative diabetic retinopathy or severe nonproliferative diabetic retinopathy and a high suspicion of NV based on clinical examination were imaged using SS-OCTA and FA at the same visit. Two separate grading sets consisting of scrambled, anonymized SS-OCTA and FA images were created. The ground truth for presence of NV was established by consensus of 2 graders with OCTA experience who did not participate in the subsequent assessment of NV in this study. The 2 anonymized image sets were graded for presence or absence of NV by 12 other graders that included 2 residents, 6 vitreoretinal fellows, and 4 vitreoretinal attending physicians. The percentage of correct grading of NV using SS-OCTA and FA was assessed for each grader and across grader training levels. RESULTS: Forty-seven eyes from 24 patients were included in this study. Overall, the mean percentage of correct NV grading was 87.8% using SS-OCTA with B-scans and 86.2% using FA (P = .92). Assessing each grader individually, there was no statistically significant asymmetry in correct grading using SS-OCTA and FA. CONCLUSIONS: Ophthalmologists across training levels were able to identify diabetic NV with equal accuracy using SS-OCTA and FA. Based on these results, SS-OCTA may be an appropriate standalone modality for diagnosing diabetic NV.

Intravitreal injections during COVID-19 outbreak: Real-world experience from an Italian tertiary referral center.

We report our experience during COVID-19 outbreak for intravitreal injections in patients with maculopathy. We proposed a treatment priority levels and timings; the "High" priority level includes all monocular patients; the "Moderate" is assigned to all patients with an active macular neovascularization; the patients affected by diabetic macular edema or retinal vein occlusion belong to the "Low" class. This organization allowed us to treat the most urgent patients although the injections performed had a 91.7% drop compared to the same period of 2019.

Effect of ICD-9 to ICD-10 Transition on Accuracy of Codes for Stage of Diabetic Retinopathy and Related Complications: Results from the CODER Study.

PURPOSE: When the International Classification of Diseases 9th Revision (ICD-9) transitioned to the International Classification of Diseases 10th Revision (ICD-10), there was a marked increase in the complexity of International Classification of Diseases (ICD) codes with potential for improved specificity in clinical database research. The purpose of this study was to characterize the accuracy of coding for stage of diabetic retinopathy (DR) and DR-related complications (including vitreous hemorrhage, retinal detachment, and neovascular glaucoma) during this transition. DESIGN: Retrospective chart review of 3 time periods corresponding to the use of ICD-9: 2014-2015; "early" use of ICD-10, 2015-2016; and "late" use of ICD-10, 2018-2019. PARTICIPANTS: Patients aged 18 years or older with a diagnosis of DR at a multispecialty academic institution. METHODS: Positive predictive value (PPV), negative predictive value (NPV), sensitivity, specificity, and kappa (κ) statistics were generated for each diagnosis. Generalized estimating equation (GEE) models were used to assess the significance of the variables. MAIN OUTCOME MEASURE: The main outcome was the proportion of agreement between the ICD code and the documented chart standard for stage of DR and DR-related complications. RESULTS: A total of 600 patients were included in the study (average age, 61 years; range, 25-93 years). Overall, there was substantial agreement between the ICD codes for stage of DR and the documented standard (κ = 0.66). The proportion of ICD codes in agreement with the documented standard diagnosis increased with time: 66.5%, 78.5%, and 83.3% for ICD-9, "early" ICD-10, and "late" ICD-10, respectively. The odds of agreement were 2.67 (95% confidence interval [CI], 1.49-4.76, P < 0.001) and 3.96 (95% CI, 2.34-6.69, P < 0.0001) times greater for "early" and "late" ICD-10 codes compared with ICD-9 codes, respectively. For specific codes, the overall PPV, NPV, sensitivity, and specificity for nonproliferative diabetic retinopathy (NPDR) and proliferative diabetic retinopathy (PDR) were excellent (>90%). The odds of agreement were 19.70 (95% CI, 11.54-33.64, P < 0.0001) times greater for PDR than NPDR. Compared with the stage of DR, DR-related diagnoses were overall less accurately coded (κ = 0.61, 0.48, and 0.52 for vitreous hemorrhage, retinal detachment, and neovascular glaucoma, respectively). CONCLUSIONS: Coding in ICD-10 is more accurate than in ICD-9, particularly for PDR compared with NPDR. The increased accuracy emphasizes the potential for ICD-10 coding to be used effectively in database research.

Long-term effects of hyperbaric oxygen therapy on visual acuity and retinopathy.

Hyperbaric oxygen (HBO2) therapy is an adjunct treatment for diabetic foot ulcers. Since plausible mechanisms of action for this treatment include increased angiogenesis and high tissue oxygen concentrations, concerns about deterioration of retinopathy have been raised. The aim of this study was to evaluate the effects of HBO2 on visual acuity (VA) and retinopathy in patients with chronic diabetic foot ulcers during a two-year follow-up period. This is a randomized, single-center, double-blinded and placebo-controlled clinical trial evaluating the effects of HBO2 in patients with diabetes mellitus and chronic foot ulcers. All study participants underwent an ophthalmological examination before the first study treatment and then at three, six, 12 and 24 months. Fifty patients with a median age of 67 years were included. Visual acuity was similar between groups and did not change during the two-year observation period. No differences in retinopathy were seen between groups; neither were any differences found in numbers or areas of bleedings, hard exudates, microaneurysms or edemas, nor between groups or visits. New clinically significant macular edema was identified in four eyes in the HBO2 group and in three eyes in the placebo group. In this population of diabetic foot ulcer patients HBO2 seems to be neutral in an ophthalmological perspective. From a retinal point of view, we could not identify any indication of harmful effects of HBO2 on the microvascular bed in the placebo group.

Density-based classification in diabetic retinopathy through thickness of retinal layers from optical coherence tomography.

Diabetic retinopathy (DR) is a severe retinal disorder that can lead to vision loss, however, its underlying mechanism has not been fully understood. Previous studies have taken advantage of Optical Coherence Tomography (OCT) and shown that the thickness of individual retinal layers are affected in patients with DR. However, most studies analyzed the thickness by calculating summary statistics from retinal thickness maps of the macula region. This study aims to apply a density function-based statistical framework to the thickness data obtained through OCT, and to compare the predictive power of various retinal layers to assess the severity of DR. We used a prototype data set of 107 subjects which are comprised of 38 non-proliferative DR (NPDR), 28 without DR (NoDR), and 41 controls. Based on the thickness profiles, we constructed novel features which capture the variation in the distribution of the pixel-wise retinal layer thicknesses from OCT. We quantified the predictive power of each of the retinal layers to distinguish between all three pairwise comparisons of the severity in DR (NoDR vs NPDR, controls vs NPDR, and controls vs NoDR). When applied to this preliminary DR data set, our density-based method demonstrated better predictive results compared with simple summary statistics. Furthermore, our results indicate considerable differences in retinal layer structuring based on the severity of DR. We found that: (a) the outer plexiform layer is the most discriminative layer for classifying NoDR vs NPDR; (b) the outer plexiform, inner nuclear and ganglion cell layers are the strongest biomarkers for discriminating controls from NPDR; and (c) the inner nuclear layer distinguishes best between controls and NoDR.

Automated classification of diabetic retinopathy through reliable feature selection.

Diabetic retinopathy (DR) is a complication of diabetes mellitus that damages the blood vessels in the retina. DR is considered a serious vision-threatening impediment that most diabetic subjects are at risk of developing. Effective automatic detection of DR is challenging. Feature extraction plays an important role in the effective classification of disease. Here we focus on a feature extraction technique that combines two feature extractors, speeded up robust features and binary robust invariant scalable keypoints, to extract the relevant features from retinal fundus images. The selection of top-ranked features using the MR-MR (maximum relevance-minimum redundancy) feature selection and ranking method enhances the efficiency of classification. The system is evaluated across various classifiers, such as support vector machine, Adaboost, Naive Bayes, Random Forest, and multi-layer perception (MLP) when giving input image features extracted from standard datasets (IDRiD, MESSIDOR, and DIARETDB0). The performances of the classifiers were analyzed by comparing their specificity, precision, recall, false positive rate, and accuracy values. We found that when the proposed feature extraction and selection technique is used together with MLP outperforms all the other classifiers for all datasets in binary and multiclass classification.