Predictive value of optical coherence tomography angiography in management of diabetic macular edema.

BACKGROUND: Optical coherence tomography angiography (OCTA) is a relatively new extension of Optical coherence tomography (OCT) that generates non-invasive, depth-resolved images of the retinal microvasculature which allows for the detection of various features of diabetic retinopathy. OBJECTIVES: This study aimed to detect biomarkers that may predict an early anatomical response to the treatment of diabetic macular edema (DME) with intravitreal ranibizumab (IVR) by means of OCTA. PATIENTS AND METHODS: This prospective interventional study was undertaken on 111 eyes of 102 naïve participants who had diabetic macular edema; enrolled patients were evaluated by taking a complete ophthalmologic history, examination and investigations by use of a pre-designed checklist involving Optical Coherence Tomography Angiography. RESULTS: Regarding the best corrected visual acuity (BCVA) the Mean ± SD was 0.704 ± 0.158 preoperatively and 0.305 ± 0.131 postoperatively in good responder patients; and was 0.661 ± 0.164 preoperatively and 0.54 ± 0.178 postoperatively in poor responders. The central macular thickness (CMT) was 436.22 ± 54.66 µm preoperatively and 308.12 ± 33.09 µm postoperatively in good responder patients; and was 387.74 ± 44.05 µm preoperatively and 372.09 ± 52.86 µm postoperatively in poor responders. By comparing the pre injection size of the foveal avascular zone area (FAZ-A) in both groups, it found that the mean ± SD of FAZ-A was 0.297 ± 0.038 mm in good responder patients compared to 0.407 ± 0.05 mm in non-responder patients. The preoperative superficial capillary plexus (SCP) foveal vascular density (VD) was 24.02 ± 3.01% in good responder patients versus 17.89 ± 3.19% um in poor responders. The preoperative SCP parafoveal VD was 43.06 ± 2.67% in good responder patients versus 37.96 ± 1.82% um in poor responders. The preoperative deep capillary plexus (DCP) foveal VD was 30.58 ± 2.89% in good responder patients versus 25.45 ± 3.14% in poor responders. The preoperative DCP parafoveal VD was 45.66 ± 2.21% in good responder patients versus 43.26 ± 2.35% um in poor responders, this was statistically significant. CONCLUSION: OCTA offers an accurate measurement for VD in the macula as well as the FAZ-A which could be used to predict an early anatomical response of anti-VEGF treatment in DME.

A new texture-based labeling framework for hyper-reflective foci identification in retinal optical coherence tomography images.

An important abnormality in Optical Coherence Tomography (OCT) images is Hyper-Reflective Foci (HRF). This anomaly can be interpreted as a biomarker of serious retinal diseases such as Age-related Macular Degeneration (AMD) and Diabetic Macular Edema (DME) or the progression of disease from an early stage to a late one. In this paper, a new method is proposed for the identification of HRFs. The new method divides the OCT B-scan into patches and separately verifies each patch to determine whether or not the patch contains an HRF. The procedure of patch verification contains a texture-based framework which assigns appropriate labels according to intensity changes to each column and row. Then, a feature vector is extracted for each patch based on the assigned labels. The feature vectors are utilized in the training step of well-known classifiers like Support Vector Machine (SVM). Then, the classifiers are used to produce the labels for the test OCT images. The new method is evaluated on a public dataset including HRF labels. The experimental results show that the new method is capable of providing outstanding results in terms of speed and accuracy.

Semi-Automated Analysis of Dome-Shaped Macula in Preterm and Full-Term Infants Using Handheld Swept-Source Optical Coherence Tomography.

Purpose: Dome-shaped macula (DSM) is known to occur in highly myopic adults and, recently, preterm infants. This study uses investigational handheld swept-source optical coherence tomography (SS-OCT) to further characterize infantile DSM. Methods: In this prospective, observational study, preterm infants undergoing retinopathy of prematurity screening and full-term infants within 72 hours of birth were imaged. Two trained graders assessed macular features, including DSM, subretinal fluid, and macular edema. A semi-automated program measured foveal immaturity, dome height, and diameter. Results: Two hundred seventeen imaging sessions from 50 full-term and 30 preterm infants were included (46% female, preterm birth weight 1038 ± 335 g, and gestational age 28.7 ± 3.1 weeks). DSM occurred in 40% preterm versus 14% full-term infants (P = 0.01). Mean postmenstrual age at first DSM diagnosis was 38.4 ± 0.0 weeks among preterm and 40.4 ± 1.1 weeks among full-term infants (P < 0.001). Dome height and diameter measured 55.67 ± 44.22 µm and 3583.15 ± 1090.35 µm for preterm versus 88.37 ± 44.73 µm and 3581.97 ± 355.07 µm for full-term infants (P = 0.24 and P = 0.96, respectively). All 27 images (11 preterm and 7 full-term infants) with 3-dimensional analysis had round dome configuration. No other associations were seen, including macular fluid (P = 0.17). Conclusions: Infants frequently exhibit DSM without an association with macular fluid. Preterm infants were more likely than full-term infants to have DSM. Unlike DSM in children and adults, infantile DSM configuration is mostly round rather than ridge-shaped.

Generating Synthesized Fluorescein Angiography Images From Color Fundus Images by Generative Adversarial Networks for Macular Edema Assessment.

Purpose: To assess the feasibility of generating synthetic fluorescein angiography (FA) images from color fundus (CF) images using pixel-to-pixel generative adversarial network (pix2pixGANs) for clinical applications. Research questions addressed image realism to retinal specialists and utility for assessing macular edema (ME) in Retinal Vein Occlusion (RVO) eyes. Methods: We used a registration-guided pix2pixGANs method trained on the CF-FA dataset from Kham Eye Centre, Kandze Prefecture People's Hospital. A visual Turing test confirmed the realism of synthetic images without novel artifacts. We then assessed the synthetic FA images for assessing ME. Finally, we quantitatively evaluated the synthetic images using Fréchet Inception distance (FID) and structural similarity measures (SSIM). Results: The raw development dataset had 881 image pairs from 349 subjects. Our approach is capable of generating realistic FA images because small vessels are clearly visible and sharp within one optic disc diameter around the macula. Two retinal specialists agreed that more than 85% of synthetic FA images have good or excellent image quality. For ME detection, accuracy was similar for real and synthetic images. FID demonstrated a 38.9% improvement over the previous state-of-the-art (SOTA), and SSIM reached 0.78 compared to the previous SOTA's 0.67. Conclusions: We developed a pix2pixGANs model translating FA images from label-free CF images, yielding reliable synthetic FA images. This suggests potential for noninvasive evaluation of ME in RVO eyes using pix2pix GANs techniques. Translational Relevance: Pix2pixGANs techniques have the potential to assist in the noninvasive clinical assessment of ME in RVO eyes.

Macular vascular and photoreceptor changes for diabetic macular edema at early stage.

This study was intended to investigate the macular vascular and photoreceptor changes for diabetic macular edema (DME) at the early stage. A total of 255 eyes of 134 diabetes mellitus patients were enrolled and underwent an ophthalmological and systemic evaluation in this cross-sectional study. Early DME was characterized by central subfoveal thickness (CST) value between 250 and 325 µm, intact ellipsoid zone, and an external limiting membrane. While non-DME was characterized by CST < 250 µm with normal retinal morphology and structure. Foveal avascular zone (FAZ) area ≤ 0.3 mm2 (P < 0.001, OR = 0.41, 95% CI 0.26-0.67 in the multivariate analysis) and HbA1c level ≤ 8% (P = 0.005, OR = 0.37, 95% CI 0.19-0.74 in multivariate analysis) were significantly associated with a higher risk of early DME. Meanwhile, no significant differences exist in cone parameters between non-DME and early DME eyes. Compared with non-DME eyes, vessel diameter, vessel wall thickness, wall-to-lumen ratio, the cross-sectional area of the vascular wall in the upper side were significantly decreased in the early DME eyes (P = 0.001, P < 0.001, P = 0.005, P = 0.003 respectively). This study suggested a vasospasm or vasoconstriction with limited further photoreceptor impairment at the early stage of DME formation. CST ≥ 250 µm and FAZ ≤ 0.3 mm2 may be the indicator for early DME detection.

Optical coherence tomography angiography for microaneurysms in anti-vascular endothelial growth factor treated diabetic macular edema.

BACKGROUND: We aimed to evaluate microaneurysms (MAs) after treatment with anti-vascular endothelial growth factor (anti-VEGF) therapy to understand causes of chronic edema and anti-VEGF resistance. METHODS: Patients with non-proliferative diabetic retinopathy, with or without macular edema were recruited. Optical coherence tomography angiography (OCTA) MAs-related parameters were observed, including the maximum diameter of overall dimensions, material presence, and flow signal within the lumen. OCTA parameters also included central macular thickness (CMT), foveal avascular zone, superficial and deep capillary plexuses, and non-flow area measurements on the superficial retinal slab. RESULTS: Overall, 48 eyes from 43 patients were evaluated. CMT differed significantly between the diabetic macular edema (DME ) and non-DME (NDME) groups at 1st, 2nd, 3rd, and 6th months of follow-up (P < 0.001; <0.001; 0.003; <0.001, respectively). A total of 55 and 59 MAs were observed in the DME (mean = 99.40 ± 3.18 µm) and NDME (mean maximum diameter = 74.70 ± 2.86 µm) groups at baseline, respectively (significant between-group difference: P < 0.001). Blood flow signal was measurable for 46 (83.6%) and 34 (59.3%) eyes in the DME and NDME groups, respectively (significant between-group difference: P < 0.001). CONCLUSIONS: Compared to the NDME group, the DME group had larger MAs and a higher blood-flow signal ratio. Following anti-VEGF therapy, changes in the diameter of MAs were observed before changes in CMT thickness.

Prediction of treatment outcome for branch retinal vein occlusion using convolutional neural network-based retinal fluorescein angiography.

Deep learning techniques were used in ophthalmology to develop artificial intelligence (AI) models for predicting the short-term effectiveness of anti-VEGF therapy in patients with macular edema secondary to branch retinal vein occlusion (BRVO-ME). 180 BRVO-ME patients underwent pre-treatment FFA scans. After 3 months of ranibizumab injections, CMT measurements were taken at baseline and 1-month intervals. Patients were categorized into good and poor prognosis groups based on macular edema at the 4th month follow-up. FFA-Net, a VGG-based classification network, was trained using FFA images from both groups. Class activation heat maps highlighted important locations. Benchmark models (DesNet-201, MobileNet-V3, ResNet-152, MansNet-75) were compared for training results. Performance metrics included accuracy, sensitivity, specificity, F1 score, and ROC curves. FFA-Net predicted BRVO-ME treatment effect with an accuracy of 88.63% and an F1 score of 0.89, with a sensitivity and specificity of 79.40% and 71.34%, respectively.The AUC of the ROC curve for the FFA-Net model was 0.71. The use of FFA based on deep learning technology has feasibility in predicting the treatment effect of BRVO-ME. The FFA-Net model constructed with the VGG model as the main body has good results in predicting the treatment effect of BRVO-ME. The typing of BRVO in FFA may be an important factor affecting the prognosis.

The effect of long-term hemodialysis on diabetic retinopathy observed by swept-source optical coherence tomography angiography.

BACKGROUND: Diabetes can cause chronic microvascular complications such as diabetic retinopathy (DR) and diabetic nephropathy (DN). DR and DN can lead to or exacerbate diabetic macular edema (DME). Hemodialysis (HD) is the main treatment method for patients with end-stage kidney disease (ESKD) secondary to DN. PURPOSE: The aim of this prospective cohort study was to determine the immediate effect of single HD session on retinal and choroidal thickness in DR patients with ESKD and the features of DR and the prevalence of DME in these patients who have received long-term HD. METHODS: Eighty-five eyes of 44 DR patients with ESKD who underwent long-term HD were examined by swept-source optical coherence tomography angiography (SS-OCTA). Based on OCTA images, the characteristics of DR and the prevalence of DME in these patients were analyzed. Changes in central retinal thickness (CRT), central retinal volume (CRV), subfoveal choroidal thickness (SFCT) and subfoveal choroidal volume (SFCV) within 30 min before and after single HD session were compared. CRT, CRV, SFCT and SFCV were compared before single HD session and before the next single HD session. RESULTS: There was no significant difference in the average CRT (251.69 ± 39.21 µm vs. 251.46 ± 39.38 µm, P = 0.286) or CRV (0.15 ± 0.62 µm vs. 0.15 ± 0.63 µm, P = 0.324) between before and after single HD session. After single HD session, SFCT (243.11 ± 77.15 µm vs. 219.20 ± 72.84 µm, P < 0.001) and SFCV (0.15 ± 0.10 µm vs. 0.13 ± 0.90 µm, P < 0.001) significantly decreased. There was no statistically significant difference in CRT (251.69 ± 39.21 µm vs. 251.11 ± 38.47 µm, P = 0.206), CRV (0.15 ± 0.62 µm vs. 0.15 ± 0.61 µm, P = 0.154), SFCT (243.11 ± 77.15 µm vs. 245.41 ± 76.23 µm, P = 0.108), or SFCV (0.15 ± 0.10 µm vs. 0.16 ± 0.10 µm, P = 0.174) before HD and before the next single HD session. On en face OCTA images, eighty-five eyes (100%) had retinal nonperfusion areas, foveal avascular zone (FAZ) enlargement, and abnormal retinal microvasculature. Based on cross-sectional OCTA images, retinal neovascularization (RNV) was confirmed in 42 eyes (49.41%), and intraretinal microvascular abnormalities (IRMAs) were detected in 85 eyes (100%). Seventeen eyes (20%) still had DME, all of which were cystoid macular edema (CME). Among eyes with DME, the epiretinal membrane (ERM) was present in 7 eyes (8.24%). CONCLUSIONS: For DR patients with ESKD who have undergone long-term HD, the choroidal thickness still changes significantly before and after single HD session, which may be related to short-term effects such as reduced blood volume and plasma osmotic pressure caused by single HD session. Although macular features seem to have stabilized in DR patients undergoing long-term dialysis, the DR of patients with ESKD should still be given attention.

Evaluation of thickness of individual macular retinal layers in diabetic eyes from optical coherence tomography.

The effect of diabetes mellitus (DM) on individual retinal layers remains incompletely understood. We evaluated the intra-retinal layer thickness alterations in 71 DM eyes with no diabetic retinopathy (DR), 90 with mild DR, and 63 with moderate DR without macular edema, using spectral-domain optical coherence tomography (SD-OCT) and the Iowa Reference Algorithm for automated retinal layer segmentation. The average thickness of 10 intra-retinal layers was then corrected for ocular magnification using axial length measurements, and pairwise comparisons were made using multivariable linear regression models adjusted for gender and race. In DM no DR eyes, significant thinning was evident in the ganglion cell layer (GCL; p < 0.001), inner nuclear layer (INL; p = 0.001), and retinal pigment epithelium (RPE; p = 0.014) compared to normal eyes. Additionally, mild DR eyes exhibited a thinner inner plexiform layer (IPL; p = 0.008) than DM no DR eyes. Conversely, moderate DR eyes displayed thickening in the INL, outer nuclear layer, IPL, and retinal nerve fiber layer (all p ≤ 0.002), with notably worse vision. These findings highlight distinctive patterns: early diabetic eyes experience thinning in specific retinal layers, while moderate DR eyes exhibit thickening of certain layers and slightly compromised visual acuity, despite the absence of macular edema. Understanding these structural changes is crucial for comprehending diabetic eye complications.

A novel approach for automatic classification of macular degeneration OCT images.

Age-related macular degeneration (AMD) and diabetic macular edema (DME) are significant causes of blindness worldwide. The prevalence of these diseases is steadily increasing due to population aging. Therefore, early diagnosis and prevention are crucial for effective treatment. Classification of Macular Degeneration OCT Images is a widely used method for assessing retinal lesions. However, there are two main challenges in OCT image classification: incomplete image feature extraction and lack of prominence in important positional features. To address these challenges, we proposed a deep learning neural network model called MSA-Net, which incorporates our proposed multi-scale architecture and spatial attention mechanism. Our multi-scale architecture is based on depthwise separable convolution, which ensures comprehensive feature extraction from multiple scales while minimizing the growth of model parameters. The spatial attention mechanism is aim to highlight the important positional features in the images, which emphasizes the representation of macular region features in OCT images. We test MSA-NET on the NEH dataset and the UCSD dataset, performing three-class (CNV, DURSEN, and NORMAL) and four-class (CNV, DURSEN, DME, and NORMAL) classification tasks. On the NEH dataset, the accuracy, sensitivity, and specificity are 98.1%, 97.9%, and 98.0%, respectively. After fine-tuning on the UCSD dataset, the accuracy, sensitivity, and specificity are 96.7%, 96.7%, and 98.9%, respectively. Experimental results demonstrate the excellent classification performance and generalization ability of our model compared to previous models and recent well-known OCT classification models, establishing it as a highly competitive intelligence classification approach in the field of macular degeneration.

Optimized deep CNN for detection and classification of diabetic retinopathy and diabetic macular edema.

Diabetic Retinopathy (DR) and Diabetic Macular Edema (DME) are vision related complications prominently found in diabetic patients. The early identification of DR/DME grades facilitates the devising of an appropriate treatment plan, which ultimately prevents the probability of visual impairment in more than 90% of diabetic patients. Thereby, an automatic DR/DME grade detection approach is proposed in this work by utilizing image processing. In this work, the retinal fundus image provided as input is pre-processed using Discrete Wavelet Transform (DWT) with the aim of enhancing its visual quality. The precise detection of DR/DME is supported further with the application of suitable Artificial Neural Network (ANN) based segmentation technique. The segmented images are subsequently subjected to feature extraction using Adaptive Gabor Filter (AGF) and the feature selection using Random Forest (RF) technique. The former has excellent retinal vein recognition capability, while the latter has exceptional generalization capability. The RF approach also assists with the improvement of classification accuracy of Deep Convolutional Neural Network (CNN) classifier. Moreover, Chicken Swarm Algorithm (CSA) is used for further enhancing the classifier performance by optimizing the weights of both convolution and fully connected layer. The entire approach is validated for its accuracy in determination of grades of DR/DME using MATLAB software. The proposed DR/DME grade detection approach displays an excellent accuracy of 97.91%.

Evaluating Ocular Blood Flow in Diabetic Macular Edema using Three-dimensional Pseudocontinuous Arterial Spin Labeling.

BACKGROUND: Alterations in ocular blood flow play an important role in the pathogenesis of diabetic macular edema; however, this remains unclear. OBJECTIVES: This study aimed to investigate ocular blood flow in eyes with or without diabetic macular edema using arterial spin labeling. METHODS: This cross-sectional study included 118 eyes of 65 patients with diabetic retinopathy analyzed between November 2018 and December 2019. We included a total of 53 eyes without diabetic macular edema (mean [SD] age, 57.83 [7.23] years; 29 men [54.7%]) and 65 eyes with diabetic macular edema (mean [SD] age, 60.11 [7.63] years; 38 men [58.5%]). Using a 3.0-T magnetic resonance imaging, participants were imaged with arterial spin labeling with multiple post-labeling delays. RESULTS: The mean ocular blood flow at post-labeling delays of 1.5 and 2.5 s was significantly lower in eyes with diabetic macular edema among patients with diabetic retinopathy compared with the remaining subgroups (P=0.022 and P <0.001, respectively). The mean ocular blood flow exhibited a significant decrease in eyes with diabetic macular edema when the post-labeling delay was set at 2.5 s in the nonproliferative and proliferative diabetic retinopathy groups, compared with the remaining subgroups (P=0.005 and P=0.002, respectively). The cutoff points of ocular blood flow at post-labeling delays of 1.5 s and 2.5 s were 9.40 and 11.10 mL/100 g/min, respectively. CONCLUSION: Three-dimensional pseudocontinuous arterial spin labeling can identify differences in the ocular blood flow of patients with diabetic eyes with and without diabetic macular edema.

Prediction of Long-Term Treatment Outcomes for Diabetic Macular Edema Using a Generative Adversarial Network.

Purpose: The purpose of this study was to analyze optical coherence tomography (OCT) images of generative adversarial networks (GANs) for the prediction of diabetic macular edema after long-term treatment. Methods: Diabetic macular edema (DME) eyes (n = 327) underwent anti-vascular endothelial growth factor (VEGF) treatments every 4 weeks for 52 weeks from a randomized controlled trial (CRTH258B2305, KINGFISHER) were included. OCT B-scan images through the foveal center at weeks 0, 4, 12, and 52, fundus photography, and retinal thickness (RT) maps were collected. GAN models were trained to generate probable OCT images after treatment. Input for each model were comprised of either the baseline B-scan alone or combined with additional OCT, thickness map, or fundus images. Generated OCT B-scan images were compared with real week 52 images. Results: For 30 test images, 28, 29, 15, and 30 gradable OCT images were generated by CycleGAN, UNIT, Pix2PixHD, and RegGAN, respectively. In comparison with the real week 52, these GAN models showed positive predictive value (PPV), sensitivity, specificity, and kappa for residual fluid ranging from 0.500 to 0.889, 0.455 to 1.000, 0.357 to 0.857, and 0.537 to 0.929, respectively. For hard exudate (HE), they were ranging from 0.500 to 1.000, 0.545 to 0.900, 0.600 to 1.000, and 0.642 to 0.894, respectively. Models trained with week 4 and 12 B-scans as additional inputs to the baseline B-scan showed improved performance. Conclusions: GAN models could predict residual fluid and HE after long-term anti-VEGF treatment of DME. Translational Relevance: The implementation of this tool may help identify potential nonresponders after long-term treatment, thereby facilitating management planning for these eyes.

A hybrid model for the detection of retinal disorders using artificial intelligence techniques.

The prevalence of vision impairment is increasing at an alarming rate. The goal of the study was to create an automated method that uses optical coherence tomography (OCT) to classify retinal disorders into four categories: choroidal neovascularization, diabetic macular edema, drusen, and normal cases. This study proposed a new framework that combines machine learning and deep learning-based techniques. The utilized classifiers were support vector machine (SVM), K-nearest neighbor (K-NN), decision tree (DT), and ensemble model (EM). A feature extractor, the InceptionV3 convolutional neural network, was also employed. The performance of the models was evaluated against nine criteria using a dataset of 18000 OCT images. For the SVM, K-NN, DT, and EM classifiers, the analysis exhibited state-of-the-art performance, with classification accuracies of 99.43%, 99.54%, 97.98%, and 99.31%, respectively. A promising methodology has been introduced for the automatic identification and classification of retinal disorders, leading to reduced human error and saved time.

A deep learning approach to hard exudates detection and disorganization of retinal inner layers identification on OCT images.

The purpose of the study was to detect Hard Exudates (HE) and classify Disorganization of Retinal Inner Layers (DRIL) implementing a Deep Learning (DL) system on optical coherence tomography (OCT) images of eyes with diabetic macular edema (DME). We collected a dataset composed of 442 OCT images on which we annotated 6847 HE and the presence of DRIL. A complex operational pipeline was defined to implement data cleaning and image transformations, and train two DL models. The state-of-the-art neural network architectures (Yolov7, ConvNeXt, RegNetX) and advanced techniques were exploited to aggregate the results (Ensemble learning, Edge detection) and obtain a final model. The DL approach reached good performance in detecting HE and classifying DRIL. Regarding HE detection the model got an AP@0.5 score equal to 34.4% with Precision of 48.7% and Recall of 43.1%; while for DRIL classification an Accuracy of 91.1% with Sensitivity and Specificity both of 91.1% and AUC and AUPR values equal to 91% were obtained. The P-value was lower than 0.05 and the Kappa coefficient was 0.82. The DL models proved to be able to identify HE and DRIL in eyes with DME with a very good accuracy and all the metrics calculated confirmed the system performance. Our DL approach demonstrated to be a good candidate as a supporting tool for ophthalmologists in OCT images analysis.

Hybrid deep learning models for the screening of Diabetic Macular Edema in optical coherence tomography volumes.

Several studies published so far used highly selective image datasets from unclear sources to train computer vision models and that may lead to overestimated results, while those studies conducted in real-life remain scarce. To avoid image selection bias, we stacked convolutional and recurrent neural networks (CNN-RNN) to analyze complete optical coherence tomography (OCT) cubes in a row and predict diabetic macular edema (DME), in a real-world diabetic retinopathy screening program. A retrospective cohort study was carried out. Throughout 4-years, 5314 OCT cubes from 4408 subjects who attended to the diabetic retinopathy (DR) screening program were included. We arranged twenty-two (22) pre-trained CNNs in parallel with a bidirectional RNN layer stacked at the bottom, allowing the model to make a prediction for the whole OCT cube. The staff of retina experts built a ground truth of DME later used to train a set of these CNN-RNN models with different configurations. For each trained CNN-RNN model, we performed threshold tuning to find the optimal cut-off point for binary classification of DME. Finally, the best models were selected according to sensitivity, specificity, and area under the receiver operating characteristics curve (AUROC) with their 95% confidence intervals (95%CI). An ensemble of the best models was also explored. 5188 cubes were non-DME and 126 were DME. Three models achieved an AUROC of 0.94. Among these, sensitivity, and specificity (95%CI) ranged from 84.1-90.5 and 89.7-93.3, respectively, at threshold 1, from 89.7-92.1 and 80-83.1 at threshold 2, and from 80.2-81 and 93.8-97, at threshold 3. The ensemble model improved these results, and lower specificity was observed among subjects with sight-threatening DR. Analysis by age, gender, or grade of DME did not vary the performance of the models. CNN-RNN models showed high diagnostic accuracy for detecting DME in a real-world setting. This engine allowed us to detect extra-foveal DMEs commonly overlooked in other studies, and showed potential for application as the first filter of non-referable patients in an outpatient center within a population-based DR screening program, otherwise ended up in specialized care.

A health economic pilot study comparing two diabetic retinopathy screening strategies.

To compare two screening strategies for diabetic retinopathy (DR), and to determine the health-economic impact of including optical coherence tomography (OCT) in a regular DR screening. This cross-sectional study included a cohort of patients (≥ 18 years) with type 1 or 2 diabetes mellitus (T1D or T2D) from a pilot DR screening program at Oslo University Hospital, Norway. A combined screening strategy where OCT was performed in addition to fundus photography for all patients, was conducted on this cohort and compared to our existing sequential screening strategy. In the sequential screening strategy, OCT was performed on a separate day only if fundus photography indicated diabetic macular edema (DME). The presence of diabetic maculopathy on fundus photography and DME on OCT was determined by two medical retina specialists. Based on the prevalence rate of diabetic maculopathy and DME from the pilot, we determined the health-economic impact of the two screening strategies. The study included 180 eyes of 90 patients. Twenty-seven eyes of 18 patients had diabetic maculopathy, and of these, 7 eyes of 6 patients revealed DME on OCT. When diabetic maculopathy was absent on fundus photographs, OCT could not reveal DME. Accordingly, 18 patients (20%) with diabetic maculopathy would have needed an additional examination with OCT in the sequential screening strategy, 6 (33%) of whom would have had DME on OCT. In an extended healthcare perspective analysis, the cost of the sequential screening strategy was higher than the cost of the combined screening strategy. There was a weak association between diabetic maculopathy on fundus photography and DME on OCT. The health economic analysis suggests that including OCT as a standard test in DR screening could potentially be cost-saving.

Applications of artificial intelligence in diagnosis of uncommon cystoid macular edema using optical coherence tomography imaging: A systematic review.

Cystoid macular edema (CME) is a sight-threatening condition often associated with inflammatory and diabetic diseases. Early detection is crucial to prevent irreversible vision loss. Artificial intelligence (AI) has shown promise in automating CME diagnosis through optical coherence tomography (OCT) imaging, but its utility needs critical evaluation. This systematic review assesses the application of AI to diagnosis CME, specifically focusing on disorders like postoperative CME (Irvine Gass syndrome) and retinitis pigmentosa without obvious vasculopathy, using OCT imaging. A comprehensive search was conducted across 6 databases (PubMed, Scopus, Web of Science, Wiley, ScienceDirect, and IEEE) from 2018 to November, 2023. Twenty-three articles met the inclusion criteria and were selected for in-depth analysis. We evaluate AI's role in CME diagnosis and its performance in "detection", "classification", and "segmentation" of OCT retinal images. We found that convolutional neural network (CNN)-based methods consistently outperformed other machine learning techniques, achieving an average accuracy of over 96 % in detecting and identifying CME from OCT images. Despite certain limitations such as dataset size and ethical concerns, the synergy between AI and OCT, particularly through CNNs, holds promise for significantly advancing CME diagnostics.

Exploration on OCT biomarker candidate related to macular edema caused by diabetic retinopathy and retinal vein occlusion in SD-OCT images.

To improve the understanding of potential pathological mechanisms of macular edema (ME), we try to discover biomarker candidates related to ME caused by diabetic retinopathy (DR) and retinal vein occlusion (RVO) in spectral-domain optical coherence tomography images by means of deep learning (DL). 32 eyes of 26 subjects with non-proliferative DR (NPDR), 77 eyes of 61 subjects with proliferative DR (PDR), 120 eyes of 116 subjects with branch RVO (BRVO), and 17 eyes of 15 subjects with central RVO (CRVO) were collected. A DL model was implemented to guide biomarker candidate discovery. The disorganization of the retinal outer layers (DROL), i.e., the gray value of the retinal tissues between the external limiting membrane (ELM) and retinal pigment epithelium (RPE), the disrupted and obscured rate of the ELM, ellipsoid zone (EZ), and RPE, was measured. In addition, the occurrence, number, volume, and projected area of hyperreflective foci (HRF) were recorded. ELM, EZ, and RPE are more likely to be obscured in RVO group and HRFs are observed more frequently in DR group (all P ≤ 0.001). In conclusion, the features of DROL and HRF can be possible biomarkers related to ME caused by DR and RVO in OCT modality.

Hypertension Likely Drives Arteriolar Wall Thickening in Preclinical Diabetic Retinopathy While Diabetes Drives Wall Thickness in Clinical Retinopathy.

Purpose: Both hypertension and diabetes are known to increase the wall-to-lumen ratio (WLR) of retinal arterioles, but the differential effects are unknown. Here, we study the timing and relative impact of hypertension versus diabetes on the WLR in diabetic retinopathy (DR) to address this unresolved question. Methods: This prospective cross-sectional study compared the retinal arteriolar WLR in 17 healthy eyes, 15 with diabetes but no apparent DR (DM no DR), and 8 with diabetic macular edema (DME) and either nonproliferative or proliferative DR. We imaged each arteriole using adaptive optics scanning laser ophthalmoscopy and measured the WLR using ImageJ. Multiple linear regression (MLR) was performed to estimate the effects of hypertension, diabetes, and age on the WLR. Results: Both subjects with DM no DR and subjects with DME had significantly higher WLR than healthy subjects (0.36 ± 0.08 and 0.42 ± 0.08 vs. 0.29 ± 0.07, 1-way ANOVA P = 0.0009). MLR in healthy subjects and subjects with DM no DR showed hypertension had the strongest effect (regression coefficient = 0.08, P = 0.009), whereas age and diabetes were not significantly correlated with WLR. MLR in all three groups together (healthy, DM no DR, and DME) showed diabetes had the strongest effect (regression coefficient = 0.05, P = 0.02), whereas age and hypertension were not significantly correlated with WLR. Conclusions: Hypertension may be an early driver of retinal arteriolar wall thickening in preclinical DR, independent of age or diabetes, whereas changes specific to DR may drive wall thickening in DME and later DR stages. Translational Relevance: We offer a framework for understanding the relative contributions of hypertension and diabetes on the vascular wall, and emphasize the importance of hypertension control early in diabetes even before DR onset.