Comparison of functional changes of retina after subthreshold and threshold pan-retinal photocoagulation in severe non-proliferative diabetic retinopathy.

PURPOSE: To find a new approach of pan-retinal photocoagulation (PRP) with less damage to the retina in the treatment of severe non-proliferative diabetic retinopathy (NPDR), this study compared functional changes in the retina after subthreshold and threshold PRP treatment in severe NPDR eyes. METHODS: Post hoc analysis of a randomized clinical trial was conducted in this study. Seventy eyes of 35 patients with bilateral, symmetric, severe NPDR were enrolled. Two eyes from the same patient were randomized into two groups, one eye received subthreshold PRP (S-PRP) and the other eye received threshold PRP (T-PRP). Comprehensive ophthalmological evaluations were performed on the baseline and every 3 months for 1 year. Visual field (VF) and full-field electroretinography (ERG) were performed on the baseline and repeated at month 12. RESULTS: During the 12-month follow-up, 4 eyes (11.4%) in the S-PRP group and 3 eyes (8.6%) in the T-PRP group progressed to proliferative diabetic retinopathy (PDR) stage, and there was no statistical difference in PDR progression rate between the two groups (P = 0.69). In addition, the changes in best-corrected visual acuity (BCVA) from baseline to month 12 between the two groups had no statistical difference (P = 0.30). From baseline to month 12, changes in central VF between the two groups had no statistical difference (P = 0.25), but changes in total score points of peripheral VF in the S-PRP group (- 242.1 ± 210.8 dB) and the T-PRP group (- 308.9 ± 209.7 dB) were statistically significant (P = 0.03). At month 12, ERG records showed that the amplitude of dark-adapted 0.01 ERG, dark-adapted 3.0 ERG, oscillatory potentials, light-adapted 3.0 ERG, and 30 Hz flicker ERG of both groups were significantly decreased from the baseline (P < 0.05). In addition, the amplitude of each ERG record in the S-PRP group decreased significantly less than those in the T-PRP group (P < 0.05). CONCLUSIONS: Subthreshold PRP is as effective as threshold PRP for preventing severe NPDR progress to PDR within 1 year with less damage to periphery VF and retinal function. CLINICALTRIALS: gov Identifier: NCT01759121.

Subthreshold Pan-Retinal Photocoagulation Using Endpoint Management Algorithm for Severe Nonproliferative Diabetic Retinopathy: A Paired Controlled Pilot Prospective Study.

PURPOSE: The aim of this study was to report the efficacy and safety profile of subthreshold pan-retinal photocoagulation (PRP) using endpoint management (EPM) algorithm compared with conventional threshold PASCAL PRP for the treatment of severe nonproliferative diabetic retinopathy (NPDR). METHODS: This was a prospective, single-center, paired randomized controlled trial of 56 eyes of 28 participants with bilateral symmetric severe NPDR. One eye of the participant was randomly assigned to receive the subthreshold EPM PRP, while the other eye of the same participant received the threshold PASCAL PRP. The primary outcome measures included the difference in the 1-year risk of progression to PDR between 2 groups, and mean changes of the logarithm of the minimal angle of resolution (logMAR) visual acuity (VA). The second outcome measures included central foveal thickness (CFT), 1-year risk of progression to PDR, and visual field (VF) parameters. RESULTS: The subthreshold EPM PRP group and the threshold PASCAL PRP group had similar 1-year risk of progression to PDR during the 12-month follow-up visits (17.86 vs. 14.29%, p > 0.05). Slightly decreased VA was found in both groups (0.08 vs. 0.09 logMAR VA); however, no statistical difference was found for neither group (p > 0.05). Similar results were found for thickened CFT for both groups (23.59 vs. 28.34 µm, p > 0.05). Specifically, although substantial loss of VF was found in the threshold PASCAL PRP group (p < 0.05), no obvious damage to VF was seen in the subthreshold EPM PRP group (p > 0.05). CONCLUSION: The subthreshold EPM PRP is noninferior to the conventional threshold PASCAL PRP in the treatment of severe NPDR during 12-month follow-up and could be an alternative treatment option for patients with severe NPDR.

Dynamic changes and correlation analysis of outer retinal microstructure in macular area of central serous chorioretinopathy patients during restoration period.

PURPOSE: To investigate the dynamic changes and possible affecting variables of outer retinal microstructure in macular area of central serous chorioretinopathy (CSC) patients. METHODS: This was a retrospective study. The data of optical coherence tomography (OCT) and autofluoroscopy (AF) of 36 CSC patients admitted to our hospital from February 2012 to February 2013 were included. Dynamic variations and possible correlated variables of central retinal thickness (CRT), subretinal fluid diameter (SRFD), ellipsoid zone (EZ), interdigitation zone (IZ) and/or hyperautofluorescent spot (HAS) were analyzed. RESULTS: The outer retinal microstructure was gradually restored along with the subretinal fluid absorption during the follow-up. EZ in 94.4% (34/36) and the IZ in 100% (36/36) eyes were completely disappeared at baseline and restored (completed or incomplete) in 88.9% (8/9) and 44.4% (4/9) eyes, respectively, after 6-month follow-up. HAS was evident in 25% eyes (8/32 eyes) at baseline, and the density was initially increased and then declined during follow-up. Correlation analysis demonstrated that the restoration of EZ and IZ was correlated with the restoration period and subretinal fluid absorption. CONCLUSION: The outer retinal microstructure was restored during the subretinal fluid absorption in CSC patients, with EZ restored earlier than IZ. The restoration period and the absorption of subretinal fluid were two closely correlated variables of macular microstructure restoration.

Anatomical and functional responses in eyes with diabetic macular edema treated with "1 + PRN" ranibizumab: one-year outcomes in population of mainland China.

BACKGROUND: To evaluate the anatomical and functional responses in eyes with diabetic macular edema (DME) treated with ranibizumab under "1 + pro re nata (PRN)" regimen. METHODS: This prospective interventional case series included 69 eyes of 69 patients with DME treated with intravitreal injections of 0.5 mg ranibizumab followed by repeated injections as needed. Best-corrected visual acuity (BCVA), central foveal thickness (CFT), subfoveal choroidal thickness (SFCT), and predictive factors for final visual outcomes were assessed. RESULTS: Logarithm of minimal angle of resolution (logMAR) BCVA improved from 0.64 ± 0.23 at baseline to 0.56 ± 0.27, 0.53 ± 0.26, 0.47 ± 0.25, 0.44 ± 0.32, 0.47 ± 0.26 and 0.46 ± 0.26 at time-point of months 1, 2, 3, 6, 9, and 12, respectively (P < 0.05 for any follow-up time-point except month 1). CFT decreased from 478.23 ± 172.31 µm at baseline to 349.74 ± 82.21 µm, 313.52 ± 69.62 µm, 292.59 ± 61.07 µm, 284.67 ± 69.85 µm, 268.33 ± 43.03 µm, and 270.39 ± 49.27 µm at above time-points, respectively (P < 0.05). The number of injections was 6.83 times over 12 months' follow-up under "1 + PRN" regimen. Multivariate analysis showed that the factors including age, BCVA at baseline, disruption of ellipsoid zone, posterior vitreous detachment (PVD), and vitreomacular traction (VMT) were correlated with the final BCVA. CONCLUSIONS: Intravitreal injections of ranibizumab under "1 + PRN" regimen is a not only effective but also safe way to improve visual acuity of DME patients. And older age, lower baseline BCVA, VMT, and disruption of ellipsoid zone are predictors for final poor BCVA while PVD is a positive predictive factor for good final BCVA. TRIAL REGISTRATION: The trial was registered retrospectively in ClinicalTrials.gov on 2 June 2019 (NCT03973138).

Effectiveness and Safety of Intravitreal Injection of Conbercept as an Initial Treatment for Exudative Circumscribed Choroidal Hemangioma.

OBJECTIVE: To investigate the effectiveness and safety of intravitreal injection of conbercept (IVC) as the initial treatment for exudative circumscribed choroidal hemangioma (CCH). METHODS: Forty-two eyes of 42 patients received 3 monthly IVC (0.5 mg/0.05 mL) as the initial treatment. Three months later, the patients were assessed for further treatment including observation, reinjection of conbercept, laser photocoagulation (if the lesion was 3,000 µm away from the macular fovea), or photodynamic therapy (PDT; if the lesion was under the macular fovea). Anatomical and functional responses including best corrected visual acuity (BCVA), central foveal thickness (CFT), and tumor size were analyzed. RESULTS: Twenty-three patients (54.76%) were sensitive to the monotherapy of IVC. Fourteen patients (33.33%) were insensitive to IVC and underwent rescue laser photocoagulation, and 5 patients (11.90%) underwent rescue PDT due to insensitivity to IVC treatment at 3 months. For subgroup analysis, although no statistical difference was found for BCVA at any follow-up time point compared to baseline, an increasing tendency of BCVA was found in the IVC group (p> 0.05). The mean CFT decreased significantly from 427.13 ± 214.74 µm at baseline to 259.83 ± 61.68 µm at 6 months in the IVC group (p< 0.05). No influence on tumor size was found in the IVC group. CONCLUSION: IVC as the initial treatment might be an option for exudative CCH.

APRPG-modified nanoliposome loaded with miR-146a-5p inhibitor suppressed choroidal neovascularization by targeting endothelial cells.

INTRODUCTION: To explore the effect and mechanism of APRPG-modified nanoliposomes loaded with miR-146a-5p inhibitor (ANL-miR-146a-5p inhibitor) on endothelial cell proliferation, migration, tube formation, and choroidal neovascularization (CNV) in mice. METHODS: ANL-miR-146a-5p inhibitors were generated by thin film hydration; in vitro, endothelial cell uptake experiment was used to detect the targeting effect of ANL-miR-146a-5p inhibitor; endothelial cells proliferation, migration, and tube formation were detected, respectively, by CCK8 assay, scratch assay, and Matrigel tube formation assay. In vivo, the mice CNV models were established by 810 nm laser photocoagulation. Mice choroidal flatmounts were performed to detect the volume of CNV after intravitreal injection of L-miR-146a-5p inhibitor, ANL-miR-146a-5p inhibitor, or normal saline; the vascular endothelial growth factor (VEGF) expression of mice choroidal tissue was detected by ELISA; HE section and electrophysiology (ERG) were performed to check the toxicity of ANL-miR-146a-5p inhibitor on mice retina. RESULTS: ANL are targeted to endothelial cells and are more targeted in inflammatory environment. At the same concentration, ANL-miR-146a-5p inhibitor's ability to inhibit endothelial cell proliferation, migration, tube formation, CNV formation, and VEGF expression is better than L-miR-146a-5p inhibitor (p < 0.05); ANL-miR-146a-5p inhibitor had no toxicity on the structure of mouse retina; ANL-miR-146a-5p inhibitor had no toxicity on the a-wave and b-wave amplitudes and b-wave implicit times (p > 0.05). CONCLUSIONS: ANL-miR-146a-5p inhibitor can more effectively down-regulate the expression level of VEGF through its targeting to endothelial cells, thereby more effectively inhibiting endothelial cell proliferation, migration, tube formation, and mice CNV formation.

A pilot prospective study of 577-nm yellow subthreshold micropulse laser treatment with two different power settings for acute central serous chorioretinopathy.

To compare the efficacy of 50% threshold power with 25% threshold power of 577-nm subthreshold micropulse laser (SMPL) for acute central serous chorioretinopathy (CSC). Prospective, interventional, non-randomized, comparative case series. A total of 54 patients (54 eyes) with acute CSC were enrolled. Twenty-four eyes received 25% threshold power and 30 eyes received 50% threshold power of 577-nm SMPL. Best-corrected visual acuity (BCVA), central macular thickness (CMT), and complete absorption of subretinal fluid (SRF) were evaluated at 1 month and 3 months. The complete absorption rate of SRF in the 50% power group was significantly greater than that in the 25% power group at 1 month (70.0% vs 25.0%, p < 0.001) and at 3 months (83.3% vs 54.2%, p < 0.001). Mean BCVA improved from 0.34 ± 0.20 LogMAR to 0.02 ± 0.13 LogMAR in the 50% power group and from 0.27 ± 0.15 LogMAR to 0.14 ± 0.21 LogMAR in the 25% power group with a significant difference between the two groups after 3 months (p = 0.027). In the 50% power group, the CMT decreased from 491.6 ± 154.8 µm at baseline to 231.3 ± 92.3 µm at 1 month and 228.2 ± 88.1 µm at 3 months, and in the 25% power group, the CMT decreased from 444.9 ± 164.1 to 306.8 ± 102.6 µm at 1 month and 254.5 ± 101.7 µm at 3 months. There was statistical difference of CMT at 1 month (p = 0.009) but no significant difference at 3 months between the two groups (p = 0.232). SMPL with 50% threshold power may be more effective than 25% threshold power for acute CSC.

Comparison of the effects of photodynamic therapy, intravitreal ranibizumab and combination for polypoidal choroidal vasculopathy under 1 + PRN regimen.

BACKGROUND: The optimal treatment for polypoidal choroidal vasculopathy (PCV) is still under debate. Little knowledge is known about the treatment effect of "1+pro re nata(PRN)" treatment regimen for PCV. The aim of this study was to compare the outcomes of photodynamic therapy (PDT), intravitreal ranibizumab injection (IVR) and combination therapy under the "1 + PRN" treatment regimen for PCV. METHODS: Fifty-seven eyes of 57 patients completed the 12 months' follow-up in this prospective study. The patients in the PDT arm(n = 23), ranibizumab arm(n = 18), or combination arm(n = 16) underwent a session of PDT, IVR or combination of both at baseline followed by additional IVR as needed. Mean change of logarithm of the minimal angle of resolution (logMAR) visual acuity (VA), central foveal thickness (CFT) and the regression rate of polyps were evaluated. Cost-benefit analysis was also performed. RESULTS: At Month 12, the mean logMAR VA improved from 0.90 ± 0.52 to 0.75 ± 0.57 in the PDT group (P < 0.05), from 0.96 ± 0.58 to 0.77 ± 0.41 in the IVR group (P < 0.05), and from 0.94 ± 0.55 to 0.72 ± 0.44 in the combination group (P < 0.05), respectively. The CFT decreased from 478.04 ± 156.70 µm, 527.5 ± 195.90 µm, and 522.63 ± 288.40 µm at the baseline to 366.43 ± 148.28 µm, 373.17 ± 134.88 µm and 328.44 ± 103.25 in the PDT group (P < 0.05), IVR group (P < 0.01), and the combination group (P < 0.05), respectively. However, no statistical difference was found between groups (P > 0.05). PDT treatment (60.87%) was superior to the IVR therapy (22.22%) in achieving complete regression of polyps (P < 0.05). Cost-benefit analysis showed that IVR treatment cost the least money for improving per 0.1logMAR units and the combination therapy demanded the least money for reducing per 100 µm of CFT. CONCLUSIONS: PDT, IVR and the combination therapy have similar efficacy in the VA improvement as well as the reduction of CFT under the "1 + PRN" treatment regimen. TRIAL REGISTRATION: Current Controlled Trials NCT03459144 . Registered retrospectively on March 2, 2018.

Quantitative assessment of retinochoroidal microvasculature in patients with carotid artery stenosis using OCT angiography.

PURPOSE: To investigate the alterations in retinochoroidal parameters measured by optical coherence tomography (OCT) and OCT angiography (OCTA) in patients with carotid artery stenosis (CAS) and assess their associations with digital subtraction angiography (DSA) data. METHOD: This study enrolled patients diagnosed with CAS and age-matched healthy controls. Both groups underwent OCT and OCTA examinations. DSA and assessment of carotid artery stenosis were performed only in the CAS group. The study evaluated various retinochoroidal parameters from OCT and OCTA, including linear vessel density (LVD), foveal avascular zone (FAZ), choroidal thickness (ChT), and retinal nerve fiber layer (RNFL) thickness. DSA-derived measures included cervical segment (C1) diameter, cavernous segment (C4) diameter, stenosis percentage, ophthalmic artery (OA) filling time, C1-OA filling time, and residual stenosis. RESULTS: A total of 42 eyes from 30 CAS patients and 60 eyes from 30 healthy controls were included. Patients with CAS displayed significantly decreased LVD compared to controls (p < 0.001). Additionally, the CAS group had thinner choroid and RNFL (p = 0.047 and p < 0.001, respectively). Macular LVD negatively correlated with both stenosis percentage and C1-OA filling time (p = 0.010 and p = 0.014, respectively). In patients who underwent carotid artery stenting, preoperative ChT significantly correlated with residual stenosis (Pearson r = -0.480, p = 0.020). CONCLUSION: OCT and OCTA provide a quantitative assessment of retinochoroidal microstructural changes associated with CAS, suggesting potential for noninvasive evaluation of the disease. This might contribute to the prevention of irreversible ocular complications and early detection of CAS. Furthermore, ChT may not only aid in diagnosing CAS more reliably but also offer prognostic information.

Digital ray: enhancing cataractous fundus images using style transfer generative adversarial networks to improve retinopathy detection.

BACKGROUND/AIMS: The aim of this study was to develop and evaluate digital ray, based on preoperative and postoperative image pairs using style transfer generative adversarial networks (GANs), to enhance cataractous fundus images for improved retinopathy detection. METHODS: For eligible cataract patients, preoperative and postoperative colour fundus photographs (CFP) and ultra-wide field (UWF) images were captured. Then, both the original CycleGAN and a modified CycleGAN (C2ycleGAN) framework were adopted for image generation and quantitatively compared using Frechet Inception Distance (FID) and Kernel Inception Distance (KID). Additionally, CFP and UWF images from another cataract cohort were used to test model performances. Different panels of ophthalmologists evaluated the quality, authenticity and diagnostic efficacy of the generated images. RESULTS: A total of 959 CFP and 1009 UWF image pairs were included in model development. FID and KID indicated that images generated by C2ycleGAN presented significantly improved quality. Based on ophthalmologists' average ratings, the percentages of inadequate-quality images decreased from 32% to 18.8% for CFP, and from 18.7% to 14.7% for UWF. Only 24.8% and 13.8% of generated CFP and UWF images could be recognised as synthetic. The accuracy of retinopathy detection significantly increased from 78% to 91% for CFP and from 91% to 93% for UWF. For retinopathy subtype diagnosis, the accuracies also increased from 87%-94% to 91%-100% for CFP and from 87%-95% to 93%-97% for UWF. CONCLUSION: Digital ray could generate realistic postoperative CFP and UWF images with enhanced quality and accuracy for overall detection and subtype diagnosis of retinopathies, especially for CFP.\ TRIAL REGISTRATION NUMBER: This study was registered with ClinicalTrials.gov (NCT05491798).

Prediction of OCT images of short-term response to anti-VEGF treatment for diabetic macular edema using different generative adversarial networks.

PURPOSE: This study sought to assess the predictive performance of optical coherence tomography (OCT) images for the response of diabetic macular edema (DME) patients to anti-vascular endothelial growth factor (VEGF) therapy generated from baseline images using generative adversarial networks (GANs). METHODS: Patient information, including clinical and imaging data, was obtained from inpatients at the Ophthalmology Department of Qilu Hospital. 715 and 103 pairs of pre-and post-treatment OCT images of DME patients were included in the training and validation sets, respectively. The post-treatment OCT images were used to assess the validity of the generated images. Six different GAN models (CycleGAN, PairGAN, Pix2pixHD, RegGAN, SPADE, UNIT) were applied to predict the efficacy of anti-VEGF treatment by generating OCT images. Independent screening and evaluation experiments were conducted to validate the quality and comparability of images generated by different GAN models. RESULTS: OCT images generated f GAN models exhibited high comparability to the real images, especially for edema absorption. RegGAN exhibited the highest prediction accuracy over the CycleGAN, PairGAN, Pix2pixHD, SPADE, and UNIT models. Further analyses were conducted based on the RegGAN. Most post-therapeutic OCT images (95/103) were difficult to differentiate from the real OCT images by retinal specialists. A mean absolute error of 26.74 ± 21.28 µm was observed for central macular thickness (CMT) between the synthetic and real OCT images. CONCLUSION: Different generative adversarial networks have different prognostic efficacy for DME, and RegGAN yielded the best performance in our study. Different GAN models yielded good accuracy in predicting the OCT-based response to anti-VEGF treatment at one month. Overall, the application of GAN models can assist clinicians in prognosis prediction of patients with DME to design better treatment strategies and follow-up schedules.

Deep Learning Performance of Ultra-Widefield Fundus Imaging for Screening Retinal Lesions in Rural Locales.

Importance: Retinal diseases are the leading cause of irreversible blindness worldwide, and timely detection contributes to prevention of permanent vision loss, especially for patients in rural areas with limited medical resources. Deep learning systems (DLSs) based on fundus images with a 45° field of view have been extensively applied in population screening, while the feasibility of using ultra-widefield (UWF) fundus image-based DLSs to detect retinal lesions in patients in rural areas warrants exploration. Objective: To explore the performance of a DLS for multiple retinal lesion screening using UWF fundus images from patients in rural areas. Design, Setting, and Participants: In this diagnostic study, a previously developed DLS based on UWF fundus images was used to screen for 5 retinal lesions (retinal exudates or drusen, glaucomatous optic neuropathy, retinal hemorrhage, lattice degeneration or retinal breaks, and retinal detachment) in 24 villages of Yangxi County, China, between November 17, 2020, and March 30, 2021. Interventions: The captured images were analyzed by the DLS and ophthalmologists. Main Outcomes and Measures: The performance of the DLS in rural screening was compared with that of the internal validation in the previous model development stage. The image quality, lesion proportion, and complexity of lesion composition were compared between the model development stage and the rural screening stage. Results: A total of 6222 eyes in 3149 participants (1685 women [53.5%]; mean [SD] age, 70.9 [9.1] years) were screened. The DLS achieved a mean (SD) area under the receiver operating characteristic curve (AUC) of 0.918 (0.021) (95% CI, 0.892-0.944) for detecting 5 retinal lesions in the entire data set when applied for patients in rural areas, which was lower than that reported at the model development stage (AUC, 0.998 [0.002] [95% CI, 0.995-1.000]; P < .001). Compared with the fundus images in the model development stage, the fundus images in this rural screening study had an increased frequency of poor quality (13.8% [860 of 6222] vs 0%), increased variation in lesion proportions (0.1% [6 of 6222]-36.5% [2271 of 6222] vs 14.0% [2793 of 19 891]-21.3% [3433 of 16 138]), and an increased complexity of lesion composition. Conclusions and Relevance: This diagnostic study suggests that the DLS exhibited excellent performance using UWF fundus images as a screening tool for 5 retinal lesions in patients in a rural setting. However, poor image quality, diverse lesion proportions, and a complex set of lesions may have reduced the performance of the DLS; these factors in targeted screening scenarios should be taken into consideration in the model development stage to ensure good performance.

DeepQuality improves infant retinopathy screening.

Image quality variation is a prominent cause of performance degradation for intelligent disease diagnostic models in clinical applications. Image quality issues are particularly prominent in infantile fundus photography due to poor patient cooperation, which poses a high risk of misdiagnosis. Here, we developed a deep learning-based image quality assessment and enhancement system (DeepQuality) for infantile fundus images to improve infant retinopathy screening. DeepQuality can accurately detect various quality defects concerning integrity, illumination, and clarity with area under the curve (AUC) values ranging from 0.933 to 0.995. It can also comprehensively score the overall quality of each fundus photograph. By analyzing 2,015,758 infantile fundus photographs from real-world settings using DeepQuality, we found that 58.3% of them had varying degrees of quality defects, and large variations were observed among different regions and categories of hospitals. Additionally, DeepQuality provides quality enhancement based on the results of quality assessment. After quality enhancement, the performance of retinopathy of prematurity (ROP) diagnosis of clinicians was significantly improved. Moreover, the integration of DeepQuality and AI diagnostic models can effectively improve the model performance for detecting ROP. This study may be an important reference for the future development of other image-based intelligent disease screening systems.

DeepFundus: A flow-cytometry-like image quality classifier for boosting the whole life cycle of medical artificial intelligence.

Medical artificial intelligence (AI) has been moving from the research phase to clinical implementation. However, most AI-based models are mainly built using high-quality images preprocessed in the laboratory, which is not representative of real-world settings. This dataset bias proves a major driver of AI system dysfunction. Inspired by the design of flow cytometry, DeepFundus, a deep-learning-based fundus image classifier, is developed to provide automated and multidimensional image sorting to address this data quality gap. DeepFundus achieves areas under the receiver operating characteristic curves (AUCs) over 0.9 in image classification concerning overall quality, clinical quality factors, and structural quality analysis on both the internal test and national validation datasets. Additionally, DeepFundus can be integrated into both model development and clinical application of AI diagnostics to significantly enhance model performance for detecting multiple retinopathies. DeepFundus can be used to construct a data-driven paradigm for improving the entire life cycle of medical AI practice.

Prediction of Visual Acuity after anti-VEGF Therapy in Diabetic Macular Edema by Machine Learning.

Purpose: To predict visual acuity (VA) 1 month after anti-vascular endothelial growth factor (VEGF) therapy in patients with diabetic macular edema (DME) by using machine learning. Methods: This retrospective study included 281 eyes with DME receiving intravitreal anti-VEGF treatment from January 1, 2019, to April 1, 2021. Eighteen features from electronic medical records and measurements data from OCT images were extracted. The data obtained from January 1, 2019, to November 1, 2020, were used as the training set; the data obtained from November 1, 2020, to April 1, 2021, were used as the validation set. Six different machine learning algorithms were used to predict VA in patients after anti-VEGF therapy. After the initial detailed investigation, we designed an optimization model for convenient application. The VA predicted by machine learning was compared with the ground truth. Results: The ensemble algorithm (linear regression + random forest regressor) performed best in VA and VA variance predictions. In the validation set, the mean absolute errors (MAEs) of VA predictions were 0.137-0.153 logMAR (within 7-8 letters), and the mean square errors (MSEs) were 0.033-0.045 logMAR (within 2-3 letters) for the 1-month VA predictions, respectively. For the prediction of VA variance at 1 month, the MAEs were 0.164-0.169 logMAR (within 9 letters), and the MSEs were 0.056-0.059 logMAR (within 3 letters), respectively. Conclusions: Our machine learning models could accurately predict VA and VA variance in DME patients receiving anti-VEGF therapy 1 month after, which would be much valuable to guide precise individualized interventions and manage expectations in clinical practice.

The Fundus Structural and Functional Predictions of DME Patients After Anti-VEGF Treatments.

Diabetic retinopathy (DR) is an important complication with a high incidence of 34.6% in the diabetic populations. DR could finally lead to vision impairment without effective interventions, during which, diabetic macular edema (DME) is a key phase causing visual loss. Up to date, antivascular endothelial growth factor (anti-VEGF) therapy is the first-line treatment for DME which has achieved relatively better clinical outcomes than traditional treatments. However, there are several kinds of anti-VEGF medicines, and patients are sensitive to different anti-VEGF treatments. In addition, its effectiveness is unstable. Considering the patients' need to accept continual anti-VEGF treatments and its price is comparatively high, it is clinically important to predict the prognosis after different anti-VEGF treatments. In our research, we used the demographic and clinical data of 254 DME patients and 2,763 optical coherence tomography (OCT) images from three countries to predict the fundus structural and functional parameters and treatment plan in 6 months after different anti-VEGF treatments. Eight baseline features combined with 11 models were applied to conduct seven prediction tasks. Accuracy (ACC), the area under curve (AUC), mean absolute error (MAE), and mean square error (MSE) were respectively used to evaluate the classification and regression tasks. The ACC and AUC of structural predictions of retinal pigment epithelial detachment were close to 1.000. The MAE and MSE of visual acuity predictions were nearly 0.3 to 0.4 logMAR. The ACC of treatment plan regarding continuous injection was approaching 70%. Our research has achieved great performance in the predictions of fundus structural and functional parameters as well as treatment plan, which can help ophthalmologists improve the treatment compliance of DME patients.

Assessment of choroidal structural changes in patients with pre- and early-stage clinical diabetic retinopathy using wide-field SS-OCTA.

Purpose: To investigate the micro-vascular changes in choroidal structures in patients with pre- and early-stage clinical diabetic retinopathy (DR) using wide-field Swept-Source Optical Coherence Tomography Angiography (SS-OCTA). Method: This observational cross-sectional study included 131 eyes of 68 subjects that were divided into healthy controls (group 1, n = 46), pre-DR (group 2, n = 43), early-stage DR (group 3, n = 42) cohorts. All participants that underwent SS-OCTA examination were inpatients in the department of Ophthalmology and the department of Endocrinology, Qilu Hospital, Shandong University, and Department of Ophthalmology, Aier Eye Hospital, Jinan, from July 11, 2021 to March 17, 2022. The choroidal vascularity index (CVI), choroidal thickness (ChT) and central macular thickness (CMT) in the whole area (diameter of 12 mm) and concentric rings with different ranges (0-3, 3-6, 6-9, and 9-12 mm) were recorded and analyzed from the OCTA image. Result: Compared with healthy eyes, decreases in CVI and ChT were found in the eyes of patients with pre-or early-stage DR. The changes were more significant in the peripheral choroid, with the most prominent abnormalities in the 9-12mm area (P < 0.001). However, there was no obvious difference in the average CMT value. Furthermore, CVI and ChT were significantly correlated with the duration of diabetes in the range of 6-9 and 9-12 mm (Ps < 0.05; Correlation coefficient = -0.549, -0.395, respectively), with the strongest correlation (Ps < 0.01; Correlation coefficient = -0.597, -0.413, respectively) observed at 9-12 mm. Conclusion: The CVI and ChT values of diabetic patients are significantly lower than in healthy controls, especially in patients with early-stage DR. In addition, the peripheral choroidal capillaries are more susceptible to early DM-induced injury than in the central area.

Predicting OCT images of short-term response to anti-VEGF treatment for retinal vein occlusion using generative adversarial network.

To generate and evaluate post-therapeutic optical coherence tomography (OCT) images based on pre-therapeutic images with generative adversarial network (GAN) to predict the short-term response of patients with retinal vein occlusion (RVO) to anti-vascular endothelial growth factor (anti-VEGF) therapy. Real-world imaging data were retrospectively collected from 1 May 2017, to 1 June 2021. A total of 515 pairs of pre-and post-therapeutic OCT images of patients with RVO were included in the training set, while 68 pre-and post-therapeutic OCT images were included in the validation set. A pix2pixHD method was adopted to predict post-therapeutic OCT images in RVO patients after anti-VEGF therapy. The quality and similarity of synthetic OCT images were evaluated by screening and evaluation experiments. We quantitatively and qualitatively assessed the prognostic accuracy of the synthetic post-therapeutic OCT images. The post-therapeutic OCT images generated by the pix2pixHD algorithm were comparable to the actual images in edema resorption response. Retinal specialists found most synthetic images (62/68) difficult to differentiate from the real ones. The mean absolute error (MAE) of the central macular thickness (CMT) between the synthetic and real OCT images was 26.33 ± 15.81 µm. There was no statistical difference in CMT between the synthetic and the real images. In this retrospective study, the application of the pix2pixHD algorithm objectively predicted the short-term response of each patient to anti-VEGF therapy based on OCT images with high accuracy, suggestive of its clinical value, especially for screening patients with relatively poor prognosis and potentially guiding clinical treatment. Importantly, our artificial intelligence-based prediction approach's non-invasiveness, repeatability, and cost-effectiveness can improve compliance and follow-up management of this patient population.