The aim of this study was to investigate the prevalence and clinical characteristics of macular microcystic edema (MME) in eyes with silicone oil (SO) endotamponade, with particular emphasis on OCT-based phenotypes and their associated structural and functional features. This retrospective cross-sectional study included eyes that underwent pars plana vitrectomy with SO for retinal detachment and had postoperative OCT imaging under SO endotamponade. Based on OCT findings, ME was classified into mixed phenotypes with both cystoid- and microcystic changes (CME/MME), and phenotypes with isolated MME. OCT biomarkers, layer-specific thickness measurements, and clinical variables were compared between phenotypes. Among 101 eyes with SO tamponade, ME was identified in 38 eyes (37.6
To develop and evaluate a deep-learning model for automated segmentation of myopic atrophic lesions on ultra-widefield (UWF) fundus images and to evaluate the effects of training-set size and color representation on segmentation performance. This study included 500 UWF images from highly myopic eyes with manually delineated myopic atrophic lesions. Independent training subsets of 50, 100, 200, 300, 400, and 500 images were constructed using a patient-level sampling strategy. For each subset size and color-space configuration (RGB, CIELAB, and luminance-only), models were trained 10 times with different random seeds. Segmentation performance was evaluated using the Dice coefficient and intersection-over-union (IoU). The effects of training-set size and color space were assessed using linear mixed-effects models, and performance variability was evaluated using Levene’s test. Training-set size significantly influenced segmentation performance (p < 0.001). RGB and CIELAB inputs performed comparably, whereas luminance-only input showed modestly lower Dice and IoU than RGB input. Model performance improved with increasing dataset size, with progressively smaller improvements beyond approximately 300–400 images. The U-Net++ model in the primary experiment trained on 500 images achieved a mean Dice coefficient of 0.860 ± 0.006 and a mean IoU of 0.754 ± 0.009, demonstrating stable and reproducible convergence across runs. Deep-learning-based segmentation of myopic atrophic lesions on UWF images is feasible. Performance gains became progressively smaller beyond approximately 300–400 annotated images under the present experimental conditions. RGB and CIELAB inputs showed comparable performance, whereas removal of chromatic information in the luminance-only condition was associated with modestly lower performance. These findings may inform efficient dataset design for quantitative UWF imaging analysis in highly myopic eyes and provide a methodological foundation for longitudinal investigation of lesion enlargement, risk-factor analysis, and spatial progression of pathologic myopia.
To evaluate the association between serum vitamin D levels and retinopathy of prematurity (ROP) through a systematic review and meta-analysis. A systematic search was conducted in PubMed, Embase, The Cochrane Library, Web of Science, and Scopus up to August 2026. All observational studies measuring vitamin D levels in infants with ROP were included. The standardized mean difference (SMD) was calculated using a random-effects model to assess differences in vitamin D levels between patients with ROP and healthy controls. The log odds ratio (OR) was computed to assess the association between vitamin D deficiency and the incidence of ROP. Ten relevant studies evaluating the relationship between vitamin D and ROP were included. Six studies, including 281 ROP infants, compared serum vitamin D levels between ROP and control groups. Five studies compared the risk of ROP development in the vitamin D-sufficient vs. deficient infants. Meta-analysis results revealed a significantly lower vitamin D level in ROP infants vs. control (SMD: -0.94, 95
Proliferative diabetic retinopathy (PDR) is an advanced neurovascular complication of diabetes mellitus that can result in irreversible visual loss. This study evaluated serum and vitreous concentrations of vascular endothelial growth factor A (VEGFA), ionized calcium-binding adapter molecule 1 (IBA-1), and retinol-binding protein 3 (RBP3), their ratios, and their associations with PDR and age. This analytical observational cross-sectional study included 30 patients with high-risk PDR undergoing pars plana vitrectomy and 10 nondiabetic controls undergoing vitrectomy for non-diabetic vitreoretinal conditions. Vitreous and serum samples were analyzed using enzyme-linked immunosorbent assays. Biomarker concentrations and ratios were compared between groups, and associations were assessed using nonparametric tests, logistic regression, and ridge-regularized logistic regression. Age-stratified analyses used three predefined groups: <45, 45–60, and > 60 years. Patients with PDR had significantly higher vitreous and serum IBA-1 concentrations, higher vitreous VEGFA concentrations, and lower vitreous RBP3 concentrations than controls (all p ≤ 0.014). Vitreous RBP3/IBA-1 and RBP3/VEGFA showed the strongest inverse associations with PDR. Vitreous IBA-1 demonstrated the strongest discrimination between the PDR and control groups (AUC = 1.000), followed by the vitreous RBP3/IBA-1 ratio (AUC = 0.993). Age-stratified analysis showed significant differences in vitreous IBA-1 concentrations across the < 45, 45–60, and > 60-year groups, with the highest concentrations observed in the > 60-year group. No significant age-group differences were observed for vitreous VEGFA or RBP3. Serum and vitreous IBA-1, VEGFA, and RBP3 profiles were associated with PDR, supporting the involvement of inflammatory, angiogenic, and neuroprotective pathways. The RBP3/IBA-1 and RBP3/VEGFA ratios may provide complementary information regarding these interacting pathways. Vitreous IBA-1 showed the strongest discrimination between the PDR and nondiabetic control groups within this cohort, while age-stratified analysis demonstrated significant differences in vitreous IBA-1 across age groups. These exploratory findings should be interpreted cautiously given the small sample size, the very high within-cohort discriminatory performance, and the absence of external validation. Validation in larger, age-matched, independent, and longitudinal cohorts is required to establish the reproducibility and clinical relevance of these biomarkers before considering any diagnostic or prognostic application.
To evaluate the choroid in multifocal retinitis assoiciated with Bartonella henselae infection. Other posterior uveitis diseases, such as toxoplasmosis, tend to present with a thickened choroid, and we aimed to evaluate whether this is the case in these patients as well. Consecutive cases of Bartonella henselae multifocal retinitis at Meir Medical Center in Kfar Saba, Israel, between January 2023 and December 2025 were reviewed retrospectively. The choroid in cases of retinitis was evaluated using OCT with enhanced depth imaging (EDI). Five patients (ten eyes) were included. All cases had a typical bilateral multifocal retinitis with small or mid-sized superficial retinal infiltrates. All were immunocompetent with a negative uveitis workup, including syphilis and rickettsia, but a positive Bartonella serology. All cases had a thickened choroid, which was either diffuse or focal, localized posterior to retinal infiltrates. The more involved eye of the same patient, with more inflammatory signs, also had a thicker choroid. The choroid became thinner after treatment. The choroid of patients with Bartonella henselae retinitis may be thickened as a result of the infection.
Anti-VEGF therapy remains the mainstay treatment for neovascular age-related macular degeneration (nAMD), yet a subset of eyes exhibit resistance to standard treatment regimens. This study evaluates the safety and efficacy of double-dose intravitreal faricimab (12 mg) in a series of eyes with nAMD refractory to bevacizumab, aflibercept, and standard-dose faricimab. This retrospective case series included 26 eyes from 24 patients with treatment-resistant nAMD who received intravitreal faricimab 12 mg every 4 weeks for three loading injections, followed by a treat-and-extend regimen when anatomic stability was achieved. Primary anatomic outcomes included central subfield thickness (CST), central foveal thickness (CFT), maximum macular thickness (MMT), pigment epithelial detachment (PED) height and volume, and intraretinal (IRF), subretinal (SRF), and total retinal fluid (TRF) volumes. Best-corrected visual acuity (BCVA) served as the functional endpoint. Safety measures included intraocular pressure (IOP), intraocular inflammation, systemic adverse events, and new-onset geographic atrophy (GA). Over a mean follow-up of 10.6 months, significant anatomical improvements were observed, including reductions in CFT (p=0.0029), MMT (p=0.0007), CST (p=0.0067), PED maximum height (p=0.0241), and TRF (p=0.0341). PED volume significantly decreased at 4 weeks (p=0.0394) but was not sustained. IRF and SRF reductions were not individually significant. BCVA, which had declined significantly in the 24 weeks prior to switching (p≈0.04), remained stable throughout treatment with double-dose faricimab (p>0.6). At final visit, 57.7
Full thickness macular holes (FTMH) exacerbated by neovascular age related macular degeneration (AMD) are a relatively rare ophthalmic entity. Due to the latter significantly lowering the threshold potential for visual recuperation, ophthalmologists are reluctant to treat these patients surgically. However time is of the essence and considering other factors, this decision should be made by an experienced vitreoretinal surgeon, who can elect the patient for vitrectomy if other ophthalmic factors are in favour and allow it. We present a case series of 4 patients diagnosed with a dual pathology of AMD and FTMH. Macular hole closure was achieved on OCT imaging and VA improed to the level which was determined by the macular state due to wet AMD. Full-thickness macular holes should be promptly treated surgically to achieve disease control, regardless of their etiology. A conservative approach may lead to further and irreversible deterioration of central visual acuity, despite reports on spontaneous closure and variable functional improvement. During vitrectomy, no additional choroidal membrane (CNV) excision was proven to improve the outcomes, which can be even worsened, therefore this modification should be avoided.
This study aims to investigate the effects of varying lengths of cilioretinal artery perfusion on central visual acuity and fundus angiography circulation time in patients with central retinal artery occlusion (CRAO). A retrospective analysis included 582 CRAO patients, including 87 patients with cilioretinal arteries. Cilioretinal arteries were classified by direction into nasal and temporal groups, with the temporal group further subdivided into four subtypes based on the optic disc-macula distance (DMD): 1/4, 2/4, 3/4, and 4/4 DMD. Differences in basic information, visual acuity, and circulation time among the groups were analyzed. Patients in the 4/4 DMD group were the oldest among the temporal subgroups (P < 0.05). The 4/4 DMD group demonstrated the best BCVA, while the Nasal-1/4DMD group had the worst (P < 0.05); notably, 13.33
Diabetic Retinopathy (DR) is a leading cause of vision impairment and a major long-term microvascular complication of diabetes. Early detection and prevention of diabetes-related complications through advanced imaging and software-assisted patient management remain important clinical priorities. Microaneurysms (MAs) are the earliest and most subtle indicators of DR, but their small size and low contrast often lead to missed detection during manual fundus examination, delaying intervention. Automated MA segmentation is therefore essential for large-scale DR screening. We conducted a controlled empirical evaluation of localized patch-based training for microaneurysm (MA) segmentation, benchmarking a compact, imbalance-aware U-Net model including learnable transposed-convolution up-sampling against a full-image U-Net trained on identical data. Fundus images and corresponding MA masks were divided into non-overlapping 256 × 256 patches, increasing MA pixel density per training sample by approximately 60-fold relative to full-image input thereby directly targeting the extreme class imbalance that causes full-image models to collapse to all-background predictions. The model was trained and evaluated on the publicly available IDRiD [Indian Diabetic Retinopathy Image Dataset] and DDR [Dataset for Diabetic Retinopathy] datasets. Performance was assessed using Intersection over Union (IoU), Dice coefficient, accuracy, recall, and precision. Performance was evaluated on a representative held-out subset of 100 images selected from an independent pool of 514 MA-annotated test images spanning the IDRiD and DDR datasets. The patch-based model achieved an overall pixel-level accuracy of 99.89
To evaluate the efficiency and safety of ChatGPT and Gemini in the surgical workflow, from patient admission to surgical planning. Twenty questions regarding preoperative and postoperative aspects of vitreoretinal surgery—most frequently asked by patients at our center—were submitted to ChatGPT (OpenAI) and Gemini (Google). Similarly, twenty surgical cases treated at our center were submitted to both platforms. The same questions and cases were evaluated by a panel of vitreoretinal surgeons with varying experience. Primary outcomes were overall reliability and inter-group agreement of the two chatbots; the secondary outcome was their ability to identify cases requiring ophthalmologist intervention. In the clinical case dataset (N = 80), ChatGPT and Gemini performed comparably, with ChatGPT slightly outperforming Gemini in discussion timeliness (p = 0.019). In the patient inquiry dataset (N = 80), Gemini provided more complete, accurate, and timely clinical indications (p ≤ 0.011). Stratified analyses by surgeon experience (Junior vs. Senior) showed consistent evaluations across experience levels. ChatGPT and Gemini perform similarly in clinical case discussions, with ChatGPT slightly better in timeliness. For patient-oriented inquiries, Gemini delivers more accurate, complete, and up-to-date responses. While AI platforms demonstrate potential, their adoption in clinical practice should be preceded by thorough and rigorous evaluation.
We compared air and gas tamponade for idiopathic full-thickness macular hole (FTMH) surgery with a 24-hour face-down protocol, in a bicentric, retrospective real-world series of 36 air- and 19 gas-treated eyes. Allocation was non-random (gas predominated earlier, air later). Preoperative symptom duration was short in both groups (median 1 [1–2] months); median hole diameter was 456 [312–611] µm (air) and 497 [364–776] µm (gas; p = 0.40). FTMH closure at 1 month was 100
Endoillumination is essential in modern vitreoretinal surgery, enabling visualization during peripheral vitreous shaving, retinal detachment repair, and other bimanual maneuvers. Chandelier illumination provides a stable wide-field light source through a separate port and should be regarded as an adjunct to conventional handheld fiber-optic endoillumination rather than as a replacement. This narrative review compares handheld, chandelier, and emerging LED-based illumination systems, focusing on surgical utility, visualization, illumination geometry, and retinal light-exposure considerations. A PubMed literature search was conducted for studies published between 1990 and April 2026. After screening 401 records and evaluating 72 full-text articles, 24 studies were included in the qualitative synthesis. These comprised in vitro investigations on retinal pigment epithelial cells, ex vivo porcine-eye irradiance measurements, prospective clinical series, and one randomized controlled trial. Across the available evidence, the most consistent safety advantage of chandelier illumination was related to delivery geometry. By increasing the working distance from the retina, chandelier systems markedly prolong retinal threshold exposure time compared with handheld endoillumination probes, largely in accordance with the inverse-square law. LED-based systems demonstrated variable safety profiles depending on spectral output, with some devices showing prominent short-wavelength peaks and others exhibiting lower blue-light fractions when compared to xenon sources. Spectral filtering, particularly amber filtering during dye-assisted procedures, further reduced photochemical hazard. Three-dimensional digital visualization systems also reduce the level of physical intraocular illumination required from conventional lighting systems while maintaining adequate surgical visualization. Additional emerging technologies, including micro-LED chip-on-tip probes, transscleral illumination, omnidirectional diffusers, and portable LED devices, suggest a broader transition toward fiber-free or fiber-optimized illumination platforms. Chandelier-based LED illumination systems appear to offer a versatile and generally safer addition to traditional wired endoilluminators. Their clinical value lies in improving peripheral visualization and facilitating complex bimanual maneuvers. Safety advantages are mainly related to illumination geometry, working distance, light distribution, and reduced need for repeated handheld probe manipulation. Nevertheless, retinal safety depends on the combined effects of emission spectrum, delivery geometry, intraocular distance, exposure duration, spectral filtering, surgical technique, and visualization system.
To present the slit-lamp illumination and visualisation system for scleral buckling for primary rhegmatogenous retinal detachment (RRD) and assess its feasibility and postoperative outcomes. In a retrospective, single-center, non-comparative case series, patients with subtotal RRD who underwent scleral buckling under a slit-lamp illumination system at Lariboisière Hospital, between May 2017 and May 2022, were included. Patients who underwent scleral buckling combined with vitrectomy, with a history of retinal surgery and with maculopathies interfering with visual acuity were excluded. The following preoperative data were collected: best-corrected visual acuity (BCVA), number and location of retinal breaks, extend of the detachment, presence of macular detachment. The fundus was visualised using a quadraspheric lens, and the slit-lamp system integrated in the surgical microscope. Anatomical success rate, BCVA improvement and complications were collected. Forty-six eyes of 46 patients with a mean age of 43.2 ± 12.8 years, followed for 8.7 ± 8.3 months were included. No procedure was switched to vitrectomy during surgery. The single surgery success rate was 93.5
To evaluate structural and functional outcomes in patients with acute Vogt-Koyanagi-Harada (VKH) disease, based on the timing of systemic corticosteroid (CS) initiation and the early introduction of noncorticosteroid immunomodulatory therapy (IMT). This retrospective single-centre study included 35 consecutive patients (70 eyes) diagnosed with acute Vogt-Koyanagi-Harada disease at a tertiary center in São Paulo, Brazil. Patients received methylprednisolone pulse therapy (1,000 mg/day, 3 days) followed by gradually tapered oral prednisone (1 mg/kg/day). Patients were classified by the timing of CS initiation. The early treatment group (ETG, n = 24) initiated CS within 30 days of onset and comprised two subgroups: CS monotherapy (Group 1, n = 9) and CS plus IMT within 3 months (Group 2, n = 15). Those treated after 30 days because of late presentation formed the late treatment group (LTG, Group 3, n = 11). Follow-up was 24 months with multimodal imaging and full-field electroretinography. At 24 months, 98.6
Circumscribed choroidal hemangioma (CCH) is a benign vascular choroidal tumor traditionally characterized on ultrasonography by dome-shaped morphology, acoustic solidity, and high internal reflectivity. Dynamic features on real-time B-scan ultrasonography have not been systematically evaluated. This study aimed to characterize the prevalence of intralesional flickering on real-time dynamic B-scan ultrasonography in treatment-naive CCH and to evaluate its relationship to tumor dimensions and exudative activity. This retrospective observational study included 29 treatment-naive eyes of 29 patients diagnosed with CCH at a tertiary ocular oncology service between 2015 and 2025. Standardized B-scan ultrasonography at 10 MHz, with real-time documentation of intralesional flickering, was performed per departmental protocol. Tumor largest basal diameter (LBD) and thickness were measured. Spectral-domain optical coherence tomography was reviewed for subretinal and intraretinal fluid. Associations between flickering status and tumor dimensions, exudative fluid, and tumor location were assessed. Mean age was 55.1 years, 75.9
To systematically review the diagnostic accuracy and inter-test agreement of optical coherence tomography (OCT) for detecting posterior vitreous detachment (PVD) in blinded, within-subject studies that use different reference standards. This systematic review followed PRISMA guidelines and was registered in PROSPERO (CRD42024580929). PubMed, Scopus, Embase, ScienceDirect, Web of Science, Google Scholar, and the Cochrane Database were searched from inception through March 2025. Eligible studies were English-language diagnostic accuracy studies evaluating PVD using blinded, within-subject designs, in which any OCT modality served as the index test and was compared with either a non-OCT reference standard or an alternative OCT protocol that differed in modality, scan protocol, diagnostic criteria, or grader type. Index and reference tests were performed on the same eyes within 24 h. Five studies (812 patients, 922 eyes) met the inclusion criteria; three evaluated diagnostic performance and two assessed inter-grader agreement. All studies used spectral-domain OCT (SD-OCT). Diagnostic performance varied substantially by imaging protocol and reference standard. High-density imaging protocols that included both the macula and optic nerve head achieved a sensitivity of 96.4
Abstract Purpose To evaluate the 2-year real-world outcomes of switching to intravitreal faricimab in patients with neovascular age-related macular degeneration (nAMD) previously treated with anti-VEGF agents and unable to extend treatment intervals (TI) beyond 12 weeks. Methods Prospective cohort study including patients with nAMD managed under a treat-and-extend regimen who required frequent anti-VEGF injections (< 12-week intervals) to maintain retinal dryness. Patients were switched to faricimab without a loading phase and followed for a minimum of 2 years. TI were extended in 4-week increments based on anatomical response. The primary outcome was change in TIl; secondary outcomes included change in best-corrected visual acuity (BCVA) and identification of predictors of response. “Optimal responders” were defined as eyes achieving ≥ 8-week interval extension. Results A total of 110 eyes from 97 patients were included (mean age 78.7 ± 9.8 years). Prior to switching, the mean TI was 6.18 ± 2.05 weeks, which increased significantly to 13.61 ± 5.62 weeks at the last follow-up visit ( p < 0.001). BCVA remained stable over 2 years (0.42 ± 0.24 vs. 0.43 ± 0.25; p = 0.0117). At final follow-up, 68.2% of eyes achieved intervals ≥ 12 weeks, 20.0% ≥16 weeks, and 24.6% reached ≥ 20-week intervals. Overall, 89.1% of eyes experienced interval extension. Optimal responders accounted for 47.3% of eyes and were significantly younger and had received fewer prior injections ( p < 0.05). Conclusions Switching to faricimab in previous treated nAMD patients enabled meaningful extension of TI over 2 years while maintaining stable visual acuity. These findings support faricimab as an effective strategy to reduce treatment burden in patients who are difficult to extend with prior anti-VEGF therapies.
To evaluate the efficacy of combined oral curcumin and intravitreal bevacizumab (IVB) in diabetic macular edema (DME). This clinical trial was conducted using a parallel-group, randomized, placebo-controlled design in patients with DME. All participants received IVB at the standard dose at monthly intervals for a total of three doses. Additionally, they were assigned to receive either oral curcumin capsules (40 mg) or a matched placebo every 12 h for three months, starting at the time of the first IVB injection. Treatment efficacy was assessed by comparing changes in foveal, perifoveal, and parafoveal retinal thickness, central macular volume (CMV), and LogMAR-corrected distance visual acuity (CDVA). Hemoglobin A1c (HbA1c) tests were performed for all patients at baseline and at three months. Fifty-two eyes of 52 patients (26 eyes per group) completed the study. The patients’ baseline characteristics were comparable between the groups. Macular thickness and CMV decreased significantly from baseline in both groups, with no statistically significant difference observed between the groups. Similarly, changes in LogMAR CDVA did not differ significantly between the intervention and control groups (MD: 0.08, [95
Diabetic tractional retinal detachment (TRD) secondary to proliferative diabetic retinopathy is a vision-threatening complication that may require pars plana vitrectomy in selected cases, particularly when the macula is threatened or detached. Silicone oil tamponade is commonly used in complex cases, although tamponade choice varies according to surgical findings and surgeon preference. To evaluate surgical outcomes associated with adjunctive dexamethasone implant during pars plana vitrectomy with silicone oil or heavy silicone oil tamponade for diabetic tractional retinal detachment. A comprehensive literature search of PubMed, Scopus, Web of Science, and ClinicalTrials.gov was performed from inception through 2025. Randomized and non-randomized comparative studies assessing PPV with silicone oil with or without adjunctive dexamethasone implant were considered. Definitions of postoperative PVR and retinal redetachment were accepted as reported by the original studies; no individual-level imaging or independent photographic regrading was available. The anatomical outcome was an exploratory composite of study-reported postoperative proliferative vitreoretinopathy (PVR) and/or retinal redetachment outcomes. Because definitions and follow-up time points differed across studies, this outcome was not treated as a uniformly graded PVR endpoint. Randomized and non-randomized comparative studies were identified; randomized trials were prioritized for the main analysis, while the non-randomized study was considered separately because of serious risk of bias. Secondary outcomes included best-corrected visual acuity (BCVA), central subfield thickness (CST), ocular hypertension, and postoperative complications. The protocol was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO: CRD420261283105) prior to the commencement of the review. Four studies involving 305 eyes were included. In the main randomized-trial analysis, two studies comprising 123 eyes contributed to the exploratory study-reported postoperative proliferative/redetachment composite. The pooled estimate was highly imprecise (RR 0.20, 95
To evaluate the short-term functional and anatomical outcomes of intravitreal aflibercept 8 mg in treatment-naïve neovascular age-related macular degeneration (nAMD), and to explore whether the early treatment response varied according to pachychoroid and polypoidal lesion status. This retrospective, multicenter observational study evaluated 45 treatment-naïve eyes of 45 patients with nAMD who received three monthly loading injections of aflibercept 8 mg. Best-corrected visual acuity (BCVA), central retinal thickness (CRT), central choroidal thickness (CCT), and exudative findings (subretinal fluid [SRF], intraretinal fluid [IRF], and subretinal hyperreflective material [SHRM]) were assessed at baseline and at 1, 2, and 4 months. Pachychoroid and polypoidal lesion status was determined using multimodal imaging. Longitudinal changes were analyzed using linear mixed-effects models, with subgroup analyses by pachychoroid and polypoidal lesion status. At 4 months, mean BCVA improved from 0.27 ± 0.33 to 0.20 ± 0.31 logMAR (P = 0.005), CRT decreased from 382.1 ± 189.7 to 231.3 ± 109.1 μm, and CCT from 263.0 ± 146.5 to 231.0 ± 127.2 μm (both P < 0.001). The dry macula rate (absence of both SRF and IRF) increased from 17.8
