Purpose: To evaluate the performance of deep learning (DL) in diagnosing glaucoma and predicting its progression using fundus photography and retinal optical coherence tomography (OCT) images. Materials and Methods: Relevant studies published up to 30 October 2024 were retrieved from PubMed, Medline, EMBASE, Cochrane Library, Web of Science, and ClinicalKey. A bivariate random-effects model was employed to calculate pooled sensitivity, specificity, positive and negative likelihood ratios, and area under the receiver operating characteristic curve (AUROC). Results: A total of 48 studies were included in the meta-analysis. DL algorithms demonstrated high diagnostic performance in glaucoma detection using fundus photography and OCT images. For fundus photography, the pooled sensitivity and specificity were 0.92 (95% CI: 0.89–0.94) and 0.93 (95% CI: 0.90–0.95), respectively, with an AUROC of 0.90 (95% CI: 0.88–0.92). For the OCT imaging, the pooled sensitivity and specificity were 0.90 (95% CI: 0.84–0.94) and 0.87 (95% CI: 0.81–0.91), respectively, with an AUROC of 0.86 (95% CI: 0.83–0.90). In predicting glaucoma progression, DL models generally showed less robust performance, with pooled sensitivities and specificities ranging lower than in diagnostic tasks. Internal validation datasets showed higher accuracy than external validation datasets. Conclusions: DL algorithms achieve excellent performance in diagnosing glaucoma using fundus photography and OCT imaging. To enhance the prediction of glaucoma progression, future DL models should integrate multimodal data, including functional assessments, such as visual field measurements, and undergo extensive validation in real-world clinical settings.

To investigate vascular abnormalities, including persistent avascular retina (PAR), in a large cohort of patients with regressed or treated retinopathy of prematurity (ROP) through long-term follow-up until school age. Retrospective large cohort study. We included pediatric patients (< 18 years old) with a history of nontreated or treated ROP (treated with either photocoagulation or intravitreal injection [IVI]) and regularly followed up until 2020. Upon enrollment, we categorized patients into 4 groups: prematurity, regressed ROP, and IVI and laser treatment ROP groups. All patients had undergone visual acuity examinations, OCT, and ultrawide-field fluorescein angiography. Percentage of eyes with PAR (greater than or equal to the area of 2 disc diameters from the ora serrata to vascular termini) and with vascular abnormalities in the peripheral and posterior retina. In total, we included 187 eyes of 95 patients. The prevalence of PAR was 0%, 33.33%, and 31.65% in eyes in the prematurity, regressed ROP, and IVI treatment groups, respectively (P < 0.0001). There was no significant difference in the percentage of PAR eyes when comparing regressed ROP group (33.33%) with the IVI treatment group (31.65%). At least 1 type of vascular abnormality was noted until school age in all (100%) treated ROP eyes. Although multivariate analysis found a significant association between IVI treatment and PAR (odds ratio: 10.28, 95% confidence interval: 3.29–32.14) until the school age of 6 to 8 years old, there were no stage 3 eyes in the spontaneously regressed group, suggesting that stage 3 ROP in the IVI group could also drive the association. Approximately one-third of ROP eyes with spontaneous regression or IVI treatment still exhibit PAR when the child reaches school age. Several distinct vascular anomalies at the vascular–avascular juncture and within the vascularized retina can persist in these children. The clinical significance of these anomalies and the decision whether to treat them warrant further investigation to optimize their outcomes. The authors have no proprietary or commercial interest in any materials discussed in this article.

To investigate the risks of metabolic acidosis and renal outcomes after topical carbonic anhydrase inhibitor (CAI) use in patients with both primary open-angle glaucoma (POAG) and advanced chronic kidney disease (CKD). Nationwide, population-based cohort study. This study was conducted with population data from Taiwan's National Health Insurance (NHI) Research Database between January 2000 and June 2009. Patients with advanced CKD who were diagnosed with glaucoma (International Classification of Diseases, Ninth Revision [ICD-9] code 365) and had been receiving eye drops for glaucoma (including carbonic anhydrase inhibitors selected by NHI drug code) were enrolled. Using Kaplan-Meier methods, we compared the cumulative incidence of mortality, long-term dialysis, and cumulative incidence of metabolic acidosis over time between CAI users and CAI non-users. Primary outcomes comprised mortality, renal outcome (progression to hemodialysis), and metabolic acidosis. In this cohort, topical CAI users had a higher incidence of long-term dialysis than non-users (incidence = 1,216.85 vs 764.17 events per 100 patient-years; adjusted hazard ratio = 1.17, 95% CI = 1.01-1.37). Hospital admissions due to metabolic acidosis were higher in CAI users compared with non-users (incidence = 21.54 vs 11.87 events per 100 patient-years; adjusted hazard ratio = 1.89, 95% CI = 1.07-3.36). Topical CAIs may be associated with higher risks of long-term dialysis and metabolic acidosis in patients with POAG and pre-dialysis advanced CKD. Therefore, topical CAIs should be used with caution in advanced CKD patients.

Purpose: To compare between oral and intravenous (IV) ultrawide-field fluorescein angiography in pediatric patients with a history of prematurity of retinopathy or prematurity. Methods: Pediatric patients (<18 year old; n = 107 patients) who underwent ultrawide-field fluorescein angiography for retinopathy of prematurity were categorized into oral and IV fluorescein angiography (FA) groups. Quality of FA images was graded on the order of retinal vessels visible. Reported outcomes were proportions of graded FA images, peak fluorescein intensity, and the time to first dye appearance and to reach peak fluorescence. Results: Image quality analysis revealed that 91.5% of IV FA images had excellent image quality compared with only 55.6% of oral FA images (P < 0.01). There were still 83.3% of oral-contrast images with good or excellent image quality. The average time required for first dye appearance and peak fluorescence were significantly shorter in the IV FA group than in the oral FA group (P < 0.01). Peak intensity was greater in the IV group (141.41 ± 29.09) than in the oral group (111.25 ± 45.68; P < 0.01). Adverse reaction rates were similar between the two groups (P = 0.22). Conclusion: Ultrawide-field fluorescein angiography provides excellent-quality imaging of the retina in the pediatric population. Overall, oral FA is still an effective and useful alternative to IV FA in children with prematurity history.

Purpose—Visual complaints are common in trauma cases. However, not every institution provides immediate ophthalmic consultations 24 h per day. Some patients may receive an ophthalmic consultation but without positive findings. We tried to evaluate risk factors for ocular emergencies in trauma patients. Then, the ophthalmologists could be selectively consulted. Methods—From January 2019 to December 2019, head injuries patients concurrent with suspected ocular injuries were retrospectively reviewed. All of the patients received comprehensive ophthalmic examinations by ophthalmologists. Patients with and without ocular injuries were compared. Specific ophthalmic evaluations that could be primarily performed by primary trauma surgeons were also analyzed in detail. Results—One hundred forty cases were studied. Eighty-nine (63.6%) patients had ocular lesions on computed tomography (CT) scans or needed ophthalmic medical/surgical intervention. Near 70% (69.7%, 62/89) of patients with ocular injuries were diagnosed by CT scans. There was a significantly higher proportion of penetrating injuries in patients with ocular injuries than in patients without ocular injuries (22.5% vs. 3.9%, p = 0.004). Among the patients with blunt injuries (N = 118), 69 (58.5%) patients had ocular injuries. These patients had significantly higher proportions of periorbital swelling (89.9% vs. 67.3%, p = 0.002) and diplopia (26.1% vs. 8.2%, p = 0.014) than patients without ocular injuries. Conclusions—In patients with head injuries, concomitant ocular injuries with indications for referral should always be considered. CT serves as a rapid and essential diagnostic tool for the evaluation of concomitant ocular injuries. Ophthalmologists could be selectively consulted for patients with penetrating injuries or specific ocular presentations, thus reducing the burden of ophthalmologists.

Tissue inhibitors of metalloproteinases (TIMPs) play a crucial role in endogenous angiogenesis besides the regulation of matrix metalloproteinase (MMP) activity. Associations between TIMP-2 gene polymorphisms and the risk of retinopathy of prematurity (ROP) were examined. Premature infants born between 2009 and 2018 were included. Five single-nucleotide polymorphisms (SNPs) of TIMP-2 were analyzed with real-time polymerase chain reaction (PCR). Multivariate logistic regression was applied to model associations between TIMP-2 polymorphisms and ROP susceptibility and severity. The GA+AA genotype in individuals with the TIMP-2 polymorphism of rs12600817 was associated with a higher risk of ROP (odds ratio [OR]: 1.518, 95% confidence interval [CI]: 1.028–2.242) compared with their wild-type genotypes. The AA genotype (OR: 1.962, 95% CI: 1.023–3.762) and the AA+GA genotype (OR: 1.686, 95% CI: 1.030–2.762) in individuals with the rs12600817 polymorphism had higher risks of severe, treatment-requiring ROP relative to their wild-type counterparts. In patients with treatment-requiring ROP, the AG+GG genotypes in the TIMP-2 polymorphism of rs2889529 were correlated with the treatment response (p = 0.035). The TIMP-2 polymorphism of rs12600817 help in predicting ROP risks in preterm infants, while the polymorphism of rs2889529 can serve as a genetic marker in evaluating the ROP treatment response.