Translational Vision Science & Technology最新文献

Purpose: To evaluate the safety and efficacy of a novel endocapsular device used during cataract surgery at the 12-month follow-up.

Methods: A cohort study was conducted in a university-affiliated private practice. Adults with age-related cataracts, intraocular lens power between 14 and 26 diopters, and no other ocular pathology were included prospectively. Exclusion criteria included diabetes, previous ocular surgery, cardiac conditions, and autoimmune diseases. One eye per patient was randomly assigned to receive the device before intraocular lens implantation. Follow-ups were conducted preoperatively and at 1 day, 1 week, and 1, 3, 6, and 12 months postoperatively. The primary safety endpoint was the incidence of adverse events; the primary efficacy endpoint was the incidence of posterior capsule opacification (PCO) postoperatively at 12 months. A control group was formed retrospectively for comparison of PCO at 12 months. Spherical equivalent, corrected distance visual acuity, and intraocular pressure were also measured.

Results: A total of 121 patients were enrolled. Sixteen adverse events occurred in 12 patients; all resolved and were deemed unrelated to the device. PCO incidence at 12 months was 0.83% in the experimental group vs. 13.0% in the control group. The spherical equivalent stabilized by 3 months. At 12 months, the mean corrected distance visual acuity and intraocular pressure were 0.03 ± 0.07 logarithm of the minimum angle of resolution and 11.23 ± 2.03 mm Hg, respectively.

Conclusions: The new device seems to be safe and have a beneficial impact on PCO up to at least 12 months postoperatively.

Translational relevance: We report our experience with a novel, safe, endocapsular open capsule device that shows promise in preventing posterior capsule opacification in real-world clinical settings.

Purpose: Previously, a variant within the Pcsk1 gene was found to segregate with the keratoconus (KC) phenotype in whole genome sequencing of a four-generation family. We aimed to evaluate a potential relation between the Pcsk1 gene and corneal phenotype in mouse models.

Methods: Two strains of Pcsk1 mice, one with a knockout (KO) and one with an N222D point mutation, were bred. Central corneal thickness (CCT) was determined using spectral domain optical coherence tomography (SD-OCT) in PC1/3+/+ (n = 12), PC1/3+/K^ (n = 14), PC1/3K^/K^ (n = 5), PC1+/+ (n = 8), PC1+/ N222D (n = 15), and PC1 N222D / N222D (n = 7) mice at 3 and 6 months of age. Pachymetry maps were generated using the Mouse Corneal Analysis Program (MCAP) to process OCT images. Hematoxylin and eosin (H&E) staining using corneal sections from these animals were used to examine morphological changes.

Results: No significant differences in corneal CCT, pachymetry, or morphology were observed among any of the mutant mice compared with their control littermates.

Conclusions: In this setting, neither the N222D point mutation nor the Pcsk1 KO affected the corneal phenotype in two mouse models. The Pcsk1 gene could contribute to keratoconus when paired with additional genetic and environmental factors, not included in this study.

Translational relevance: Genetic factors are known to contribute to the pathogenesis of keratoconus. Although a variant in the Pcsk1 gene has been shown to segregate with keratoconus in a family, there is yet no evidence of Pcsk1 gene mutation correlating with pathogenic corneal phenotype in mouse models.

Purpose: A significant portion of patients with glaucoma exhibit normal intraocular pressure (IOP), suggesting additional risk factors other than elevated IOPs that are critical for the pathogenesis of glaucoma. This study aimed to identify blood biomarkers associated with glaucoma and construct a predictive model for glaucoma.

Methods: Peripheral blood samples from patients with glaucoma and 175 age- and sex-matched healthy controls were analyzed using an automated hematology analyzer. Red blood cell membrane stability was analyzed by an osmotic fragility test. Red blood cell morphologies were analyzed using Giemsa staining. A least absolute shrinkage and selection operator (LASSO) regression model was constructed to predict glaucoma.

Results: LASSO regression analysis of blood biomarkers identified important biomarkers for distinguishing patients with glaucoma from healthy controls, which also implied that red blood cells from patients with glaucoma were more prone to rupture. The erythrocyte osmotic fragility (EOF) test revealed significantly higher erythrocyte fragility in patients with glaucoma compared with healthy controls (P < 0.001). To improve the prediction accuracy, a new LASSO regression model for glaucoma prediction was constructed by integrating EOF. The probability of glaucoma (p) was calculated. The revised predictive model achieved an accuracy of 88.68% with a sensitivity of 92.59% and specificity of 84.62%. Validation using an independent dataset demonstrated the great predictive performance of the predictive model, with an area under the ROC (receiver operating characteristic) curve (AUC) value of 0.97.

Conclusions: Increased EOF is a risk factor for glaucoma. EOF, along with several blood biomarkers, helps diagnose glaucoma.

Translational relevance: Measurement of EOF could facilitate clinical glaucoma diagnosis.

Purpose: This study compares the readability and quality of ChatGPT-4o-generated and American Academy of Ophthalmology (AAO) patient handouts in English and Spanish, as AAO materials are expert developed but not publicly accessible.

Methods: Ten AAO handouts on common ocular conditions were obtained in English and Spanish. ChatGPT-4o was queried to create handouts at an 8th-grade reading level of comparable length. English readability was assessed with Simple Measure of Gobbledygook (SMOG) and Flesch-Kincaid Reading Ease, and Spanish readability was assessed with five validated metrics, including the Szigriszt-Pazos Perspicuity Index (SPPI) and the Flesch-Szigriszt Index (INFLESZ). Grade levels were measured by the Flesch-Kincaid Grade Level and Crawford-Nivel-de-Grado. Three ophthalmologist graders per language assessed content with the Quality of Generated Language Outputs for Patients (QGLOP), indicated their preferred handout, and attempted to identify its source.

Results: ChatGPT-4o handouts had English readability similar to that of AAO (SMOG 7.86 ± 0.53 vs. 7.49 ± 0.76; P = 0.35) but significantly lower Spanish SPPI scores (60.26 ± 4.47 vs. 64.82 ± 2.97; P = 0.04). Grade levels were comparable. English QGLOP scores were similar (14.73 vs. 14.87; P = 0.97), but Spanish ChatGPT handouts scored higher in all categories (15.83 vs. 14.23; P < 0.001). ChatGPT-4o handouts were correctly identified by 17 of 30 in English (56.7%) and by 3 of 30 in Spanish (10.0%). They were favored 23 out of 30 times in English (76.7%) and 29 out of 30 times in Spanish (96.7%).

Conclusions: ChatGPT-generated materials matched or exceeded AAO handouts in readability and content quality, with Spanish versions being most preferred.

Translational relevance: Integrating artificial intelligence-generated patient education materials into ophthalmic care can enhance health literacy for multilingual patient populations.

Purpose: The purpose of this study was to evaluate the accuracy and perceptual quality of red-green-blue (RGB) fundus images generated from red-green (RG) images using Pix2Pix-based generative adversarial network (GAN).

Methods: RG images were extracted from original RGB images recorded by a newer model of the Optos system by turning off the blue laser source. A conditional GAN based on the Pix2Pix architecture was trained to convert this extracted RG image to an RGB image. The performance was evaluated quantitatively by using a five-fold cross-validation and assessments with structural similarity index measure (SSIM), peak signal-to-noise ratio (PSNR), learned perceptual image patch similarity (LPIPS), mean absolute error (MAE), and root mean square error (RMSE). To assess the perceptual accuracy, 39 ophthalmologists examined 60 randomly selected images, 30 true and 30 generated RGB images, using Google Form-based interface. They were instructed to classify each image as a true or an artificial intelligence (AI)-generated image.

Results: The generated images had high similarity to the original RGB images (SSIM = 0.97 ± 0.00; PSNR = 35.74 ± 0.17 decibel [dB]; LPIPS = 0.09 ± 0.00; MAE = 1.76 ± 0.04; and RMSE = 4.25 ± 0.10). The overall correct classification rate was 57.1%. Certain lesion types, such as epiretinal membrane and optic disc cupping, had higher discrimination rates. The receiver operating characteristic (ROC) analysis across all evaluations yielded an area under the curve (AUC) of 0.497 (P = 0.314).

Conclusions: Pix2Pix-based GANs can generate perceptually and structurally consistent RGB images from RG images without lesion-specific attention mechanisms.

Translational relevance: This method enables a reinterpretation of legacy RG fundus images and may support future AI diagnostic applications.

Purpose: The purpose of this study was to prioritize mitochondria-related causal genes (MRGs) involved in myopia by integrating MRGs-related methylation quantitative trait loci (mQTL), expression quantitative trait loci (eQTL), and protein quantitative trait loci (pQTL) data with myopia genome-wide association study (GWAS) data.

Methods: UK Biobank data were utilized for discovery, and FinnGen data for validation. Summary Data-based Mendelian Randomization (SMR) assessed associations between blood-derived MRG-related QTLs and myopia risk, followed by colocalization analysis to evaluate shared genetic etiology. Multi-omics integration linked methylation, expression, and disease risk.

Results: Our integrative SMR analysis of MRGs prioritized BDH1 as the sole robust candidate. We identified a significant association between increased methylation at a promoter-proximal site (cg02569554) and reduced BDH1 expression (odds ratio [OR] = 0.522, 95% confidence interval [CI] = 0.425-0.641, false discovery rate [FDR] = 8.79 × 10⁻⁹). Although other nominal associations were observed, they did not survive multiple testing correction and are thus considered exploratory.

Conclusions: This study provides integrative genetic evidence that BDH1 may contribute to myopia through methylation-regulated expression. However, the findings remain preliminary due to limited cross-layer consistency and require replication and functional validation.

Translational relevance: This study prioritizes BDH1 as a mitochondria-related gene that may influence myopia risk via methylation and expression changes, providing a candidate target for mechanistic research and therapeutic intervention.

Purpose: The purpose of this study was to assess whether vitrectomy size and cutting rate have an impact on cell survival of healthy lymphocytes and lymphoma cells during diagnostic vitrectomy in an in vitro model.

Methods: In the first experiment, healthy lymphocytes were isolated from blood samples of one healthy individual. In the second experiment, cultivated MYD88-positive lymphoma cells derived from a commercially available cell line were used. Suspensions with defined cell concentrations (1 × 103 cells/mL, 1 × 104 cells/mL, and 1 × 106 cells/mL) were subjected to vitrectomy under controlled conditions, using aspiration only or varying cutting rates of 1500/minutes and 5000/minutes with vitrectomy sizes of 20G, 23G, 25G, and 27G. Three technical replicates were performed for each condition. Cell integrity in post-vitrectomy samples was assessed via fluorescence activated cell sorting (FACS). To verify statistically significant differences between groups, the Kruskal-Wallis test was used.

Results: No correlation was observed between lymphocyte survival and either gauge size or lower cutting rates in healthy lymphocytes. However, a significant correlation between vitrectomy size and cell survival was observed for lymphoma cells (Kendall's tau τ = -0.253, P = 0.00048, Kruskal-Wallis χ² = 13.337, P = 0.00396).

Conclusions: Our study found that larger vitrectomy lumen diameters were associated with higher lymphoma cell survival rates, suggesting the use of 23G and 25G instruments in clinical practice for suspected cases.

Translational relevance: This clinical research should provide a guideline for the investigation of suspected lymphoma cases.

Background: Distinguishing uveal melanoma (UM) from uveal nevus (UN) is often challenging yet crucial for appropriate management. Machine learning (ML), particularly deep learning (DL), has emerged as a promising solution to this binary classification task. This study aimed to examine the ability of artificial intelligence (AI) models to differentiate UM from UN based on fundus photographs.

Methods: We systematically searched four databases through July 6, 2025, for studies developing ML models for distinguishing UM from UN using fundus photographs as input. The risk of bias and applicability concerns were assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 tool. The results of primary studies were pooled using random-effects meta-analysis.

Results: Our review included seven articles with 6208 participants in total. Six studies used DL and one applied conventional ML. Only two studies conducted external validation. The proposed algorithms demonstrated a strong performance, achieving a pooled area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity of 0.915, 85.3%, 83.7%, and 87.7%, respectively. In subgroup analysis, externally validated models retained a promising discriminative ability with a subtotal pooled AUC of 0.873.

Conclusions: ML algorithms showcase consistently high performance in differentiating UM from UN based on fundus photographs, supporting their potential role as adjunctive tools in clinical practice. Their objective, reproducible assessments may improve referrals, guide clinical decision-making, and boost diagnostic confidence across health care providers. However, existing evidence is highly heterogeneous and constrained by small dataset sizes and limited external validation. Addressing these gaps through multicenter collaborations and data-sharing initiatives could yield more accurate, robust, and generalizable models.

Translational relevance: This work bridges computational research and ophthalmologic care by demonstrating the potential of AI-based analysis of fundus photographs to assist in differentiating uveal melanocytic tumors.

Purpose: Polygenic risk score for glaucoma is an emerging tool showing effective risk stratification. Previous studies have shown public acceptability for the test as a hypothetical scenario. However, the potential uptake of the test and associated predictors has not been reported.

Methods: Individuals who had previously participated in a questionnaire-based study evaluating attitudes toward polygenic testing for glaucoma were invited to participate in a prospective research study for which they will undergo polygenic testing for glaucoma. Individuals had to be over the age of 50 years and living in South Australia to be eligible and invited to participate. Uptake was assessed by decision to consent to the research project and provide a sample for polygenic testing. Multivariable logistic regression was performed to identify predictor variables associated with uptake.

Results: Of the 418 individuals who completed the initial questionnaire, 292 were eligible to participate, of which 143 enrolled, yielding an uptake rate of 49.0%. The mean age of the enrolled cohort was 67.5 years, 67.1% were female, 83.2% were from an urban residency, and 58.7% had a family history of glaucoma. Interest in testing was not associated with participation (P = 0.61). Predictors of uptake included a positive family history (P < 0.001) urban residency (P = 0.004), higher education (P = 0.036), and obtaining more information (P = 0.027).

Conclusions: This is the first study to demonstrate the potential uptake of polygenic testing for glaucoma among an unaffected population.

Translational relevance: These findings provide useful insights into the predictors of glaucoma polygenic testing to support effective clinical implementation.

Purpose: To present a non-mydriatic hyperspectral retinal camera based on spectral scanning, developed to achieve a practical balance among imaging performance, acquisition speed, and system simplicity for advanced retinal diagnostics.

Methods: The system integrates LED-based broadband illumination, linear variable filters, custom optics, a monochrome sensor, and a motorized three-dimensional stage to capture high-resolution hyperspectral data across 29 wavebands from 450 to 850 nm. Optical performance was evaluated using standard metrics including spectral resolution, irradiance, uniformity, resolving power, field of view, and chromatic focal compensation. Imaging was performed on model eyes and human subjects to assess spectral signature capture and repeatability.

Results: The system achieved a 40° field of view and a spectral resolution ranging from 20 to 80 nm. Chromatic focal correction and illumination uniformity were maintained across the spectral range. In vivo imaging demonstrated the ability to capture distinct spectral signatures of anatomical structures and ocular pathologies. Test-retest assessments showed high repeatability, with spectral variation below 5%. The device operated under non-mydriatic conditions with acquisition times of approximately 300 ms.

Conclusions: The prototype demonstrates reliable and repeatable hyperspectral imaging of the retina in a compact and semi-automated form factor. The system offers a foundation for further optimization, including improved spectral precision, artifact reduction, and increased field of view.

Translational relevance: This technology enables non-invasive, high-content retinal imaging suitable for integration into clinical workflows.