Ophthalmology science最新文献

{"title":"Heterogeneous Visual Function Deficits in Intermediate Age-Related Macular Degeneration: A MACUSTAR Report","authors":"Hannah M.P. Dunbar PhD ,&nbsp;David P. Crabb PhD ,&nbsp;Charlotte Behning MSc ,&nbsp;Alison M. Binns PhD ,&nbsp;Amina Abdirahman ,&nbsp;Jan H. Terheyden MD ,&nbsp;Stephen Poor MRCOphth ,&nbsp;Robert P. Finger MD, PhD ,&nbsp;Sergio Leal MD ,&nbsp;Adnan Tufail MD, FRCOphth ,&nbsp;Frank G. Holz MD ,&nbsp;Matthias Schmid PhD ,&nbsp;Ulrich F.O. Luhmann PhD ,&nbsp;MACUSTAR Consortium","doi":"10.1016/j.xops.2025.100708","DOIUrl":"10.1016/j.xops.2025.100708","url":null,"abstract":"<div><h3>Objective</h3><div>To examine the extent to which visual function in Beckman age-related macular degeneration (AMD) disease stages differs from age-similar peers with no AMD and, using reference limits derived from those with no AMD, test the hypothesis that people with intermediate AMD (iAMD) have heterogeneous visual function deficits.</div></div><div><h3>Design</h3><div>Cross-sectional analyses of a range of baseline visual function measures from the MACUSTAR study—an international, multicenter (n = 20), noninterventional clinical trial.</div></div><div><h3>Participants</h3><div>Five hundred eighty-five participants with iAMD (67% female, mean [standard deviation] age 72 [7] years) were recruited alongside 56 with no AMD (59% female, 68 [6]), 34 with early AMD (79% female, 72 [6]), and 43 with late AMD (49% female, 75 [6]).</div></div><div><h3>Methods</h3><div>Participants performed best-corrected visual acuity (BCVA), low luminance visual acuity (LLVA), Moorfields acuity test (MAT), Pelli-Robson contrast sensitivity (PR-CS), small print standardized International reading speed test (SPS), mesopic and scotopic average threshold (MesAT and ScoAT; macular integrity assessment, iCare), and rod intercept time (RIT; AdaptDx, Lumithera).</div></div><div><h3>Main Outcome Measures</h3><div>Relationship between each visual function measure and disease classification was examined by linear regression adjusted for age, sex, and phakic status. No AMD data were used to estimate normal reference limits for each visual function test. Intermediate AMD scores were dichotomized against reference limits, and the proportion worse than each limit was calculated.</div></div><div><h3>Results</h3><div>Relative to no AMD, SPS was significantly worse in early AMD (<em>P</em> = 0.001); all measures except SPS were significantly reduced in iAMD (<em>P</em> &lt; 0.02), and all measures were markedly reduced in late AMD (<em>P</em> &lt; 0.0001). Thirty-one point three percent of iAMD participants breached reference limits for PR-CS, 29.4% for RIT, 24.1% for LLVA, 23.2% for MAT, 20.5% for BCVA, 19.8% for MesAT, 17.9% for ScoAT, and 12.6% for SPS. Of the iAMD participants, 69.6% and 42.7% breached ≥1 and ≥2 reference limits, respectively, whereas 33.6% and 5.7% would be expected by chance.</div></div><div><h3>Conclusions</h3><div>A large proportion of people with structurally defined iAMD exhibit heterogeneous visual function deficits outside normal reference limits. This observation may be relevant for the design and inclusion criteria of future interventional trials.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 4","pages":"Article 100708"},"PeriodicalIF":3.2,"publicationDate":"2025-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143681912","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhanced Macular Telangiectasia Type 2 Detection: Leveraging Self-Supervised Learning and Ensemble Models","authors":"Shahrzad Gholami PhD ,&nbsp;Lea Scheppke PhD ,&nbsp;Meghana Kshirsagar PhD ,&nbsp;Yue Wu PhD ,&nbsp;Rahul Dodhia PhD ,&nbsp;Roberto Bonelli PhD ,&nbsp;Irene Leung BA ,&nbsp;Ferenc B. Sallo MD, PhD ,&nbsp;Alyson Muldrew PhD ,&nbsp;Catherine Jamison MSc ,&nbsp;Tunde Peto MD, PhD ,&nbsp;Juan Lavista Ferres PhD ,&nbsp;William B. Weeks MD, PhD ,&nbsp;Martin Friedlander MD, PhD ,&nbsp;Aaron Y. Lee MD, MSCI ,&nbsp;Lowy Medical Research Institute","doi":"10.1016/j.xops.2025.100710","DOIUrl":"10.1016/j.xops.2025.100710","url":null,"abstract":"<div><h3>Objective</h3><div>To investigate an ensemble-based approach utilizing deep learning models for accurate and interpretable detection of macular telangiectasia (MacTel) type 2 on OCT imaging.</div></div><div><h3>Design</h3><div>Retrospective analysis of OCT scans, model development, and assessment.</div></div><div><h3>Participants</h3><div>A total of 5200 OCT images from participants in the MacTel Registry conducted by the Lowy Medical Research Institute and from the University of Washington (780 MacTel patients and 1900 non-MacTel patients).</div></div><div><h3>Methods, Intervention, or Testing</h3><div>We trained multiple individual MacTel vs. non-MacTel classification models using traditional supervised learning and self-supervised learning (SSL) and ensembled them using average weighting methods. We investigated diverse methodologies for constructing the ensemble, including varied architectural configurations and learning paradigms of individual models, and manipulating the amount of labeled data accessible for training. Model performance was compared against human expert graders on held-out test set data. Model interpretability was investigated using gradient-weighted class activation maps (Grad-CAM) visualization and by evaluating interrater agreement.</div></div><div><h3>Main Outcome Measures</h3><div>For model performance, area under the receiver operating characteristic curve (AUROC), area under the precision–recall curve (AUPRC), accuracy, sensitivity, and specificity were reported. For interpretability, interrater agreements and Grad-CAM visualization results were evaluated.</div></div><div><h3>Results</h3><div>Despite access to only 419 OCT volumes, including 185 MacTel patients within the 10% labeled training dataset, the ensemble model demonstrated a performance level (AUROC 0.972 [95% confidence interval (CI), 0.971–0.973], AUPRC 0.967 [95% CI, 0.965–0.969], accuracy 91.7%, sensitivity 0.905, and specificity 0.925) comparable to the human experts ensemble (AUROC 0.977 [95% CI, 0.975–0.978], AUPRC 0.987 [95% CI, 0.986–0.987], accuracy 96.8%, sensitivity 0.929, and specificity 1) on a test set of 500 patients. The individual models did not achieve the same performance levels when evaluated separately.</div></div><div><h3>Conclusions</h3><div>Even with limited data, combining SSL with ensemble approaches improved MacTel classification accuracy and interpretation compared to the individual models. Self-supervised learning captures meaningful representations from unlabeled data, a key benefit in the setting of limited data such as with rare diseases.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 4","pages":"Article 100710"},"PeriodicalIF":3.2,"publicationDate":"2025-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143681922","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Modeling Patterns of Medication Adherence in Primary Open-Angle Glaucoma","authors":"Shervonne Poleon PhD ,&nbsp;Michael Twa PhD, OD ,&nbsp;Yu-Mei Schoenberger-Godwin PhD ,&nbsp;Mathew Fifolt PhD ,&nbsp;Lyne Racette PhD","doi":"10.1016/j.xops.2025.100706","DOIUrl":"10.1016/j.xops.2025.100706","url":null,"abstract":"<div><h3>Objective</h3><div>To use group-based trajectory modeling to identify patterns of medication adherence in patients with primary open-angle glaucoma (POAG) and to identify factors associated with each pattern.</div></div><div><h3>Design</h3><div>Prospective cohort study.</div></div><div><h3>Participants</h3><div>Seventy-two patients with POAG who were enrolled in a National Institutes of Health–funded progression study at the University of Alabama at Birmingham were included in this study. Patients were required to be &gt;18 years of age, have a diagnosis of POAG, and be prescribed hypotensive eye drops to treat their glaucoma.</div></div><div><h3>Methods</h3><div>Fifty-two weeks of mean weekly medication adherence data were collected using Medication Event Monitoring Systems. Group-based trajectory modeling was used to estimate models with 2, 3, 4, 5, and 6 medication adherence trajectory groups. Self-reported race and illness perception were included as covariates. The model with the lowest Bayesian information criterion (which provides a measure of the trade-off between model fit and model complexity) and the highest number of clinically relevant trajectory groups was deemed optimal.</div></div><div><h3>Main Outcome Measures</h3><div>Medication adherence trajectory groups.</div></div><div><h3>Results</h3><div>The Bayesian information criterion was −1041.1 for the 2-group model, −755.9 for the 3-group model, −643.8 for the 4-group model, −590.4 for the 5-group model, and −559.0 for the 6-group model. We identified the 4-group model as the most optimal. The 4 trajectory groups estimated by this model were near-perfect adherence (51.8% of participants), good adherence (23.2% of participants), declining adherence (18.1% of participants), and poor adherence (6.9% of participants). Compared with the poor adherence group, a higher illness perception score predicted a lower probability of membership in the good (B = −0.276, <em>P</em> = 0.042) and declining (B = −0.303, <em>P</em> = 0.028) adherence groups.</div></div><div><h3>Conclusions</h3><div>Medication adherence is an important clinical outcome that is associated with disease severity and disease progression in POAG. Further investigation of this important topic may reveal other shared clinical characteristics that can be used to identify patients who may be at risk for adverse clinical outcomes such as glaucoma progression.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 4","pages":"Article 100706"},"PeriodicalIF":3.2,"publicationDate":"2025-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143681921","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploring the Causal Relationships between Lipid Biomarkers and Anti-VEGF Treatment Response in Patients with Neovascular Age-related Macular Degeneration","authors":"Feixiang He MS ,&nbsp;Qifang Chen MS ,&nbsp;Peilin Gu MS ,&nbsp;Xuemei Liu MS ,&nbsp;Yinglian Chen PhD ,&nbsp;Ting Liu PhD, MD ,&nbsp;Chongyi Li PhD, MD","doi":"10.1016/j.xops.2025.100711","DOIUrl":"10.1016/j.xops.2025.100711","url":null,"abstract":"<div><h3>Purpose</h3><div>To identify the connections between lipid biomarkers and the anti-VEGF therapy response in patients with neovascular age-related macular degeneration (nAMD).</div></div><div><h3>Design</h3><div>A bidirectional and multivariable Mendelian randomization study.</div></div><div><h3>Participants</h3><div>The summary statistics for anti-VEGF nAMD treatment response included a total of 128 responders, 51 nonresponders, and 6 908 005 genetic variants available for analysis. The sample size of lipid biomarkers is 441 016 and 12 321 875 genetic variants available for analysis.</div></div><div><h3>Methods</h3><div>Two-sample Mendelian randomization (MR) method was conducted to exhaustively appraise the causalities among 13 lipid biomarkers and the risk of different anti-VEGF treatment responses (including visual acuity [VA] and central retinal thickness [CRT]) for nAMD subtypes.</div></div><div><h3>Main Outcome Measures</h3><div>Thirteen lipid biomarkers, VA, and CRT.</div></div><div><h3>Results</h3><div>A positive causal relationship was identified between triglycerides (TGs), apolipoproteins (Apos) E2, ApoE3, total cholesterol (TC), and VA response to anti-VEGF therapy in patients with nAMD, as confirmed by MR-Egger, weighted median, and weighted mode models. The MR-Egger model yielded statistically significant results for TC, ApoA-I, ApoB, and ApoA-V in relation to the CRT response to anti-VEGF treatment in patients with nAMD. In the reverse MR, the MR-Egger model identified significant causal relationships between ApoA-I, low-density lipoprotein cholesterol (LDL-c), ApoE3, and ApoF and the VA response. However, this was not the case in the weighted median and weighted mode models. In the MR-Egger model, ApoB, LDL-c, ApoE3, and ApoM were identified as significantly influencing the CRT response. In the multisample MR analysis, TC, high-density lipoprotein cholesterol, LDL-c, and TG were found to be causally related to VA response, and TC was also identified as being causally related to the CRT response to anti-VEGF therapy in patients with nAMD.</div></div><div><h3>Conclusions</h3><div>This MR study suggests unidirectional causality between TG and ApoE3 and the response to anti-VEGF treatment in patients with nAMD.</div></div><div><h3>Financial Disclosure(s)</h3><div>The author(s) have no proprietary or commercial interest in any materials discussed in this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 4","pages":"Article 100711"},"PeriodicalIF":3.2,"publicationDate":"2025-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143697536","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Submacular Choroidal Arteries","authors":"Michel Paques MD, PhD ,&nbsp;Zosia Bratasz PhD ,&nbsp;Leo Puyo PhD ,&nbsp;Céline Chaumette BS ,&nbsp;Daniela Castro Farias MD ,&nbsp;Michael Atlan PhD ,&nbsp;Sarah Mrejen MD","doi":"10.1016/j.xops.2025.100709","DOIUrl":"10.1016/j.xops.2025.100709","url":null,"abstract":"<div><h3>Objective</h3><div>To document the aspect, topography and morphometry of normal human choroidal arteries in the posterior pole by laser Doppler holography (LDH) and OCT.</div></div><div><h3>Design</h3><div>Cross-sectional study.</div></div><div><h3>Subjects</h3><div>Fifty-four eyes of 27 healthy subjects.</div></div><div><h3>Methods</h3><div>A prototypic LDH system captured the laser Doppler shift of the choroidal circulation within the central 20°. Doppler shifts were filtered to extract high velocity vessels. Images of choroidal arteries identified by LDH were subsequently registered with <em>en face</em> and cross-sectional OCT images. Subsequently, the diameters of macular choroidal arteries and their correlation to central choroidal thickness was measured on OCT B-scans.</div></div><div><h3>Main Outcome Measures</h3><div>Spatial disposition, distribution, and diameters of choroidal arteries.</div></div><div><h3>Results</h3><div>Choroidal arteries were identified by LDH and OCT from their emergence from short posterior ciliary arteries (sPCAs), and could be traced to second and third divisions. In the 8 eyes that underwent LDH, 7 of 8 (88%) showed a horizontal first-order artery within 0.5 disc diameter from the fovea. OCT B-scans showed that first-order arteries were located along the sclera-choroid interface; around arteries, the choroidal tissue formed a pyramid-shaped avascular structure with a posterior base contiguous and isoreflective to the sclera. In a cohort of 49 eyes, the diameter of horizontal submacular arteries (average [± standard deviation] 136.3 μm [±47]; range, 70–209 μm) was weakly correlated to central choroidal thickness (<em>P</em> = 0.09).</div></div><div><h3>Conclusions</h3><div>First-order choroidal arteries emerging from sPCAs are located along the sclerochoroidal interface and are surrounded by a pyramid-shaped avascular space, which contributes to differentiate them from veins. The majority of normal eye show a submacular first-order artery running horizontally toward the temporal periphery. These results will pave the way for a better knowledge of diseases affecting the choroidal circulation.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 3","pages":"Article 100709"},"PeriodicalIF":3.2,"publicationDate":"2025-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143551934","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"In Vivo Noncontact Imaging of Conjunctival Goblet Cells with Customized Widefield Fluorescence Microscopy","authors":"Yushuang Liu MS ,&nbsp;Zhengyu Duan PhD ,&nbsp;Zhongzhou Luo BEng,&nbsp;Runze Zhang MS,&nbsp;Jiaxiong Li MS,&nbsp;Jinze Zhang PhD,&nbsp;Zeyu Meng MS,&nbsp;Bowen Wang MD, PhD,&nbsp;Jin Yuan MD, PhD,&nbsp;Peng Xiao PhD","doi":"10.1016/j.xops.2025.100712","DOIUrl":"10.1016/j.xops.2025.100712","url":null,"abstract":"<div><h3>Purpose</h3><div>Conjunctival goblet cells (CGCs) play a crucial role in maintaining ocular surface health by producing mucins. However, assessing CGC changes in ocular diseases remains limited by invasive techniques and subjective evaluations. This study aims to develop a noncontact cellular resolution fluorescence microscopy for in vivo CGC imaging and investigate CGC dynamics in a dry eye disease (DED) mouse model.</div></div><div><h3>Design</h3><div>Experimental study.</div></div><div><h3>Subjects</h3><div>Freshly ex vivo porcine eyes, New Zealand white rabbits, and C57BL/6 mice.</div></div><div><h3>Methods</h3><div>Based on the intrinsic fluorescence properties of moxifloxacin, a high-resolution noncontact widefield fluorescence microscopy (WFFM) was customized with an all-in-focus algorithm to optimize in vivo CGC imaging over the curved conjunctival surface. A DED mouse model was established by topically applying 0.2% benzalkonium chloride (BAC) to the ocular surface daily for 7 days, followed by a 7-day recovery period without BAC. In vivo CGC alterations were assessed using WFFM on days 0, 3, 7, and 14. Additional assessments included the phenol red thread tear test, corneal sodium fluorescein staining, and periodic acid–Schiff (PAS) assay.</div></div><div><h3>Main Outcome Measures</h3><div>Conjunctival goblet cell density and area ratio.</div></div><div><h3>Results</h3><div>The WFFM system achieved a cellular resolution of 1 μm and a field of view of 1.4 mm × 1.4 mm. Imaging validation in mice and rabbits allowed for the distinguishing and quantitative assessment of individual CGCs or clusters on the curved conjunctival surface in vivo. Significant reductions in CGC density and area ratio on days 3 and 7 after BAC induction were observed in DED mouse in vivo with WFFM, with their values returning to the baseline 7 days after BAC removal, which was consistent with PAS staining results.</div></div><div><h3>Conclusions</h3><div>The customized WFFM enables in vivo cellular imaging of CGCs, offering a safe and accurate method for continuous monitoring of CGC pathophysiology in ocular surface diseases such as DED.</div></div><div><h3>Financial Disclosures</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 3","pages":"Article 100712"},"PeriodicalIF":3.2,"publicationDate":"2025-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143570501","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"How Foundational Is the Retina Foundation Model? Estimating RETFound’s Label Efficiency on Binary Classification of Normal versus Abnormal OCT Images","authors":"David Kuo MD ,&nbsp;Qitong Gao PhD ,&nbsp;Dev Patel MS ,&nbsp;Miroslav Pajic PhD ,&nbsp;Majda Hadziahmetovic MD","doi":"10.1016/j.xops.2025.100707","DOIUrl":"10.1016/j.xops.2025.100707","url":null,"abstract":"&lt;div&gt;&lt;h3&gt;Objective&lt;/h3&gt;&lt;div&gt;While the availability of public internet-scale datasets of images and language has catalyzed remarkable progress in machine learning, medical datasets are constrained by regulations protecting patient privacy and the time and cost required for curation and labeling. Self-supervised learning or pretraining has demonstrated great success in learning meaningful representations from large unlabeled datasets to enable efficient learning on downstream tasks. In ophthalmology, the RETFound model, a large vision transformer (ViT-L) model trained by masked autoencoding on 1.6 million color fundus photos and OCT B-scans, is the first model pretrained at such scale for ophthalmology, demonstrating strong performance on downstream tasks from diabetic retinopathy grading to stroke detection. Here, we measure the label efficiency of the RETFound model in learning to identify normal vs. abnormal OCT B-scans obtained as part of a pilot study for primary care-based diabetic retinopathy screening in North Carolina.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Design&lt;/h3&gt;&lt;div&gt;The 1150 TopCon Maestro OCT central B-scans (981 normal and 169 abnormal) were randomly split 80/10/10 into training, validation, and test datasets. Model training and hyperparameter tuning were performed on the training set guided by validation set performance. The best performing models were then evaluated on the final test set.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Subjects&lt;/h3&gt;&lt;div&gt;Six hundred forty-seven patients with diabetes in the Duke Health System participating in primary care diabetic retinopathy screening contributed 1150 TopCon Maestro OCT central B-scans.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Methods&lt;/h3&gt;&lt;div&gt;Three models (ResNet-50, ViT-L, and RETFound) were fine-tuned on the full training dataset of 915 OCT B-scans and on smaller training data subsets of 500, 250, 100, and 50 OCT B-scans, respectively, across 3 random seeds.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Main Outcome Measures&lt;/h3&gt;&lt;div&gt;Mean accuracy, area under the receiver operator curve (AUROC), area under the precision recall curve (AUPRC), F1 score, precision, and recall on the final held-out test set were reported for each model.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Results&lt;/h3&gt;&lt;div&gt;Across 3 random seeds and all training dataset sizes, RETFound outperformed both ResNet-50 and ViT-L on all evaluation metrics on the final held-out test dataset. Large vision transformer and ResNet-50 performed comparably at the largest training dataset sizes of 915 and 500 OCT B-scans; however, ResNet-50 suffered more pronounced performance degradation at the smallest dataset sizes of 100 and 50 OCT B-scans.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Conclusions&lt;/h3&gt;&lt;div&gt;Our findings validate the benefits of RETFound's additional retina-specific pretraining. Further research is needed to establish best practices for fine-tuning RETFound to downstream tasks.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;Financial Disclosure(s)&lt;/h3&gt;&lt;div&gt;Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.&lt;/div&gt;","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 3","pages":"Article 100707"},"PeriodicalIF":3.2,"publicationDate":"2025-01-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143592630","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Sectoral Changes in Neuroretinal Rim Pallor Across Refractive Error","authors":"Fabian Yii ,&nbsp;Samuel Gibbon MSc ,&nbsp;Tom MacGillivray PhD","doi":"10.1016/j.xops.2025.100705","DOIUrl":"10.1016/j.xops.2025.100705","url":null,"abstract":"<div><h3>Purpose</h3><div>To investigate the association between spherical equivalent refraction (SER) and pallor in different neuroretinal rim (NRR) sectors.</div></div><div><h3>Design</h3><div>Population-based cross-sectional study.</div></div><div><h3>Participants</h3><div>Normal eyes of 24 057 healthy participants aged 40 to 70 years from the UK Biobank.</div></div><div><h3>Methods</h3><div>Pallor in different NRR sectors was quantitatively derived from color fundus photographs using automated software. We first examined the association between SER and pallor in each NRR sector—controlling for age, sex, ethnicity (White vs. non-White), intraocular pressure, and mean blood pressure. We then incorporated disc–fovea distance (the shortest distance from the center of the disc to the fovea) and temporal arterial/venous concavity (extent to which the temporal artery/vein curved inwardly toward the fovea) as additional independent variables, as these features have been suggested to reflect the degree of axonal stretching at the posterior pole.</div></div><div><h3>Main Outcome Measures</h3><div>Pallor in the temporal, temporal inferior, nasal inferior, nasal, nasal superior, and temporal superior sectors of the NRR.</div></div><div><h3>Results</h3><div>Moving from the temporal sector to the temporal superior sector, NRR pallor varied in an asymmetrical U-shaped pattern, with the least pallor observed nasally. White participants tended to have paler NRR, but the association between SER and pallor did not differ between ethnic groups (no interaction effect between SER and ethnicity). Decreasing SER was associated with increasing pallor in all 6 NRR sectors (all <em>P</em> &lt; 0.001), but the temporal (ß: −0.009, 95% confidence interval: −0.011 to −0.008) and temporal inferior (ß: −0.008, 95% confidence interval: −0.009 to −0.007) sectors exhibited the steepest increase. The rate of increase diminished by half toward the more nasal/central sectors, and by another half in the nasal-most sector. Consistent with these changes, increasing disc–fovea distance and temporal arterial/venous concavity resulted in up to 4 times as much pallor temporally compared with nasally. These retinal changes accounted for approximately ≥50% of the effect of SER on NRR pallor.</div></div><div><h3>Conclusions</h3><div>Decreasing SER increases NRR pallor approximately 4 times faster temporally than nasally. The association between SER and NRR pallor is primarily attributable to changes in disc–fovea distance and temporal arterial/venous concavity. These findings suggest that the papillomacular nerve fiber bundle, linked to the temporal NRR, is most susceptible to myopic stretching.</div></div><div><h3>Financial Disclosure(s)</h3><div>The author(s) have no proprietary or commercial interest in any materials discussed in this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 3","pages":"Article 100705"},"PeriodicalIF":3.2,"publicationDate":"2025-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143611596","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Bile Acid Metabolism Changes in Patients with a CRB1-Associated Inherited Retinal Degeneration","authors":"Lude Moekotte MD, PhD ,&nbsp;Joke H. de Boer MD, PhD ,&nbsp;Sanne Hiddingh MSc ,&nbsp;Bram Gerritsen PhD ,&nbsp;Jutta Lintelmann PhD ,&nbsp;Alexander Cecil PhD ,&nbsp;L. Ingeborgh van den Born MD, PhD ,&nbsp;Xuan-Thanh-An Nguyen MD, PhD ,&nbsp;Camiel J.F. Boon MD, PhD ,&nbsp;Maria M. van Genderen MD, PhD ,&nbsp;Jonas J.W. Kuiper PhD","doi":"10.1016/j.xops.2025.100704","DOIUrl":"10.1016/j.xops.2025.100704","url":null,"abstract":"<div><h3>Purpose</h3><div>To compare the plasma metabolic profile of patients with a <em>Crumbs homolog 1</em>-associated inherited retinal degeneration (<em>CRB1</em>-IRD) with that of healthy controls (HCs).</div></div><div><h3>Design</h3><div>A case-control study.</div></div><div><h3>Participants</h3><div>A cohort of 30 Dutch patients with <em>CRB1</em>-IRD and 29 Dutch HCs.</div></div><div><h3>Methods</h3><div>The MxP Quant 500 Kit was used for measuring metabolite concentrations. We fitted a linear regression model with adjustments for age and sex based on the concentration of metabolites in micromolar (micromoles per liter) or on the sums and ratios of metabolites to determine differences between patients and controls.</div></div><div><h3>Main Outcome Measures</h3><div>Plasma concentration of 619 metabolites.</div></div><div><h3>Results</h3><div>Overrepresentation of pathways among metabolites associated strongest to <em>CRB1</em>-IRDs (<em>P</em> &lt; 0.05, n = 62) identified amino acid pathways (such as β-alanine, histidine, and glycine/serine) and bile acid biosynthesis, driven by a decrease in deoxycholic acid derivatives produced by gut microbiota. Enrichment analysis of metabolic classes across the plasma metabolic profile further identified significant positive enrichment for lipid metabolites glycerophospholipids, cholesterol esters, and ceramides, and significant depletion for bile acid metabolites. Further investigation of the sums and ratios (i.e., metabolism indicators) ascertained a significant decrease in intestinal <em>microbial-dependent secondary</em> bile acid classes.</div></div><div><h3>Conclusions</h3><div>Lipid metabolic alterations and decreased microbiota-related secondary bile acid concentrations indicate significant alterations in gut <em>metabolism</em> in patients with a <em>CRB1</em>-IRD.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 4","pages":"Article 100704"},"PeriodicalIF":3.2,"publicationDate":"2025-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143833484","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantitative Mapping of Posterior Eye Curvature in Children Using Distortion-Corrected OCT: Insights into Temporal Region Morphology","authors":"Takahiro Hiraoka MD, PhD ,&nbsp;Masato Tamura MS ,&nbsp;Yoshikiyo Moriguchi PhD ,&nbsp;Riku Kuji MS ,&nbsp;Toshihiro Mino PhD ,&nbsp;Masahiro Akiba PhD ,&nbsp;Yosuke Takahashi OD ,&nbsp;Kenichi Yoshino MD, PhD ,&nbsp;Yoshimi Sugiura MD, PhD ,&nbsp;Toshifumi Mihashi PhD ,&nbsp;Tetsuro Oshika MD, PhD","doi":"10.1016/j.xops.2024.100695","DOIUrl":"10.1016/j.xops.2024.100695","url":null,"abstract":"<div><h3>Purpose</h3><div>To explore the curvature distribution in the posterior eye among school-aged children using distortion-corrected widefield OCT and its relationship with biometric variables.</div></div><div><h3>Design</h3><div>Cross-sectional, observational study.</div></div><div><h3>Participants</h3><div>Eighty-eight children.</div></div><div><h3>Methods</h3><div>A swept-source-OCT prototype system with a repetition rate of 400 kHz was used to capture widefield retinal OCT images with a field-of-view of 68 × 68 degrees. The acquired OCT images were corrected for distortion using ocular optical information obtained separately for each eye. The posterior eye curvature was represented by the Gaussian curvature which was derived from Bruch’s membrane segmentation. The mean Gaussian curvature was assessed across 4 regions set centered on the fovea, considering axial length (AL), refractive error, age, and choroidal thickness (ChT). Additionally, we identified the entry site of the long posterior ciliary artery (LPCA) into the choroid.</div></div><div><h3>Main Outcome Measures</h3><div>Local curvature distribution in the posterior eye.</div></div><div><h3>Results</h3><div>A total of 176 eyes were imaged, with 7 excluded due to low image quality. Analysis of 169 OCT images revealed bilateral symmetry in choroidal vascular patterns and posterior eye curvature. Significant correlations were noted between mean curvature and AL, refractive error, and ChT in the superior, macular, and inferior regions. However, the temporal region exhibited reversed correlation trends. A local maximum curvature point was commonly observed in the temporal region, potentially linked to the LPCA entry site.</div></div><div><h3>Conclusions</h3><div>Our study provided a quantitative analysis of posterior eye curvature in children, highlighting a local maximum curvature point in the temporal region. Interestingly, the relationships between mean curvature and biometric variables in the temporal region contradicted those in the other 3 regions. Further investigation is necessary to elucidate its origin and implications for ocular development.</div></div><div><h3>Financial Disclosure(s)</h3><div>Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.</div></div>","PeriodicalId":74363,"journal":{"name":"Ophthalmology science","volume":"5 3","pages":"Article 100695"},"PeriodicalIF":3.2,"publicationDate":"2025-01-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143511010","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}