Shyamanga Borooah

Background: Late-Onset Retinal Degeneration (L-ORD) is an inherited retinal dystrophy (IRD) that typically manifests with delayed dark adaptation by the fifth decade of life and progresses to initial central visual impairment after age 60 years. Here, we describe the clinical and molecular characterisation of two C1QTNF5 missense variants identified in individuals with IRD that do not share the characteristic retinal phenotype of L-ORD and exhibit variable age of onset and disease severity. Methods: Probands from two unrelated families underwent genetic testing and comprehensive ophthalmic phenotyping, including multimodal retinal imaging and electrophysiological assessment. This allowed comparison to the previously reported ophthalmic phenotype in two additional families for which the genetic cause of IRD had been previously reported. Genetic analysis was performed using next-generation sequencing, with variants classified according to ACMG/AMP guidelines. In silico analyses assessed evolutionary conservation, predicted pathogenicity, and structural consequences. To experimentally assess variant effects, wild-type and mutant C1QTNF5 constructs were expressed in hTERT-RPE1 cells, and protein folding, stability, and secretion were evaluated. Results: A de novo heterozygous variant, c.548G>A p.(G183E) in C1QTNF5 in a proband, was classified as likely pathogenic. The genetic cause of IRD in the other unrelated probands was C1QTNF5 c.538C>G, p.(Q180E), a known pathogenic variant. The missense variants affect highly conserved residues and are absent from population databases. Structural modelling of both variants predicted destabilisation of C1QTNF5 for both variants, although significantly less severe than for previously identified pathogenic L-ORD variants. In vitro analyses revealed reduced stability of both mutant proteins expressed alone, but when co-expressed with wild-type (WT) C1QTNF5, differences between these variants were observed. In contrast to L-ORD variants, C1QTNF5 containing p.G183E or p.Q180E, could multimerise, generating the expected ratio of trimer and high molecular weight C1QTNF5 multimers. C1QTNF5 p.Q180E/WT co-assembly destabilised the protein and decreased the amount of total C1QTNF5 secreted, but p.G183E was secreted at an equivalent level to WT C1QTNF5. Conclusions: These findings expand the variant and mechanistic spectrum of C1QTNF5-associated retinal disease and highlight C1QTNF5 protein misfolding as a key pathogenic mechanism. Integrating clinical phenotyping with structural and functional analyses is critical for accurate variant interpretation and molecular diagnosis in C1QTNF5-associated retinal disease.

Background: PRPH2- associated retinal degeneration (PARD) is one of the most common causes of inherited retinal disease and can impair visual function and quality of life. While genetic and clinical features of PARD have been described, patient-reported experiences of visual symptoms remain less well characterized. This study aims to evaluate self-reported visual function and quality of life among individuals with genetically confirmed PARD. Methods: This cross-sectional study surveyed patients with confirmed PARD at a single academic center using the National Eye Institute Visual Function Questionnaire-25 (VFQ-25). The questionnaire was administered digitally and assessed 11 vision-related domains, including general vision, near and distance activities, and vision-related mental health. A composite score was calculated as the mean of all subscale scores excluding the general health item. Descriptive statistics were used to summarize responses, with higher scores indicating better perceived visual function. Results: Twenty-two participants completed the VFQ-25. The median (IQR) composite score was 76.9 (16.4), reflecting moderate perceived visual dysfunction. The highest-rated domains were ocular pain (87.5 (37.5)) and color vision (100.0 (0.0)). Domains related to functional activities demonstrated greater impairment, including general vision (80.0 (20.0)), near activities (79.2 (39.6)), distance activities (75.0 (29.2)), and driving (75.0 (33.3)). The vision-specific mental health domain had the lowest score (59.4 (40.6)), underscoring the psychological impact of PARD. Conclusions: Individuals with PARD report significant challenges in visual function and emotional well-being. These results highlight the importance of incorporating patient-reported outcome measures into the clinical characterization of PARD and the design of future therapeutic studies.

The purpose of this study was to compare standard 85 kilohertz (kHz) and newer high-speed 125 kHz optical coherence tomography angiography (OCTA) scanning modes in healthy and diseased eyes, assessing differences in acquisition time, image quality, and quantitative vascular metrics. This retrospective cross-sectional study included 200 eyes from 100 consecutive patients with a wide range of retinal diseases and healthy controls. Each eye underwent OCTA imaging at both 85 kHz and 125 kHz using 10 × 10 mm fields of view. Acquisition time was measured manually and automatically using the Heidelberg log file. Three retinal slabs, the superficial capillary plexus (SCP), deep capillary plexus (DCP), and choriocapillaris (CC),were analyzed. Two masked retinal specialists graded images for clinical utility and motion artifacts. Quantitative vascular metrics were extracted using AngioTool version 2.0, and Q-scores were recorded for all scans. Main outcome measures included acquisition time, Q-score, qualitative grading, and quantitative vascular parameters across the three slabs. High-speed 125 kHz scans significantly reduced mean acquisition time compared with 85 kHz (P < 0.001). Q-scores did not differ significantly between modes (P = 0.37). Qualitative grading favored 125 kHz scans, showing improved interpretability and fewer motion artifacts. Quantitative vascular metrics across SCP, DCP, and CC demonstrated no significant differences between scan speeds. High-speed 125 kHz OCTA markedly decreases acquisition time without compromising image quality, Q-score, or quantitative vascular parameters, supporting its use in routine and research imaging. Adoption of 125 kHz OCTA may enhance clinical workflow and patient comfort while preserving quantitative reliability, enabling more efficient high-speed imaging in practice.

Purpose: This study validates a deep learning–based artificial intelligence tool for quantifying macular edema (ME) intraretinal fluid (IRF) volumes in retinitis pigmentosa, and through longitudinal analysis of IRF, provides new insight into treatment efficacy and disease natural history. Methods: This retrospective, longitudinal study identified patients with retinitis pigmentosa with ME. A commercially available retinal analysis tool quantified IRF and was validated for segmentation of ME using spectral-domain optical coherence tomography volume scans. Baseline analysis of IRF versus traditional central subfield thickness and longitudinal analyses of IRF versus treatment and best-corrected visual acuity were performed. Results: Forty-four patients were identified. For treatment studies, 52 eyes were in the treated group and 14 eyes in the untreated ME group. Mean follow-up was 5.3 exams (3.7, 6.9) over 2.3 years (1.7, 3.0). Software validation compared automated and manual IRF segmentation of 490 image pairs, finding a Dice coefficient of 0.928 (95% confidence interval: 0.92‒0.99). Cohort mean IRF volume was 230.85 nL (57.42, 403.91) at baseline. Intraretinal fluid change in eyes treated with topical carbonic anhydrase inhibitors was −2.1 nL/year ( P = 0.81). Oral acetazolamide-treated eyes had significant IRF reduction (−33.6 nL/year, P = 0.009) and significant improvements in best-corrected visual acuity (logMAR/year; Early Treatment Diabetic Retinopathy Study letters equivalent) (−0.041; +2 letters) ( P = 0.025). Conclusion: A deep learning tool was able to rapidly and accurately quantify IRF in retinitis pigmentosa–associated ME. Using this analysis tool, we confirmed that treatment with acetazolamide led to significant reduction in long-term IRF. Structural changes (IRF) only translated to significant functional improvements (best-corrected visual acuity) in eyes treated with acetazolamide.

This study aimed to assess how multimodal large language models (MLLM) diagnose and differentiate Pentosan Polysulfate (PPS) Maculopathy from other phenotypic mimics. A retrospective review of clinical records and multimodal retinal imaging was conducted with patients from the Shiley Eye Institute and Casey Eye Institute. Four MLLMs (ChatGPT-4o, Claude 3.5 Sonnet, Google Gemini 1.5 Pro, Perplexity Llama 3.1 Sonar/Default) along with human retinal specialists answered prompts based on retinal imaging and demographic data. Performance was evaluated using accuracy, sensitivity and specificity estimates. The study included 126 eyes from 63 patients, with 36 eyes with PPS maculopathy, 50 eyes with Stargardt disease, and 40 eyes with PRPH2 -associated multifocal pattern dystrophy. MLLMs showed improved accuracy and sensitivity when answer choices were restricted, with ChatGPT consistently performing best when all imaging modalities were prompted together. The inclusion of demographic data further enhanced performance in prompts with limited answer choices. Human retinal specialist evaluations aligned with MLLM performance trends and also improved with demographic data. While MLLMs show diagnostic potential, further refinement is needed before clinical implementation. These findings highlight the importance of prompt design and demographic data to optimize MLLM performance with retinal imaging modalities.

To evaluate the diagnostic and treatment recommendation performance of multimodal large language models (MLLMs) in identifying and classifying retinal lesions as choroidal nevus or melanoma, as well as compare their performance with expert human graders. This retrospective cross-sectional study included 48 eyes from 47 patients diagnosed with either choroidal nevus or melanoma. Patient demographics, including age, sex, ethnicity, best-corrected visual acuity (BCVA), and symptoms, were documented. Color fundus, autofluorescence, optical coherence tomography, and B-scan images were collected. The ocular images and patient characteristics were presented to ChatGPT 4.0, Gemini Advanced 1.5 Pro, and Perplexity Pro. Responses were recorded and compared with the clinical diagnoses and treatment recommendations made by two expert human graders. Diagnostic and treatment agreement, accuracy, sensitivity, and specificity were analyzed. Gemini consistently outperformed ChatGPT and Perplexity across diagnostic and treatment prompts. The highest model performance was observed for prompts requesting treatment recommendations with clinical information, where Gemini achieved the highest accuracy (0.725), followed by Perplexity (0.647) and ChatGPT (0.314). Performance was lowest for prompts requiring strict clinical criteria, with all models showing poor sensitivity. Both human graders outperformed all MLLMs in accuracy and sensitivity on most prompts (P < 0.005). Accuracy did not improve when provided demographic or clinical data, except for Gemini. Human graders outperform current MLLMs, which show only moderate ability to diagnose choroidal nevi or melanoma from imaging. This study highlights limitations and potential of MLLMs in aiding diagnosis and treatment of choroidal lesions.

This single-center retrospective study evaluated the performance of four multimodal large language models (MLLMs) (ChatGPT-4o, Claude 3.5 Sonnet, Google Gemini 1.5 Pro, Perplexity Sonar Large) in detecting and grading the severity of age-related macular degeneration (AMD) from ultrawide field fundus images. Images from 76 patients (136 eyes; mean age 81.1 years; 69.7% female) seen at the University of California San Diego were graded independently for AMD severity by two junior retinal specialists (and an adjudicating senior retina specialist for disagreements) using the Age-Related Eye Disease Study (AREDS) classification. The cohort included 17 (12.5%) eyes with ‘No AMD’, 18 (13.2%) with ‘Early AMD’, 50 (36.8%) with ‘Intermediate AMD’, and 51 (37.5%) with ‘Advanced AMD’. Between December 2024 and February 2025, each MLLM was prompted with single images and standardized queries to assess the primary outcomes of accuracy, sensitivity, and specificity in binary disease classification, disease severity grading, open-ended diagnosis, and multiple-choice diagnosis (with distractor diseases). Secondary outcomes included precision, F1 scores, Cohen’s kappa, model performance comparisons, and error analysis. ChatGPT-4o demonstrated the highest accuracy for binary disease classification [mean 0.824 (95% confidence interval (CI)): 0.743, 0.875)], followed by Perplexity Sonar Large [mean 0.815 (95% CI: 0.744, 0.879)], both of which were significantly more accurate (P < 0.00033) Than Gemini 1.5 Pro [mean 0.669 (95% CI: 0.581, 0.743)] and Claude 3.5 Sonnet [mean 0.301 (95% CI: 0.221, 0.375)]. For severity grading, Perplexity Sonar Large was most accurate [mean 0.463 (95% CI: 0.368, 0.537)], though differences among models were not statistically significant. ChatGPT-4o led in open-ended and multiple-choice diagnostic tasks. In summary, while MLLMs show promise for automated AMD detection and grading from fundus images, their current reliability is insufficient for clinical application, highlighting the need for further model development and validation.

This paper presents the first study to apply tunable wavelet units (UwUs) for 3D retinal layer segmentation from Optical Coherence Tomography (OCT) volumes. To overcome the limitations of conventional max-pooling, we integrate three wavelet-based downsampling modules, OrthLattUwU, BiorthLattUwU, and LS-BiorthLattUwU, into a motion-corrected MGU-Net architecture. These modules use learnable lattice filter banks to preserve both low- and high-frequency features, enhancing spatial detail and structural consistency. Evaluated on the Jacobs Retina Center (JRC) OCT dataset, our framework shows significant improvement in accuracy and Dice score, particularly with LS-BiorthLattUwU, highlighting the benefits of tunable wavelet filters in volumetric medical image segmentation.