Vision-Language Models for Feature Detection of Macular Diseases on Optical Coherence Tomography.

Importance: Vision-language models (VLMs) are a novel artificial intelligence technology capable of processing image and text inputs. While demonstrating strong generalist capabilities, their performance in ophthalmology has not been extensively studied. Objective: To assess the performance of the Gemini Pro VLM in expert-level tasks for macular diseases from optical coherence tomography (OCT) scans. Design, Setting, and Participants: This was a cross-sectional diagnostic accuracy study evaluating a generalist VLM on ophthalmology-specific tasks using the open-source Optical Coherence Tomography Image Database. The dataset included OCT B-scans from 50 unique patients: healthy individuals and those with macular hole, diabetic macular edema, central serous chorioretinopathy, and age-related macular degeneration. Each OCT scan was labeled for 10 key pathological features, referral recommendations, and treatments. The images were captured using a Cirrus high definition OCT machine (Carl Zeiss Meditec) at Sankara Nethralaya Eye Hospital, Chennai, India, and the dataset was published in December 2018. Image acquisition dates were not specified. Exposures: Gemini Pro, using a standard prompt to extract structured responses on December 15, 2023. Main Outcomes and Measures: The primary outcome was model responses compared against expert labels, calculating F1 scores for each pathological feature. Secondary outcomes included accuracy in diagnosis, referral urgency, and treatment recommendation. The model's internal concordance was evaluated by measuring the alignment between referral and treatment recommendations, independent of diagnostic accuracy. Results: The mean F1 score was 10.7% (95% CI, 2.4-19.2). Measurable F1 scores were obtained for macular hole (36.4%; 95% CI, 0-71.4), pigment epithelial detachment (26.1%; 95% CI, 0-46.2), subretinal hyperreflective material (24.0%; 95% CI, 0-45.2), and subretinal fluid (20.0%; 95% CI, 0-45.5). A correct diagnosis was achieved in 17 of 50 cases (34%; 95% CI, 22-48). Referral recommendations varied: 28 of 50 were correct (56%; 95% CI, 42-70), 10 of 50 were overcautious (20%; 95% CI, 10-32), and 12 of 50 were undercautious (24%; 95% CI, 12-36). Referral and treatment concordance were very high, with 48 of 50 (96%; 95 % CI, 90-100) and 48 of 49 (98%; 95% CI, 94-100) correct answers, respectively. Conclusions and Relevance: In this study, a generalist VLM demonstrated limited vision capabilities for feature detection and management of macular disease. However, it showed low self-contradiction, suggesting strong language capabilities. As VLMs continue to improve, validating their performance on large benchmarking datasets will help ascertain their potential in ophthalmology.

Risk Stratification for Diabetic Retinopathy Screening Order Using Deep Learning: A Multicenter Prospective Study.

Purpose: Real-world evaluation of a deep learning model that prioritizes patients based on risk of progression to moderate or worse (MOD+) diabetic retinopathy (DR). Methods: This nonrandomized, single-arm, prospective, interventional study included patients attending DR screening at four centers across Thailand from September 2019 to January 2020, with mild or no DR. Fundus photographs were input into the model, and patients were scheduled for their subsequent screening from September 2020 to January 2021 in order of predicted risk. Evaluation focused on model sensitivity, defined as correctly ranking patients that developed MOD+ within the first 50% of subsequent screens. Results: We analyzed 1,757 patients, of which 52 (3.0%) developed MOD+. Using the model-proposed order, the model's sensitivity was 90.4%. Both the model-proposed order and mild/no DR plus HbA1c had significantly higher sensitivity than the random order (P < 0.001). Excluding one major (rural) site that had practical implementation challenges, the remaining sites included 567 patients and 15 (2.6%) developed MOD+. Here, the model-proposed order achieved 86.7% versus 73.3% for the ranking that used DR grade and hemoglobin A1c. Conclusions: The model can help prioritize follow-up visits for the largest subgroups of DR patients (those with no or mild DR). Further research is needed to evaluate the impact on clinical management and outcomes. Translational Relevance: Deep learning demonstrated potential for risk stratification in DR screening. However, real-world practicalities must be resolved to fully realize the benefit.

Ageing changes in retinal outer nuclear layer thickness and cone photoreceptor density using adaptive optics-free imaging.

PURPOSE: To investigate age-related changes of the outer nuclear layer (ONL) thickness and cone density, and their associations in healthy participants using a modified, narrow scan-angle Heidelberg Retina Angiograph (HRA2). METHODS: Retinal cones were imaged outside the fovea at 8.8° eccentricity and cone density was compared to ONL thickness measurements obtained by Spectral-Domain Optical Coherence Tomography (SD-OCT) at the same locations. Fifty-six eyes of 56 healthy participants with a median age (interquartile range, IQR) of 37 years (29-55) were included. RESULTS: Median (IQR) cone count was 7,472 (7,188, 7,746) cones/mm2 and median (IQR) ONL thickness was 56 (52, 60) µm for healthy participants. Both cone density and ONL thickness were negatively associated with age: cone density, R2 = 0.16 (F(1,54) = 10.41, P = 0.002); ONL thickness, R2 = 0.12 (F(1,54) = 7.41, P = 0.009). No significant association was seen between cone density and ONL thickness (R2 = 0.03; F(1,54) = 1.66, P = 0.20). CONCLUSION: Cone density was lower, and ONL thinner, in older compared to younger participants, therefore, image-based structural measures should be compared to age-related data. However, cone density and ONL thickness were not strongly associated, indicating that determinants of ONL thickness measurements other than cone density measurements, and including measurement error, have a major influence.

Deep Learning to Detect OCT-derived Diabetic Macular Edema from Color Retinal Photographs: A Multicenter Validation Study.

PURPOSE: To validate the generalizability of a deep learning system (DLS) that detects diabetic macular edema (DME) from 2-dimensional color fundus photographs (CFP), for which the reference standard for retinal thickness and fluid presence is derived from 3-dimensional OCT. DESIGN: Retrospective validation of a DLS across international datasets. PARTICIPANTS: Paired CFP and OCT of patients from diabetic retinopathy (DR) screening programs or retina clinics. The DLS was developed using data sets from Thailand, the United Kingdom, and the United States and validated using 3060 unique eyes from 1582 patients across screening populations in Australia, India, and Thailand. The DLS was separately validated in 698 eyes from 537 screened patients in the United Kingdom with mild DR and suspicion of DME based on CFP. METHODS: The DLS was trained using DME labels from OCT. The presence of DME was based on retinal thickening or intraretinal fluid. The DLS's performance was compared with expert grades of maculopathy and to a previous proof-of-concept version of the DLS. We further simulated the integration of the current DLS into an algorithm trained to detect DR from CFP. MAIN OUTCOME MEASURES: The superiority of specificity and noninferiority of sensitivity of the DLS for the detection of center-involving DME, using device-specific thresholds, compared with experts. RESULTS: The primary analysis in a combined data set spanning Australia, India, and Thailand showed the DLS had 80% specificity and 81% sensitivity, compared with expert graders, who had 59% specificity and 70% sensitivity. Relative to human experts, the DLS had significantly higher specificity (P = 0.008) and noninferior sensitivity (P < 0.001). In the data set from the United Kingdom, the DLS had a specificity of 80% (P < 0.001 for specificity of >50%) and a sensitivity of 100% (P = 0.02 for sensitivity of > 90%). CONCLUSIONS: The DLS can generalize to multiple international populations with an accuracy exceeding that of experts. The clinical value of this DLS to reduce false-positive referrals, thus decreasing the burden on specialist eye care, warrants a prospective evaluation.

Intravitreal injections: past trends and future projections within a UK tertiary hospital.

AIMS: To describe past trends and future projections for the number of intravitreal injections being administered at a large tertiary hospital in London, United Kingdom. METHODS: Retrospective data from Moorfields Eye Hospital were collected using the electronic medical record system. Descriptive statistics were used to visualise overall trends. Time series forecasting was used to predict the number of injections that will be administered up to and including the year 2029. RESULTS: The number of injections has increased nearly 11-fold from 2009 to 2019, with a total of 44,924 injections delivered in 2019. The majority of injections were given for the treatment of neovascular age-related macular degeneration. Aflibercept formed 87% of injections administered in 2019. The number of injections is predicted to continue to increase every year, with nearly 83,000 injections forecasted in the year 2029. CONCLUSION: The demand for intravitreal injections has increased substantially over the last decade and is predicted to further increase. Healthcare systems will need to adapt to accommodate the high demand. Other solutions may include longer-acting therapies to reduce the treatment burden.

Perfluorocarbons Therapeutics in Modern Cancer Nanotechnology for Hypoxiainduced Anti-tumor Therapy.

With an estimated failure rate of about 90%, immunotherapies that are intended for the treatment of solid tumors have caused an anomalous rise in the mortality rate over the past decades. It is apparent that resistance towards such therapies primarily occurs due to elevated levels of HIF-1 (Hypoxia-induced factor) in tumor cells, which are caused by disrupted microcirculation and diffusion mechanisms. With the advent of nanotechnology, several innovative advances were brought to the fore; and, one such promising direction is the use of perfluorocarbon nanoparticles in the management of solid tumors. Perfluorocarbon nanoparticles enhance the response of hypoxia-based agents (HBAs) within the tumor cells and have been found to augment the entry of HBAs into the tumor micro-environment. The heightened penetration of HBAs causes chronic hypoxia, thus aiding in the process of cell quiescence. In addition, this technology has also been applied in photodynamic therapy, where oxygen self-enriched photosensitizers loaded perfluorocarbon nanoparticles are employed. The resulting processes initiate a cascade, depleting tumour oxygen and turning it into a reactive oxygen species eventually to destroy the tumour cell. This review elaborates on the multiple applications of nanotechnology based perfluorocarbon formulations that are being currently employed in the treatment of tumour hypoxia.

Validation and Clinical Applicability of Whole-Volume Automated Segmentation of Optical Coherence Tomography in Retinal Disease Using Deep Learning.

IMPORTANCE: Quantitative volumetric measures of retinal disease in optical coherence tomography (OCT) scans are infeasible to perform owing to the time required for manual grading. Expert-level deep learning systems for automatic OCT segmentation have recently been developed. However, the potential clinical applicability of these systems is largely unknown. OBJECTIVE: To evaluate a deep learning model for whole-volume segmentation of 4 clinically important pathological features and assess clinical applicability. DESIGN, SETTING, PARTICIPANTS: This diagnostic study used OCT data from 173 patients with a total of 15 558 B-scans, treated at Moorfields Eye Hospital. The data set included 2 common OCT devices and 2 macular conditions: wet age-related macular degeneration (107 scans) and diabetic macular edema (66 scans), covering the full range of severity, and from 3 points during treatment. Two expert graders performed pixel-level segmentations of intraretinal fluid, subretinal fluid, subretinal hyperreflective material, and pigment epithelial detachment, including all B-scans in each OCT volume, taking as long as 50 hours per scan. Quantitative evaluation of whole-volume model segmentations was performed. Qualitative evaluation of clinical applicability by 3 retinal experts was also conducted. Data were collected from June 1, 2012, to January 31, 2017, for set 1 and from January 1 to December 31, 2017, for set 2; graded between November 2018 and January 2020; and analyzed from February 2020 to November 2020. MAIN OUTCOMES AND MEASURES: Rating and stack ranking for clinical applicability by retinal specialists, model-grader agreement for voxelwise segmentations, and total volume evaluated using Dice similarity coefficients, Bland-Altman plots, and intraclass correlation coefficients. RESULTS: Among the 173 patients included in the analysis (92 [53%] women), qualitative assessment found that automated whole-volume segmentation ranked better than or comparable to at least 1 expert grader in 127 scans (73%; 95% CI, 66%-79%). A neutral or positive rating was given to 135 model segmentations (78%; 95% CI, 71%-84%) and 309 expert gradings (2 per scan) (89%; 95% CI, 86%-92%). The model was rated neutrally or positively in 86% to 92% of diabetic macular edema scans and 53% to 87% of age-related macular degeneration scans. Intraclass correlations ranged from 0.33 (95% CI, 0.08-0.96) to 0.96 (95% CI, 0.90-0.99). Dice similarity coefficients ranged from 0.43 (95% CI, 0.29-0.66) to 0.78 (95% CI, 0.57-0.85). CONCLUSIONS AND RELEVANCE: This deep learning-based segmentation tool provided clinically useful measures of retinal disease that would otherwise be infeasible to obtain. Qualitative evaluation was additionally important to reveal clinical applicability for both care management and research.

Predicting Incremental and Future Visual Change in Neovascular Age-Related Macular Degeneration Using Deep Learning.

PURPOSE: To evaluate the predictive usefulness of quantitative imaging biomarkers, acquired automatically from OCT scans, of cross-sectional and future visual outcomes of patients with neovascular age-related macular degeneration (AMD) starting anti-vascular endothelial growth factor (VEGF) therapy. DESIGN: Retrospective cohort study. PARTICIPANTS: Treatment-naive, first-treated eyes of patients with neovascular AMD between 2007 and 2017 at Moorfields Eye Hospital (a large, United Kingdom single center) undergoing anti-VEGF therapy. METHODS: Automatic segmentation was carried out by applying a deep learning segmentation algorithm to 137 379 OCT scans from 6467 eyes of 3261 patients with neovascular AMD. After applying selection criteria, 926 eyes of 926 patients were analyzed. MAIN OUTCOME MEASURES: Correlation coefficients (R2 values) and mean absolute error (MAE) between quantitative OCT (qOCT) parameters and cross-sectional visual function, as well as the predictive value of these parameters for short-term visual change, that is, incremental visual acuity (VA) resulting from an individual injection, as well as VA at distant time points (up to 12 months after baseline). RESULTS: Visual acuity at distant time points could be predicted: R2 = 0.80 (MAE, 5.0 Early Treatment Diabetic Retinopathy Study [ETDRS] letters) and R2 = 0.7 (MAE, 7.2 ETDRS letters) after injection at 3 and at 12 months after baseline (P < 0.001 for both), respectively. Best performing models included both baseline qOCT parameters and treatment response. Furthermore, we present proof-of-principle evidence that the incremental change in VA from an injection can be predicted: R2 = 0.14 (MAE, 5.6 ETDRS letters) for injection 2 and R2 = 0.11 (MAE, 5.0 ETDRS letters) for injection 3 (P < 0.001 for both). CONCLUSIONS: Automatic segmentation enables rapid acquisition of quantitative and reproducible OCT biomarkers with potential to inform treatment decisions in the care of neovascular AMD. This furthers development of point-of-care decision-aid systems for personalized medicine.

Optical coherence tomography in the 2020s-outside the eye clinic.

Optical coherence tomography (OCT) is a paragon of success in the translation of biophotonics science to clinical practice. OCT systems have become ubiquitous in eye clinics but access beyond this is limited by their cost, size and the skill required to operate the devices. Remarkable progress has been made in the development of OCT technology to improve the speed of acquisition, the quality of images and into functional extensions of OCT such as OCT angiography. However, more needs to be done to radically improve the access to OCT by addressing its limitations and enable penetration outside of typical clinical settings and into underserved populations. Beyond high-income countries, there are 6.5 billion people with similar eye-care needs, which cannot be met by the current generation of bulky, expensive and complex OCT systems. In addition, advancing the portability of this technology to address opportunities in point-of-care diagnostics, telemedicine and remote monitoring may aid development of personalised medicine. In this review, we discuss the major milestones in OCT hardware development to reach those beyond the eye clinic.

Ten-year survival trends of neovascular age-related macular degeneration at first presentation.

BACKGROUND: To describe 10-year trends in visual outcomes, anatomical outcomes and treatment burden of patients receiving antivascular endothelial growth factor (anti-VEGF) therapy for neovascular age-related macular degeneration (nAMD). METHODS: Retrospective cohort study of treatment-naïve, first-affected eyes with nAMD started on ranibizumab before January 1, 2009. The primary outcome was time to best-corrected visual acuity (BCVA) falling ≤35 ETDRS letters after initiating anti-VEGF therapy. Secondary outcomes included time to BCVA reaching ≥70 letters, proportion of eyes with BCVA ≥70 and ≤35 letters in 10 years, mean trend of BCVA and central retinal thickness over 10 years, and mean number of injections. RESULTS: For our cohort of 103 patients, Kaplan-Meier analyses demonstrated median time to BCVA reaching ≤35 and ≥70 letters were 37.8 (95% CI 22.2 to 65.1) and 8.3 (95% CI 4.8 to 20.9) months after commencing anti-VEGF therapy, respectively. At the final follow-up, BCVA was ≤35 letters and ≥70 letters in 41.1% and 21%, respectively, in first-affected eyes, while this was the case for 5.4% and 48.2%, respectively, in a patient's better-seeing eye. Mean injection number was 37.0±24.2 per eye and 53.6±30.1 at patient level (63.1% of patients required injections in both eyes). CONCLUSIONS: The chronicity of nAMD disease and its management highlights the importance of long-term visual prognosis. Our analyses suggest that one in five patients will retain good vision (BCVA ≥70 ETDRS letters) in the first-affected eye at 10 years after starting anti-VEGF treatment; yet, one in two patients will have good vision in their better-seeing eye. Moreover, our data suggest that early treatment of nAMD is associated with better visual outcomes.

Quantitative Analysis of OCT for Neovascular Age-Related Macular Degeneration Using Deep Learning.

PURPOSE: To apply a deep learning algorithm for automated, objective, and comprehensive quantification of OCT scans to a large real-world dataset of eyes with neovascular age-related macular degeneration (AMD) and make the raw segmentation output data openly available for further research. DESIGN: Retrospective analysis of OCT images from the Moorfields Eye Hospital AMD Database. PARTICIPANTS: A total of 2473 first-treated eyes and 493 second-treated eyes that commenced therapy for neovascular AMD between June 2012 and June 2017. METHODS: A deep learning algorithm was used to segment all baseline OCT scans. Volumes were calculated for segmented features such as neurosensory retina (NSR), drusen, intraretinal fluid (IRF), subretinal fluid (SRF), subretinal hyperreflective material (SHRM), retinal pigment epithelium (RPE), hyperreflective foci (HRF), fibrovascular pigment epithelium detachment (fvPED), and serous PED (sPED). Analyses included comparisons between first- and second-treated eyes by visual acuity (VA) and race/ethnicity and correlations between volumes. MAIN OUTCOME MEASURES: Volumes of segmented features (mm3) and central subfield thickness (CST) (µm). RESULTS: In first-treated eyes, the majority had both IRF and SRF (54.7%). First-treated eyes had greater volumes for all segmented tissues, with the exception of drusen, which was greater in second-treated eyes. In first-treated eyes, older age was associated with lower volumes for RPE, SRF, NSR, and sPED; in second-treated eyes, older age was associated with lower volumes of NSR, RPE, sPED, fvPED, and SRF. Eyes from Black individuals had higher SRF, RPE, and serous PED volumes compared with other ethnic groups. Greater volumes of the majority of features were associated with worse VA. CONCLUSIONS: We report the results of large-scale automated quantification of a novel range of baseline features in neovascular AMD. Major differences between first- and second-treated eyes, with increasing age, and between ethnicities are highlighted. In the coming years, enhanced, automated OCT segmentation may assist personalization of real-world care and the detection of novel structure-function correlations. These data will be made publicly available for replication and future investigation by the AMD research community.

The Utility of Corneal Nerve Fractal Dimension Analysis in Peripheral Neuropathies of Different Etiology.

Purpose: Quantification of corneal confocal microscopy (CCM) images has shown a significant reduction in corneal nerve fiber length (CNFL) in a range of peripheral neuropathies. We assessed whether corneal nerve fractal dimension (CNFrD) analysis, a novel metric to quantify the topological complexity of corneal subbasal nerves, can differentiate peripheral neuropathies of different etiology. Methods: Ninety patients with peripheral neuropathy, including 29 with diabetic peripheral neuropathy (DPN), 34 with chronic inflammatory demyelinating polyneuropathy (CIDP), 13 with chemotherapy-induced peripheral neuropathy (CIPN), 14 with human immunodeficiency virus-associated sensory neuropathy (HIV-SN), and 20 healthy controls (HCs), underwent CCM for estimation of corneal nerve fiber density (CNFD), CNFL, corneal nerve branch density (CNBD), CNFrD, and CNFrD adjusted for CNFL (ACNFrD). Results: In patients with DPN, CIDP, CIPN, or HIV-SN compared to HCs, CNFD (P = 0.004-0.0001) and CNFL (P = 0.05-0.0001) were significantly lower, with a further significant reduction among subgroups. CNFrD was significantly lower in patients with CIDP compared to HCs and patients with HIV-SN (P = 0.02-0.0009) and in patients with DPN compared to HCs and patients with HIV-SN, CIPN, or CIDP (P = 0.001-0.0001). ACNFrD was lower in patients with CIPN, CIDP, or DPN compared to HCs (P = 0.03-0.0001) and in patients with DPN compared to those with HIV-SN, CIPN, or CIDP (P = 0.01-0.005). Conclusions: CNFrD can detect a distinct pattern of corneal nerve loss in patients with DPN or CIDP compared to those with CIPN or HIV-SN and controls. Translational Relevance: Various peripheral neuropathies are characterized by a comparable degree of corneal nerve loss. Assessment of corneal nerve topology by CNFrD could be useful in differentiating neuropathies based on the pattern of loss.

Predicting conversion to wet age-related macular degeneration using deep learning.

Progression to exudative 'wet' age-related macular degeneration (exAMD) is a major cause of visual deterioration. In patients diagnosed with exAMD in one eye, we introduce an artificial intelligence (AI) system to predict progression to exAMD in the second eye. By combining models based on three-dimensional (3D) optical coherence tomography images and corresponding automatic tissue maps, our system predicts conversion to exAMD within a clinically actionable 6-month time window, achieving a per-volumetric-scan sensitivity of 80% at 55% specificity, and 34% sensitivity at 90% specificity. This level of performance corresponds to true positives in 78% and 41% of individual eyes, and false positives in 56% and 17% of individual eyes at the high sensitivity and high specificity points, respectively. Moreover, we show that automatic tissue segmentation can identify anatomical changes before conversion and high-risk subgroups. This AI system overcomes substantial interobserver variability in expert predictions, performing better than five out of six experts, and demonstrates the potential of using AI to predict disease progression.

In Vivo Confocal Microscopy Morphologic Features and Cyst Density in Acanthamoeba Keratitis.

PURPOSE: To correlate in vivo confocal microscopy morphologic features (IVCM-MF) and Acanthamoeba cyst density (ACD) with final best-corrected visual acuity (BCVA) in Acanthamoeba keratitis (AK). DESIGN: Retrospective cohort study. METHODS: Patient demographics, treatment outcome, and corresponding IVCM-MF performed at the acute stage of infection were analyzed. Inclusion criteria were microbiological positive AK cases seen at Moorfields Eye Hospital between February 2013 and October 2017. Statistical significance was assessed by multinomial regression and multiple linear regression analysis. Main outcome measure was final BCVA. RESULTS: A total of 157 eyes (157 patients) had AK. Absence of single-file round/ovoid objects was associated with a BCVA of 6/36 to 6/9 (odds ratio [OR] 8.13; 95% confidence interval [CI], 1.55-42.56, P = .013) and ≥6/6 (OR 10.50; 95% CI, 2.12-51.92, P = .004) when compared to no perception of light to 6/60. Absence of rod/spindle objects was associated with a BCVA of ≥6/6 (OR 4.55; 95% CI, 1.01-20.45, P = .048). Deep stromal/ring infiltrate was associated with single-file round/ovoid objects (OR 7.78; 95% CI, 2.69-22.35, P < .001), rod/spindle objects (OR 7.05; 95% CI, 2.11-23.59, P = .002), and binary round/ovoid objects (OR 3.45; 95% CI, 1.17-10.14, P = .024). There was a positive association between ACD and treatment duration (ß = 0.14, P = .049), number of IVCM-MF (ß = 0.34, P = .021), and clusters of round/ovoid objects (ß = 0.29, P = .002). CONCLUSIONS: Specific IVCM-MF correlate with ACD and clinical staging of disease, and are prognostic indicators for a poorer visual outcome.

Automated Pupillometry Using a Prototype Binocular Optical Coherence Tomography System.

PURPOSE: To determine the test-retest reliability and diagnostic accuracy of a binocular optical coherence tomography (OCT) prototype (Envision Diagnostics, El Segundo, California, USA) for pupillometry. DESIGN: Assessment of diagnostic reliability and accuracy. METHODS: Fifty participants with relative afferent pupillary defects (RAPDs) confirmed using the swinging flashlight method (mean age 49.6 years) and 50 healthy control subjects (mean age 31.3 years) were examined. Participants twice underwent an automated pupillometry examination using a binocular OCT system that presents a stimulus and simultaneously captures OCT images of the iris-pupil plane of both eyes. Participants underwent a single examination on the RAPDx (Konan Medical, Irvine, California, USA), an automated infrared pupillometer. Pupil parameters including maximum and minimum diameter, and anisocoria were measured. The magnitude of RAPD was calculated using the log of the ratio of the constriction amplitude between the eyes. A pathological RAPD was above ±0.5 log units on both devices. RESULTS: The intraclass correlation coefficient was >0.90 for OCT-derived maximum pupil diameter, minimum pupil diameter, and anisocoria. The RAPDx had a sensitivity of 82% and a specificity of 94% for detection of RAPD whereas the binocular OCT had a sensitivity of 74% and specificity of 86%. The diagnostic accuracy of the RAPDx and binocular OCT was 88% (95% confidence interval 80%-94%) and 80% (95% confidence interval 71%-87%) respectively. CONCLUSIONS: Binocular OCT-derived pupil parameters had excellent test-retest reliability. The diagnostic accuracy of RAPD was inferior to the RAPDx and is likely related to factors such as eye movement during OCT capture. As OCT becomes ubiquitous, OCT-derived measurements may provide an efficient method of objectively quantifying the pupil responses.

Moorfields AMD database report 2: fellow eye involvement with neovascular age-related macular degeneration.

BACKGROUND/AIMS: Neovascular age-related macular degeneration (nAMD) is frequently bilateral, and previous reports on 'fellow eyes' have assumed sequential treatment after a period of treatment of the first eye only. The aim of our study was to analyse baseline characteristics and visual acuity (VA) outcomes of fellow eye involvement with nAMD, specifically differentiating between sequential and non-sequential (due to macular scarring in the first eye) antivascular endothelial growth factor treatment and timelines for fellow eye involvement. METHODS: Retrospective, electronic medical record database study of the Moorfields AMD database of 6265 patients/120 286 single entries with data extracted between 21 October 2008 and 9 August 2018. The data set for analysis consisted of 1180 sequential, 807 non-sequential and 3410 unilateral eyes. RESULTS: Mean VA (ETDRS letters±SD) of sequentially treated fellow eyes at baseline was significantly higher (63±13), VA gain over 2 years lower (0.37±14) and proportion of eyes with good VA (≥70 letters) higher (46%) than the respective first eyes (baseline VA 54±16, VA gain at 2 years 5.6±15, percentage of eyes with good VA 39%). Non-sequential fellow eyes showed baseline characteristics and VA outcomes similar to first eyes. Fellow eye involvement rate was 32% at 2 years, and median time interval to fellow eye involvement was 71 (IQR: 27-147) weeks. CONCLUSION: This report shows that sequentially treated nAMD fellow eyes have better baseline and final VA than non-sequentially treated eyes after 2 years of treatment. Sequentially treated eyes also had a greater proportion with good VA after 2 years.

One- and two-year visual outcomes from the Moorfields age-related macular degeneration database: a retrospective cohort study and an open science resource.

OBJECTIVES: To analyse treatment outcomes and share clinical data from a large, single-centre, well-curated database (8174 eyes/6664 patients with 120 756 single entries) of patients with neovascular age-related macular degeneration (AMD) treated with anti-vascular endothelial growth factor (VEGF). By making our depersonalised raw data openly available, we aim to stimulate further research in AMD, as well as set a precedent for future work in this area. SETTING: Retrospective, comparative, non-randomised electronic medical record (EMR) database cohort study of the UK Moorfields AMD database with data extracted between 2008 and 2018. PARTICIPANTS: Including one eye per patient, 3357 eyes/patients (61% female). Extraction criteria were ≥1 ranibizumab or aflibercept injection, entry of 'AMD' in the diagnosis field of the EMR and a minimum of 1 year of follow-up. Exclusion criteria were unknown date of first injection and treatment outside of routine clinical care at Moorfields before the first recorded injection in the database. MAIN OUTCOME MEASURES: Primary outcome measure was change in VA at 1 and 2 years from baseline as measured in Early Treatment Diabetic Retinopathy Study letters. Secondary outcomes were the number of injections and predictive factors for VA gain. RESULTS: Mean VA gain at 1 year and 2 years were +5.5 (95% CI 5.0 to 6.0) and +4.9 (95% CI 4.2 to 5.6) letters, respectively. Fifty-four per cent of eyes gained ≥5 letters at 2 years, 63% had stable VA (±≤14 letters), 44% of eyes maintained good VA (≥70 letters). Patients received a mean of 7.7 (95% CI 7.6 to 7.8) injections during year 1 and 13.0 (95% CI 12.8 to 13.2) injections over 2 years. Younger age, lower baseline VA and more injections were associated with higher VA gain at 2 years. CONCLUSION: This study benchmarks high quality EMR study results of real life AMD treatment and promotes open science in clinical AMD research by making the underlying data publicly available.

Automated deep learning design for medical image classification by health-care professionals with no coding experience: a feasibility study.

BACKGROUND: Deep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of automated deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding-and no deep learning-expertise. METHODS: We used five publicly available open-source datasets: retinal fundus images (MESSIDOR); optical coherence tomography (OCT) images (Guangzhou Medical University and Shiley Eye Institute, version 3); images of skin lesions (Human Against Machine [HAM] 10000), and both paediatric and adult chest x-ray (CXR) images (Guangzhou Medical University and Shiley Eye Institute, version 3 and the National Institute of Health [NIH] dataset, respectively) to separately feed into a neural architecture search framework, hosted through Google Cloud AutoML, that automatically developed a deep learning architecture to classify common diseases. Sensitivity (recall), specificity, and positive predictive value (precision) were used to evaluate the diagnostic properties of the models. The discriminative performance was assessed using the area under the precision recall curve (AUPRC). In the case of the deep learning model developed on a subset of the HAM10000 dataset, we did external validation using the Edinburgh Dermofit Library dataset. FINDINGS: Diagnostic properties and discriminative performance from internal validations were high in the binary classification tasks (sensitivity 73·3-97·0%; specificity 67-100%; AUPRC 0·87-1·00). In the multiple classification tasks, the diagnostic properties ranged from 38% to 100% for sensitivity and from 67% to 100% for specificity. The discriminative performance in terms of AUPRC ranged from 0·57 to 1·00 in the five automated deep learning models. In an external validation using the Edinburgh Dermofit Library dataset, the automated deep learning model showed an AUPRC of 0·47, with a sensitivity of 49% and a positive predictive value of 52%. INTERPRETATION: All models, except the automated deep learning model trained on the multilabel classification task of the NIH CXR14 dataset, showed comparable discriminative performance and diagnostic properties to state-of-the-art performing deep learning algorithms. The performance in the external validation study was low. The quality of the open-access datasets (including insufficient information about patient flow and demographics) and the absence of measurement for precision, such as confidence intervals, constituted the major limitations of this study. The availability of automated deep learning platforms provide an opportunity for the medical community to enhance their understanding in model development and evaluation. Although the derivation of classification models without requiring a deep understanding of the mathematical, statistical, and programming principles is attractive, comparable performance to expertly designed models is limited to more elementary classification tasks. Furthermore, care should be placed in adhering to ethical principles when using these automated models to avoid discrimination and causing harm. Future studies should compare several application programming interfaces on thoroughly curated datasets. FUNDING: National Institute for Health Research and Moorfields Eye Charity.

Clinically applicable deep learning for diagnosis and referral in retinal disease.

The volume and complexity of diagnostic imaging is increasing at a pace faster than the availability of human expertise to interpret it. Artificial intelligence has shown great promise in classifying two-dimensional photographs of some common diseases and typically relies on databases of millions of annotated images. Until now, the challenge of reaching the performance of expert clinicians in a real-world clinical pathway with three-dimensional diagnostic scans has remained unsolved. Here, we apply a novel deep learning architecture to a clinically heterogeneous set of three-dimensional optical coherence tomography scans from patients referred to a major eye hospital. We demonstrate performance in making a referral recommendation that reaches or exceeds that of experts on a range of sight-threatening retinal diseases after training on only 14,884 scans. Moreover, we demonstrate that the tissue segmentations produced by our architecture act as a device-independent representation; referral accuracy is maintained when using tissue segmentations from a different type of device. Our work removes previous barriers to wider clinical use without prohibitive training data requirements across multiple pathologies in a real-world setting.