External dacryocystorhinostomy: A critical overview of the current evidence.

PURPOSE: External dacryocystorhinostomy (ext-DCR) is still considered the gold standard in the treatment of distal lacrimal duct obstruction. Despite an overall high success rate, some patients do relapse after surgery and the ideal perioperative management has not been identified yet. The purpose of this study is to critically discuss the published evidence in the last five years on the medical and surgical management of external dacryocystorhinostomy. Furthermore, the drawbacks of the available literature and the perspectives in this field will be also highlighted. METHODS: Systematic review of the literature in the last 5 years by using PubMed and Google Scholar. RESULTS: A total of 64 articles were included. Many technical modifications and surgical refinements have been proposed but a formal comparison among the various techniques is hampered by methodological heterogeneity. The use of local anesthesia, and the perioperative adjunctive techniques to reduce the risk of restenosis are also gaining popularity but the level of evidence remains weak. CONCLUSION: Ext-DCR offers satisfactory clinical outcomes even though there are many gray areas that need to be addressed in future high-quality studies.

Submacular hemorrhage during neovascular age-related macular degeneration: a meta-analysis and meta-regression on the use of tPA and anti-VEGFs.

BACKGROUND: Submacular hemorrhage (SMH) associated with neovascular age-related macular degeneration (nAMD) precipitates rapid visual decline and impacts quality of life. Treatments vary, but combined recombinant tissue plasminogen activator (tPA) and anti-vascular endothelial growth factor (anti-VEGF) therapy has gained prominence as a viable treatment option. OBJECTIVES: This study aims to evaluate the efficacy of the combination of tPA and anti-VEGF. METHODS: We conducted a systematic review meta-analysis following PRISMA guidelines, focusing on studies examining tPA and anti-VEGF therapy in SMH secondary to nAMD. Outcomes measured were change in best-corrected visual acuity (BCVA) and success rate of SMH displacement. Meta-regression assessed the relative efficacy of intravitreal and subretinal delivery. RESULTS: Out of 257 initial reports, 22 studies involving 29 patient populations met inclusion criteria. Our analysis showed significant improvement in BCVA and a high rate of successful SMH displacement with combined tPA and anti-VEGF therapy. No significant differences were found between subretinal and intravitreal tPA administration. Furthermore, when evaluating the effects of subretinal versus intravitreal anti-VEGF administration in patients treated with subretinal tPA, the results indicated similar efficacy. CONCLUSIONS: Combined tPA and anti-VEGF therapy is effective in managing SMH in nAMD patients, significantly improving visual acuity and SMH displacement. The location of tPA and anti-VEGF delivery did not significantly impact outcomes.

Intravitreal aflibercept 8 mg in neovascular age-related macular degeneration (PULSAR): 48-week results from a randomised, double-masked, non-inferiority, phase 3 trial.

BACKGROUND: Intravitreal aflibercept 8 mg could improve treatment outcomes and provide sustained disease control in patients with neovascular age-related macular degeneration (nAMD), with extended dosing compared with aflibercept 2 mg. METHODS: PULSAR is a phase 3, randomised, three-group, double-masked, non-inferiority, 96-week trial conducted across 223 sites worldwide. Adults with nAMD were randomised 1:1:1 to aflibercept 8 mg every 12 weeks (8q12), aflibercept 8 mg every 16 weeks (8q16), or aflibercept 2 mg every 8 weeks (2q8), following three initial monthly doses in all groups. From week 16, patients in the aflibercept 8 mg groups had their dosing interval shortened if pre-specified dose regimen modification criteria denoting disease activity were met. The primary endpoint was change from baseline in best-corrected visual acuity (BCVA) at week 48. All patients with at least one dose of study treatment were included in the efficacy and safety analyses. This trial is registered with ClinicalTrials.gov (NCT04423718) and is ongoing. FINDINGS: Of 1011 patients randomised to aflibercept 8q12 (n=336), 8q16 (n=338), or 2q8 (n=337) between Aug 11, 2020, and July 30, 2021, 1009 patients received study treatment (aflibercept 8q12 n=335; aflibercept 8q16 n=338; and aflibercept 2q8 n=336). Aflibercept 8q12 and 8q16 showed non-inferior BCVA gains versus aflibercept 2q8 (mean BCVA change from baseline +6·7 [SD 12·6] and +6·2 [11·7] vs +7·6 [12·2] letters). The least squares mean differences between aflibercept 8q12 versus 2q8 and 8q16 versus 2q8, respectively, were -0·97 (95% CI -2·87 to 0·92) and -1·14 (-2·97 to 0·69) letters (non-inferiority margin at 4 letters). The incidence of ocular adverse events in the study eye was similar across groups (aflibercept 8q12 n=129 [39%]; aflibercept 8q16 n=127 [38%]; and aflibercept 2q8 n=130 [39%]). INTERPRETATION: Aflibercept 8 mg showed efficacy and safety with extended dosing intervals, which has the potential to improve the management of patients with nAMD. FUNDING: Bayer AG and Regeneron Pharmaceuticals.

Behind the mask: a critical perspective on the ethical, moral, and legal implications of AI in ophthalmology.

PURPOSE: This narrative review aims to provide an overview of the dangers, controversial aspects, and implications of artificial intelligence (AI) use in ophthalmology and other medical-related fields. METHODS: We conducted a decade-long comprehensive search (January 2013-May 2023) of both academic and grey literature, focusing on the application of AI in ophthalmology and healthcare. This search included key web-based academic databases, non-traditional sources, and targeted searches of specific organizations and institutions. We reviewed and selected documents for relevance to AI, healthcare, ethics, and guidelines, aiming for a critical analysis of ethical, moral, and legal implications of AI in healthcare. RESULTS: Six main issues were identified, analyzed, and discussed. These include bias and clinical safety, cybersecurity, health data and AI algorithm ownership, the "black-box" problem, medical liability, and the risk of widening inequality in healthcare. CONCLUSION: Solutions to address these issues include collecting high-quality data of the target population, incorporating stronger security measures, using explainable AI algorithms and ensemble methods, and making AI-based solutions accessible to everyone. With careful oversight and regulation, AI-based systems can be used to supplement physician decision-making and improve patient care and outcomes.

A mini-invasive surgical technique for Carlevale IOL implantation: case series study and description of concomitant surgery.

PURPOSE: To examine the feasibility and outcomes of a modified technique for the implantation of scleral fixated Carlevale intraocular lens (IOL) (I71 FIL SSF. Soleko IOL Division, Pontecorvo, Italy), and to analyze the occurrence of adverse events. METHODS: This is a retrospective observational study conducted revising patients charts from 2018 to 2023. Thirty-five eyes of 33 patients were included. Patients requiring IOL explantation had either IOL dislocation or opacification. The implantation of the Carlevale IOL was performed with the subconjunctival positioning of the anchors without any scleral flap. All maneuvers were performed transconjunctivally. The anatomical outcomes considered were IOL positioning, and the absence of postoperative complications. The functional outcomes analyzed were best correctedvisual acuity (BCVA) and refraction. RESULTS: In all the cases, the IOL was well positioned and centered postoperatively. No cases of conjunctival erosion were recorded. The best corrected visual acuity (BCVA) was 0.9±0.6 logMar (mean±standard deviation) preoperatively and 0.5±0.5 logMar (mean±standard deviation) postoperatively. The mean preoperative spherical equivalent was +6.8±7.7 dioptres, while postoperatively it was -1.1±1.6 dioptres. The most frequent procedure associated to secondary IOL implantation was posterior vitrectomy (25 eyes, 71.4%), which was performed with 25-gauge transconjunctival cannulas in the ciliary sulcus. The follow-up period was 24.5±16.9 months (mean±standard deviation). CONCLUSION: The described mini-invasive technique for Carlevale IOL implantation is safe and effective. It can be recommended either as a stand-alone operation or associated to concurrent surgical procedures.

Towards a Long-Read Sequencing Approach for the Molecular Diagnosis of RPGRORF15 Genetic Variants.

Sequencing of the low-complexity ORF15 exon of RPGR, a gene correlated with retinitis pigmentosa and cone dystrophy, is difficult to achieve with NGS and Sanger sequencing. False results could lead to the inaccurate annotation of genetic variants in dbSNP and ClinVar databases, tools on which HGMD and Ensembl rely, finally resulting in incorrect genetic variants interpretation. This paper aims to propose PacBio sequencing as a feasible method to correctly detect genetic variants in low-complexity regions, such as the ORF15 exon of RPGR, and interpret their pathogenicity by structural studies. Biological samples from 75 patients affected by retinitis pigmentosa or cone dystrophy were analyzed with NGS and repeated with PacBio. The results showed that NGS has a low coverage of the ORF15 region, while PacBio was able to sequence the region of interest and detect eight genetic variants, of which four are likely pathogenic. Furthermore, molecular modeling and dynamics of the RPGR Glu-Gly repeats binding to TTLL5 allowed for the structural evaluation of the variants, providing a way to predict their pathogenicity. Therefore, we propose PacBio sequencing as a standard procedure in diagnostic research for sequencing low-complexity regions such as RPGRORF15, aiding in the correct annotation of genetic variants in online databases.

Explainable artificial intelligence model for the detection of geographic atrophy using colour retinal photographs.

OBJECTIVE: To develop and validate an explainable artificial intelligence (AI) model for detecting geographic atrophy (GA) via colour retinal photographs. METHODS AND ANALYSIS: We conducted a prospective study where colour fundus images were collected from healthy individuals and patients with retinal diseases using an automated imaging system. All images were categorised into three classes: healthy, GA and other retinal diseases, by two experienced retinologists. Simultaneously, an explainable learning model using class activation mapping techniques categorised each image into one of the three classes. The AI system's performance was then compared with manual evaluations. RESULTS: A total of 540 colour retinal photographs were collected. Data was divided such that 300 images from each class trained the AI model, 120 for validation and 120 for performance testing. In distinguishing between GA and healthy eyes, the model demonstrated a sensitivity of 100%, specificity of 97.5% and an overall diagnostic accuracy of 98.4%. Performance metrics like area under the receiver operating characteristic (AUC-ROC, 0.988) and the precision-recall (AUC-PR, 0.952) curves reinforced the model's robust achievement. When differentiating GA from other retinal conditions, the model preserved a diagnostic accuracy of 96.8%, a precision of 90.9% and a recall of 100%, leading to an F1-score of 0.952. The AUC-ROC and AUC-PR scores were 0.975 and 0.909, respectively. CONCLUSIONS: Our explainable AI model exhibits excellent performance in detecting GA using colour retinal images. With its high sensitivity, specificity and overall diagnostic accuracy, the AI model stands as a powerful tool for the automated diagnosis of GA.

Update on coronavirus disease 2019: Ophthalmic Manifestations and Adverse Reactions to Vaccination.

The coronavirus disease 2019 (COVID-19) pandemic caused by the severe acute respiratory syndrome coronavirus 2 was one of the most devastating public health issues in recent decades. The ophthalmology community is as concerned about the COVID-19 pandemic as the global public health community is, as COVID-19 was recognized to affect multiple organs in the human body, including the eyes, early in the course of the outbreak. Ophthalmic manifestations of COVID-19 are highly variable and could range from mild ocular surface abnormalities to potentially sight and life-threatening orbital and neuro-ophthalmic diseases. Furthermore, ophthalmic manifestations may also be the presenting or the only findings in COVID-19 infections. Meanwhile, global vaccination campaigns to attain herd immunity in different populations are the major strategy to mitigate the pandemic. As novel vaccinations against COVID-19 emerged, so were reports on adverse ophthalmic reactions potentially related to such. As the world enters a post-pandemic state where COVID-19 continues to exist and evolve as an endemic globally, the ophthalmology community ought to be aware of and keep abreast of the latest knowledge of ophthalmic associations with COVID-19 and its vaccinations. This review is a summary of the latest literature on the ophthalmic manifestations of COVID-19 and the adverse ophthalmic reactions related to its vaccinations.

Real-world practice patterns of eplerenone use for central serous chorioretinopathy.

PURPOSE: To report eplerenone use by retina specialists worldwide for central serous chorioretinopathy (CSCR). METHODS: A self-reporting questionnaire was distributed to retina specialists worldwide to gather clinicians' perspectives on CSCR cases treated, eplerenone dosage and duration, reasons to use it, and side effects. RESULTS: The survey included 241 retina specialists (122 Indian and 119 international) with an average experience of 15.69 ± 9.59 years. Oral eplerenone was used to treat CSCR by 149 (62%) participants. Only 6% (n = 9) had easy access to verteporfin dye and photodynamic therapy. 30 (20%) of the 149 respondents changed their treatment with eplerenone after VICI trial results. Eplerenone was prescribed mostly for chronic CSCR (n = 86, 58%), regardless of involvement laterality. 62% (n = 92) had fewer than 25% CSCR cases treated with eplerenone. 85 (57%) respondents used eplerenone only when other treatments failed, while 36 (24%) used it as first-line treatment. 73 (49%) respondents, prescribed eplerenone at a 50 mg daily dose and 137 (92%) retina specialists used eplerenone for 0-3 months. The drug's efficacy dissatisfied 21 (14%) study participants. 124 (83%) study participants did not encounter any ocular or systemic side effects with eplerenone use. Eplerenone related kidney and electrolyte issues were noted by 11 (7%) study participants. CONCLUSION: The treatment of CSCR varies around the world and is primarily influenced by the photodynamic therapy availability and the findings of VICI trial. Despite the limited benefit of eplerenone reported by the VICI trial, it is still used as evidenced by real-world experience. TRIAL REGISTRATION NUMBER: Not applicable.

The Present and Future of Optic Pathway Glioma Therapy.

Optic pathway gliomas (OPGs) encompass two distinct categories: benign pediatric gliomas, which are characterized by favorable prognosis, and malignant adult gliomas, which are aggressive cancers associated with a poor outcome. Our review aims to explore the established standards of care for both types of tumors, highlight the emerging therapeutic strategies for OPG treatment, and propose potential alternative therapies that, while originally studied in a broader glioma context, may hold promise for OPGs pending further investigation. These potential therapies encompass immunotherapy approaches, molecular-targeted therapy, modulation of the tumor microenvironment, nanotechnologies, magnetic hyperthermia therapy, cyberKnife, cannabinoids, and the ketogenic diet. Restoring visual function is a significant challenge in cases where optic nerve damage has occurred due to the tumor or its therapeutic interventions. Numerous approaches, particularly those involving stem cells, are currently being investigated as potential facilitators of visual recovery in these patients.

Ocular complications of diabetes mellitus in a pediatric population and proposals for screening and follow-up programs.

Background: Diabetes mellitus (DM) is one of the world's greatest health emergencies of the 21st century. Ocular complications of DM are commonly chronic and progressive, but vision loss can be effectively prevented or delayed with early detection and timely treatment. Therefore, regular comprehensive ophthalmologic examinations are mandatory. Ophthalmic screening and dedicated follow-up for adults with DM are well established, whereas, there is no consensus on optimal recommendations for the pediatric population, reflecting the lack of clarity about the current burden of disease in this age group. Objectives: To determine the epidemiology of ocular complications of diabetes and to assess optical coherence tomography (OCT) and optical coherence tomography angiography (OCTA) macular features in a pediatric population with DM. To review ophthalmological screening and follow-up plans for the diabetic pediatric population. Design: Observational study. Methods: Retrospective consecutive cohort study of all 165 diabetic patients (330 eyes) aged 0-18 years, examined between January 2006 and September 2018 at the Pediatric Department of 'S. Maria della Misericordia' Udine Hospital who underwent at least one complete ophthalmologic examination at the Ophthalmology University Clinic at the Udine Hospital. OCT and OCTA data were available for 37 patients (72 eyes, 2 excluded). The associations between ocular complications and selected potential risk factors were evaluated by univariate analyses. Results: No patient had signs of ocular diabetic complications or any macular morphological or micro-vascular impairment, regardless of any potential risk factor. The prevalence of strabismus and refractive errors in the study group, was found to be similar to non-diabetic pediatric populations. Conclusion: Screening and follow-up of ocular diabetic complications in children and adolescents could be performed less frequently than in adults with diabetes. There is no need to screen potentially treatable visual disorders in diabetic children earlier or more frequently than in the healthy children thus reducing time spent in hospital and permitting a better tolerance to medical examinations in diabetic pediatric patients. We described the OCT and OCTA patterns in a pediatric population with DM.

The impact of artificial intelligence on retinal disease management: Vision Academy retinal expert consensus.

PURPOSE OF REVIEW: The aim of this review is to define the "state-of-the-art" in artificial intelligence (AI)-enabled devices that support the management of retinal conditions and to provide Vision Academy recommendations on the topic. RECENT FINDINGS: Most of the AI models described in the literature have not been approved for disease management purposes by regulatory authorities. These new technologies are promising as they may be able to provide personalized treatments as well as a personalized risk score for various retinal diseases. However, several issues still need to be addressed, such as the lack of a common regulatory pathway and a lack of clarity regarding the applicability of AI-enabled medical devices in different populations. SUMMARY: It is likely that current clinical practice will need to change following the application of AI-enabled medical devices. These devices are likely to have an impact on the management of retinal disease. However, a consensus needs to be reached to ensure they are safe and effective for the overall population.

Current status and practical considerations of artificial intelligence use in screening and diagnosing retinal diseases: Vision Academy retinal expert consensus.

PURPOSE OF REVIEW: The application of artificial intelligence (AI) technologies in screening and diagnosing retinal diseases may play an important role in telemedicine and has potential to shape modern healthcare ecosystems, including within ophthalmology. RECENT FINDINGS: In this article, we examine the latest publications relevant to AI in retinal disease and discuss the currently available algorithms. We summarize four key requirements underlining the successful application of AI algorithms in real-world practice: processing massive data; practicability of an AI model in ophthalmology; policy compliance and the regulatory environment; and balancing profit and cost when developing and maintaining AI models. SUMMARY: The Vision Academy recognizes the advantages and disadvantages of AI-based technologies and gives insightful recommendations for future directions.

Pharmacokinetic and Pharmacodynamic Rationale for Extending VEGF Inhibition Increasing Intravitreal Aflibercept Dose.

BACKGROUND: The effects of various dosages and treatment regimens on intravitreal aflibercept concentrations and the proportion of free vascular endothelial growth factor (VEGF) to total VEGF were evaluated using a drug and disease assessment model. The 8 mg dosage received specific attention. METHODS: A time-dependent mathematical model was developed and implemented using Wolfram Mathematica software v12.0. This model was used to obtain drug concentrations after multiple doses of different aflibercept dosages (0.5 mg, 2 mg, and 8 mg) and to estimate the time-dependent intravitreal free VEGF percentage levels. A series of fixed treatment regimens were modeled and evaluated as potential clinical applications. RESULTS: The simulation results indicate that 8 mg aflibercept administered at a range of treatment intervals (between 12 and 15 weeks) would allow for the proportion of free VEGF to remain below threshold levels. Our analysis indicates that these protocols maintain the ratio of free VEGF below 0.001%. CONCLUSIONS: Fixed q12-q15 (every 12-15 weeks) 8 mg aflibercept regimens can produce adequate intravitreal VEGF inhibition.

The ideal intravitreal injection setting: office, ambulatory surgery room or operating theatre? A narrative review and international survey.

PURPOSE: This study reviews evidence and provides recommendations for the ideal setting of intravitreal injection (IVI) administration of vascular endothelial growth factor (VEGF) inhibitors. METHODS: A multi-step approach was employed, including content analysis of regulations and guidelines, a systematic literature review, and an international survey assessing perioperative complications and endophthalmitis incidence in relation to injection settings. The literature review searched PubMed and Cochrane databases from 2006 to 2022, focusing on studies reporting correlations between complications and treatment settings. The survey utilized a web-based questionnaire distributed to clinical sites and the international ophthalmic community, with data managed using electronic capture tools. RESULTS: We reviewed regulations and guidelines from 23 countries across five continents, finding significant variation in IVI administration settings. In most countries, IVI is primarily administered in outpatient clean rooms (96%) or offices (39%), while in others, it is restricted to ambulatory surgery rooms or hospital-based operating theatres (4%). The literature review found that endophthalmitis risk after IVI is generally low (0.01% to 0.26% per procedure), with no significant difference between office-based and operating room settings. The international survey (20 centers, 96,624 anti-VEGF injections) found low overall incidences of severe perioperative systemic adverse events and endophthalmitis, independent of injection settings. CONCLUSION: No significant differences in perioperative complications were observed among various settings, including operating theatres, ambulatory surgery rooms, offices, hospitals, or extra-hospital environments. Choosing the appropriate clinical setting can optimize patient management, potentially increasing effectiveness, quality, productivity, and capacity.

Biosimilars for retinal diseases: United States-Europe awareness survey (Bio-USER - survey).

PURPOSE: To assess the awareness of biosimilar intravitreal anti-VEGF agents among retina specialists practicing in the United States (US) and Europe. METHODS: A 16-question online survey was created in English and distributed between Dec 01, 2021 and Jan 31, 2022. A total of 112 respondents (retinal physicians) from the US and Europe participated. RESULTS: The majority of the physicians (56.3%) were familiar with anti-VEGF biosimilars. A significant number of physicians needed more information (18.75%) and real world data (25%) before switching to a biosimilar. About one half of the physicians were concerned about biosimilar safety (50%), efficacy (58.9 %), immunogenicity (50%), and their efficacy with extrapolated indications (67.8 %). Retinal physicians from the US were less inclined to shift from off-label bevacizumab to biosimilar ranibizumab or on-label bevacizumab (if approved) compared to physicians from Europe (p=0.0001). Furthermore, physicians from the US were more concerned about biosimilar safety (p=0.0371) and efficacy compared to Europe (p= 0.0078). CONCLUSIONS: The Bio-USER survey revealed that while the majority of retinal physicians need additional information regarding the safety, efficacy and immunogenicity when making clinical decisions regarding their use. Retinal physicians from US are more comfortable in continuing to use off-label bevacizumab compared to physicians from Europe.

Vitrectomized vs non-vitrectomized eyes in DEX implant treatment for DMO-Is there any difference? the VITDEX study.

OBJECTIVE: We aimed to compare visual and anatomical outcome in vitrectomized and non-vitrectomized eyes treated with dexamethasone (DEX) implant due to diabetic macular oedema (DMO). DESIGN: Multicenter, retrospective, interventional study. PARTICIPANTS: 236 eyes from 234 patients with DMO with or without previous vitrectomy performed with follow-up of 12 months. METHODS: Records were reviewed for cases of DMO treated with DEX implant in vitrectomized and not vitrectomized eyes. Best corrected visual acuity (BCVA), central subfoveal thickness (CST), and intraocular pressure (IOP) were recorded at baseline and 12 months after treatment with DEX implants. Correlations between vitreous status and visual and anatomical outcome, as well as safety profile were analysed. MAIN OUTCOME MEASURES: BCVA and CST over follow-up period. SECONDARY OUTCOMES: cataract rate formation, intraocular pressure increase, number of implants needed. RESULTS: The non-vitrectomized group included 130 eyes (55.1%), the vitrectomized group included 106 eyes (44.9%). The groups were well balanced for age and gender (p = 0.540, and p = 0.053, respectively). Both groups showed statistically significant improvement in BCVA and CST (for all groups: p < 0.001). There was no significant difference between the groups in terms of change in vision (p = 0.89) and anatomy (p = 0.65). The mean number of DEX implants given during follow-up was 3.5 in both groups, and there was no significant difference between the groups (p = 0.81). CONCLUSION: We demonstrated similar anatomical and functional efficacy of DEX implant in non-vitrectomized and vitrectomized eyes. Its efficacy was not influenced by full vitrectomy for diabetic retinopathy complications. Safety profile was well balanced between groups.

Optical Coherence Tomography Findings in Rhegmatogenous Retinal Detachment: A Systematic Review.

Rhegmatogenous retinal detachment is a sight-threatening condition that may lead to blindness if left untreated. Surgical treatments may vary and are tailored to a single patient. Anatomical and functional results may vary, due to factors that are currently under study. Optical coherence tomography (OCT) allows a detailed visualization of the retinal structure. Some studies have been performed using OCT on eyes with retinal detachment. We performed a review on the subject. Several data have been obtained using different OCT applications. Some alterations may represent potential biomarkers since they are associated with visual and anatomical prognoses. Increased knowledge on the subject may be helpful to choose among different surgical strategies and endotamponades. More research on the topic is needed.

Detecting glaucoma from multi-modal data using probabilistic deep learning.

Objective: To assess the accuracy of probabilistic deep learning models to discriminate normal eyes and eyes with glaucoma from fundus photographs and visual fields. Design: Algorithm development for discriminating normal and glaucoma eyes using data from multicenter, cross-sectional, case-control study. Subjects and participants: Fundus photograph and visual field data from 1,655 eyes of 929 normal and glaucoma subjects to develop and test deep learning models and an independent group of 196 eyes of 98 normal and glaucoma patients to validate deep learning models. Main outcome measures: Accuracy and area under the receiver-operating characteristic curve (AUC). Methods: Fundus photographs and OCT images were carefully examined by clinicians to identify glaucomatous optic neuropathy (GON). When GON was detected by the reader, the finding was further evaluated by another clinician. Three probabilistic deep convolutional neural network (CNN) models were developed using 1,655 fundus photographs, 1,655 visual fields, and 1,655 pairs of fundus photographs and visual fields collected from Compass instruments. Deep learning models were trained and tested using 80% of fundus photographs and visual fields for training set and 20% of the data for testing set. Models were further validated using an independent validation dataset. The performance of the probabilistic deep learning model was compared with that of the corresponding deterministic CNN model. Results: The AUC of the deep learning model in detecting glaucoma from fundus photographs, visual fields, and combined modalities using development dataset were 0.90 (95% confidence interval: 0.89-0.92), 0.89 (0.88-0.91), and 0.94 (0.92-0.96), respectively. The AUC of the deep learning model in detecting glaucoma from fundus photographs, visual fields, and both modalities using the independent validation dataset were 0.94 (0.92-0.95), 0.98 (0.98-0.99), and 0.98 (0.98-0.99), respectively. The AUC of the deep learning model in detecting glaucoma from fundus photographs, visual fields, and both modalities using an early glaucoma subset were 0.90 (0.88,0.91), 0.74 (0.73,0.75), 0.91 (0.89,0.93), respectively. Eyes that were misclassified had significantly higher uncertainty in likelihood of diagnosis compared to eyes that were classified correctly. The uncertainty level of the correctly classified eyes is much lower in the combined model compared to the model based on visual fields only. The AUCs of the deterministic CNN model using fundus images, visual field, and combined modalities based on the development dataset were 0.87 (0.85,0.90), 0.88 (0.84,0.91), and 0.91 (0.89,0.94), and the AUCs based on the independent validation dataset were 0.91 (0.89,0.93), 0.97 (0.95,0.99), and 0.97 (0.96,0.99), respectively, while the AUCs based on an early glaucoma subset were 0.88 (0.86,0.91), 0.75 (0.73,0.77), and 0.92 (0.89,0.95), respectively. Conclusion and relevance: Probabilistic deep learning models can detect glaucoma from multi-modal data with high accuracy. Our findings suggest that models based on combined visual field and fundus photograph modalities detects glaucoma with higher accuracy. While probabilistic and deterministic CNN models provided similar performance, probabilistic models generate certainty level of the outcome thus providing another level of confidence in decision making.