Repeatability of a Virtual Reality Headset Perimeter in Glaucoma and Ocular Hypertensive Patients.

Purpose: The VisuALL S is an automated, static threshold, virtual reality-based perimeter for mobile evaluation of the visual field. We examined same-day and 3-month repeatability. Methods: Adult participants with a diagnosis of glaucoma or ocular hypertension underwent two VisuALL 24-2 Normal T- Full threshold strategy tests at baseline and one additional exam at 3 months for each eligible eye. Spearman, intraclass correlation coefficients (ICCs), and Bland-Altman plots were used to assess the correlation of individual point sensitivities and mean deviation (MD) among three tests. Results: Eighty-eight eyes (44 participants) were included. Average age was 68.1 ± 14.3 years, and 60.7% were male. VisuALL MD was highly correlated between tests (intravisit: r = 0.89, intervisit: r = 0.82; P < 0.001 for both). Bland-Altman analysis showed an average difference in intravisit MD of -0.67 dB (95% confidence interval [CI], -6.04 to 4.71 dB) and -0.15 dB (95% CI, -8.04 to 7.73 dB) for intervisit exams. Eight-five percent of pointwise intravisit ICCs were above 0.75 (range, 0.63 to 0.93), and 65% of pointwise intervisit ICCs were above 0.75 (range, 0.55 to 0.91). Conclusions: VisuALL demonstrated high correlation of MD between tests and good repeatability for individual point sensitivities among three tests in 3 months, except at the points around the blind spot and superiorly. Translational Relevance: The preliminary reproducibility results for VisuALL are encouraging. Its portable design makes it a potentially useful tool for patients with glaucoma, enabling more frequent assessments both at home and in clinical settings.

Using Electronic Health Record Data to Determine the Safety of Aqueous Humor Liquid Biopsies for Molecular Analyses.

Purpose: Knowing the surgical safety of anterior chamber liquid biopsies will support the increased use of proteomics and other molecular analyses to better understand disease mechanisms and therapeutic responses in patients and clinical trials. Manual review of operative notes from different surgeons and procedures in electronic health records (EHRs) is cumbersome, but free-text software tools could facilitate efficient searches. Design: Retrospective case series. Participants: A total of 1418 aqueous humor liquid biopsies from patients undergoing intraocular surgery. Methods: Free-text EHR searches were performed using the Stanford Research Repository cohort discovery tool to identify complications associated with anterior chamber paracentesis and subsequent endophthalmitis. Complications of the surgery unrelated to the biopsy were not reviewed. Main Outcome Measures: Biopsy-associated intraoperative complications and endophthalmitis. Results: A total of 1418 aqueous humor liquid biopsies were performed by 17 experienced surgeons. EHR free-text searches were 100% error-free for surgical complications, >99% for endophthalmitis (<1% false positive), and >93.6% for anesthesia type, requiring manual review for only a limited number of cases. More than 85% of cases were performed under local anesthesia without ocular muscle akinesia. Although the most common indication was cataract (50.1%), other diagnoses included glaucoma, diabetic retinopathy, uveitis, age-related macular degeneration, endophthalmitis, retinitis pigmentosa, and uveal melanoma. A 50- to 100-µL sample was collected in all cases using either a 30-gauge needle or a blunt cannula via a paracentesis. The median follow-up was >7 months. There was only one minor complication (0.07%) identified: a case of a small tear in Descemet membrane without long-term sequelae. No other complications occurred, including other corneal injuries, lens or iris trauma, hyphema, or suprachoroidal hemorrhage. There was no case of postoperative endophthalmitis. Conclusions: Anterior chamber liquid biopsy during intraocular surgery is a safe procedure and may be considered for large-scale collection of aqueous humor samples for molecular analyses. Free-text EHR searches are an efficient approach to reviewing intraoperative procedures. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Comparing a Head-Mounted Smartphone Visual Field Analyzer to Standard Automated Perimetry in Glaucoma: A Prospective Study.

PRECIS: Wang et al compare an FDA-registered head-mounted smartphone device (PalmScan VF2000) with standard automated perimetry (SAP) in glaucoma patients and find that the head-mounted device may not fully recapitulate SAP testing. PURPOSE: This study prospectively compared visual field testing using the PalmScan VF2000 Visual Field Analyzer, a head-mounted smartphone device, with standard automated perimetry (SAP). METHODS: Patients with glaucoma undergoing Humphrey Field Analyzer SAP testing were asked to complete in-office PalmScan testing using a Samsung S5 smartphone in a virtual reality-style headset. Glaucoma severity was defined as SAP mean deviation (MD) >-6 dB for mild, between -6 and -12 dB for moderate, and <-12 dB for severe. Global parameters MD and pattern standard deviation (PSD) from PalmScan and SAP were compared using t-tests and Bland-Altman analyses. Bland-Altmann analyses of PalmScan and SAP MD were conducted for the superonasal, superotemporal, inferonasal, and inferotemporal visual field quadrants. The repeatability of PalmScan was assessed using Spearman's correlations and intraclass correlation coefficients (ICCs). RESULTS: Fifty-one patients (51 eyes) completed both SAP and PalmScan testing and met criteria for analysis. Compared to SAP, global MD and PSD measurements from PalmScan differed by an average of +0.62±0.26 dB (range: -3.25 to +4.60 dB) and -1.00±0.24 dB (range: -6.03 to +2.77 dB), respectively, while MD scores from individual visual field quadrants differed by as much as -6.58 to +11.43 dB. Agreement of PalmScan and SAP in classifying glaucoma severity was 86.3% across all eyes. PalmScan and SAP identified the same quadrant as having the worst visual field defect in 66.7% of eyes. CONCLUSIONS: Despite advantages in cost and accessibility, the PalmScan head-mounted perimetry device may not be able to fully recapitulate SAP testing.

Liquid-biopsy proteomics combined with AI identifies cellular drivers of eye aging and disease in vivo.

Single-cell analysis in living humans is essential for understanding disease mechanisms, but it is impractical in non-regenerative organs, such as the eye and brain, because tissue biopsies would cause serious damage. We resolve this problem by integrating proteomics of liquid biopsies with single-cell transcriptomics from all known ocular cell types to trace the cellular origin of 5,953 proteins detected in the aqueous humor. We identified hundreds of cell-specific protein markers, including for individual retinal cell types. Surprisingly, our results reveal that retinal degeneration occurs in Parkinson's disease, and the cells driving diabetic retinopathy switch with disease stage. Finally, we developed artificial intelligence (AI) models to assess individual cellular aging and found that many eye diseases not associated with chronological age undergo accelerated molecular aging of disease-specific cell types. Our approach, which can be applied to other organ systems, has the potential to transform molecular diagnostics and prognostics while uncovering new cellular disease and aging mechanisms.

Developing a privacy-preserving deep learning model for glaucoma detection: a multicentre study with federated learning.

BACKGROUND: Deep learning (DL) is promising to detect glaucoma. However, patients' privacy and data security are major concerns when pooling all data for model development. We developed a privacy-preserving DL model using the federated learning (FL) paradigm to detect glaucoma from optical coherence tomography (OCT) images. METHODS: This is a multicentre study. The FL paradigm consisted of a 'central server' and seven eye centres in Hong Kong, the USA and Singapore. Each centre first trained a model locally with its own OCT optic disc volumetric dataset and then uploaded its model parameters to the central server. The central server used FedProx algorithm to aggregate all centres' model parameters. Subsequently, the aggregated parameters are redistributed to each centre for its local model optimisation. We experimented with three three-dimensional (3D) networks to evaluate the stabilities of the FL paradigm. Lastly, we tested the FL model on two prospectively collected unseen datasets. RESULTS: We used 9326 volumetric OCT scans from 2785 subjects. The FL model performed consistently well with different networks in 7 centres (accuracies 78.3%-98.5%, 75.9%-97.0%, and 78.3%-97.5%, respectively) and stably in the 2 unseen datasets (accuracies 84.8%-87.7%, 81.3%-84.8%, and 86.0%-87.8%, respectively). The FL model achieved non-inferior performance in classifying glaucoma compared with the traditional model and significantly outperformed the individual models. CONCLUSION: The 3D FL model could leverage all the datasets and achieve generalisable performance, without data exchange across centres. This study demonstrated an OCT-based FL paradigm for glaucoma identification with ensured patient privacy and data security, charting another course toward the real-world transition of artificial intelligence in ophthalmology.

Biobanking of Human Aqueous and Vitreous Liquid Biopsies for Molecular Analyses.

A critical challenge in translational research is establishing a viable and efficient interface between patient care in the operating room (OR) and the research laboratory. Here, we developed a protocol for acquiring high-quality liquid biopsies for molecular analyses from the aqueous humor and the vitreous from patients undergoing eye surgery. In this workflow, a Mobile Operating Room Lab Interface (MORLI) cart equipped with a computer, a barcode scanner, and lab instruments, including onboard cold storage, is used to obtain and archive human biological samples. A web-based data privacy-compliant database enables annotating each sample over its lifetime, and a cartesian coordinate system allows tracking each barcoded specimen in storage, enabling quick and accurate retrieval of samples for downstream analyses. Molecular characterization of human tissue samples not only serves as a diagnostic tool (e.g., to distinguish between infectious endophthalmitis and other non-infectious intraocular inflammation) but also represents an important component of translational research, allowing the identification of new drug targets, development of new diagnostic tools, and personalized therapeutics.

Comparison of Ophthalmologist and Large Language Model Chatbot Responses to Online Patient Eye Care Questions.

Importance: Large language models (LLMs) like ChatGPT appear capable of performing a variety of tasks, including answering patient eye care questions, but have not yet been evaluated in direct comparison with ophthalmologists. It remains unclear whether LLM-generated advice is accurate, appropriate, and safe for eye patients. Objective: To evaluate the quality of ophthalmology advice generated by an LLM chatbot in comparison with ophthalmologist-written advice. Design, Setting, and Participants: This cross-sectional study used deidentified data from an online medical forum, in which patient questions received responses written by American Academy of Ophthalmology (AAO)-affiliated ophthalmologists. A masked panel of 8 board-certified ophthalmologists were asked to distinguish between answers generated by the ChatGPT chatbot and human answers. Posts were dated between 2007 and 2016; data were accessed January 2023 and analysis was performed between March and May 2023. Main Outcomes and Measures: Identification of chatbot and human answers on a 4-point scale (likely or definitely artificial intelligence [AI] vs likely or definitely human) and evaluation of responses for presence of incorrect information, alignment with perceived consensus in the medical community, likelihood to cause harm, and extent of harm. Results: A total of 200 pairs of user questions and answers by AAO-affiliated ophthalmologists were evaluated. The mean (SD) accuracy for distinguishing between AI and human responses was 61.3% (9.7%). Of 800 evaluations of chatbot-written answers, 168 answers (21.0%) were marked as human-written, while 517 of 800 human-written answers (64.6%) were marked as AI-written. Compared with human answers, chatbot answers were more frequently rated as probably or definitely written by AI (prevalence ratio [PR], 1.72; 95% CI, 1.52-1.93). The likelihood of chatbot answers containing incorrect or inappropriate material was comparable with human answers (PR, 0.92; 95% CI, 0.77-1.10), and did not differ from human answers in terms of likelihood of harm (PR, 0.84; 95% CI, 0.67-1.07) nor extent of harm (PR, 0.99; 95% CI, 0.80-1.22). Conclusions and Relevance: In this cross-sectional study of human-written and AI-generated responses to 200 eye care questions from an online advice forum, a chatbot appeared capable of responding to long user-written eye health posts and largely generated appropriate responses that did not differ significantly from ophthalmologist-written responses in terms of incorrect information, likelihood of harm, extent of harm, or deviation from ophthalmologist community standards. Additional research is needed to assess patient attitudes toward LLM-augmented ophthalmologists vs fully autonomous AI content generation, to evaluate clarity and acceptability of LLM-generated answers from the patient perspective, to test the performance of LLMs in a greater variety of clinical contexts, and to determine an optimal manner of utilizing LLMs that is ethical and minimizes harm.

Real-world agreement of same-visit Tono-Pen vs Goldmann applanation intraocular pressure measurements using electronic health records.

Purpose: To compare intraocular pressure (IOP) obtained with Tono-Pen (TP) and Goldmann applanation (GAT) using large-scale electronic health records (EHR). Design: Retrospective cohort study. Methods: A single pair of eligible TP/GAT IOP readings was randomly selected from the EHR for each ophthalmology patient at an academic ophthalmology center (2013-2022), yielding 4550 eligible measurements. We used Bland-Altman analysis to describe agreement between TP/GAT IOP differences and mean IOP measurements. We also used multivariable logistic regression to identify factors associated with different IOP readings in the same eye, including demographics, glaucoma diagnosis, and central corneal thickness (CCT). Primary outcome metrics were discrepant measurements between TP and GAT as defined by two methods: Outcome A (normal TP despite elevated GAT measurements), and Outcome B (TP and GAT IOP differences ≥6 mmHg). Result: The mean TP/GAT IOP difference was 0.15 mmHg ( ± 5.49 mmHg 95% CI). There was high correlation between the measurements (r = 0.790, p < 0.001). We found that TP overestimated pressures at IOP <16.5 mmHg and underestimated at IOP >16.5 mmHg (Fig. 4). Discrepant measurements accounted for 2.6% (N = 116) and 5.2% (N = 238) for outcomes A and B respectively. Patients with thinner CCT had higher odds of discrepant IOP (OR 0.88 per 25 µm increase, CI [0.84-0.92], p < 0.0001; OR 0.88 per 25 µm increase, CI [0.84-0.92], p < 0.0001 for outcomes A and B respectively). Conclusion: In a real-world academic practice setting, TP and GAT IOP measurements demonstrated close agreement, although 2.6% of measurements showed elevated GAT IOP despite normal TP measurements, and 5.2% of measurements were ≥6 mmHg apart.

Managing primary open-angle glaucoma in the setting of suboptimal surgical outcomes in the fellow eye.

A 62-year-old woman with mild myopia presented to her local optometrist for a routine examination and was found to have intraocular pressure (IOP) of 30 mm Hg in both eyes and cupped nerves. She had a family history of glaucoma in her father. She was started on latanoprost in both eyes and was referred for a glaucoma evaluation. On initial evaluation, her IOP was 25 mm Hg in the right eye and 26 mm Hg in the left eye. Central corneal thickness measured 592 µm in the right eye and 581 µm in the left eye. Her angles were open to gonioscopy without any peripheral anterior synechia. She had 1+ nuclear sclerosis with a corrected distance visual acuity (CDVA) of 20/25 in the right eye and 20/30- in the left eye and uncorrected near visual acuity of J1+ in each eye. Her nerves were 0.85 mm in the right eye and 0.75 mm in the left eye. Optical coherence tomography (OCT) showed retinal nerve fiber layer thinning and a dense superior arcuate scotoma into fixation in her right eye, and superior and inferior arcuate scotomas in her left eye (Figures 1 and 2JOURNAL/jcrs/04.03/02158034-202307000-00019/figure1/v/2023-06-26T195222Z/r/image-tiffJOURNAL/jcrs/04.03/02158034-202307000-00019/figure2/v/2023-06-26T195222Z/r/image-tiff, Supplemental Figures 1 and 2, available at http://links.lww.com/JRS/A882 and http://links.lww.com/JRS/A883). She was successively trialed on fixed combination brimonidine-timolol, dorzolamide, and netarsudil, in addition to her latanoprost, but her IOP remained in the mid- to upper 20s in both eyes. The addition of acetazolamide lowered the pressure to 19 mm Hg in both eyes, but she tolerated it poorly. Methazolamide was also attempted with similar side effects. We elected to perform left eye cataract surgery combined with 360-degree viscocanaloplasty and insertion of a Hydrus microstent (Alcon Laboratories, Inc.). Surgery was uncomplicated with IOP of 16 mm Hg on postoperative day 1 with no glaucoma medications. However, by postoperative week 3, IOP returned to 27 mm Hg, and despite restarting latanoprost-netarsudil and finishing her steroid taper, IOP remained at 27 mm Hg by postoperative week 6. Brimonidine-timolol was added back to her left eye regimen and at postoperative week 8, IOP had elevated to 45 mm Hg. Maximizing her therapy with the addition of topical dorzolamide and oral methazolamide brought her IOP back down to 30 mm Hg. At that point, the decision was made to proceed with trabeculectomy of the left eye. The trabeculectomy was uneventful. However, postoperative attempts to augment filtration were rendered less successful by extremely thick Tenon layer. At her most recent follow-up the pressure in the left eye was mid-teens with brimonidine-timolol and dorzolamide. Her right eye IOP is in the upper 20s on maximum topical therapy. Knowing her postoperative course in the left eye, how would you manage the right eye? In addition to currently available options, would you consider a supraciliary shunt such as the MINIject (iSTAR) if such a device were U.S. Food and Drug Administration (FDA)-approved?

Three-Dimensional Multi-Task Deep Learning Model to Detect Glaucomatous Optic Neuropathy and Myopic Features From Optical Coherence Tomography Scans: A Retrospective Multi-Centre Study.

Purpose: We aim to develop a multi-task three-dimensional (3D) deep learning (DL) model to detect glaucomatous optic neuropathy (GON) and myopic features (MF) simultaneously from spectral-domain optical coherence tomography (SDOCT) volumetric scans. Methods: Each volumetric scan was labelled as GON according to the criteria of retinal nerve fibre layer (RNFL) thinning, with a structural defect that correlated in position with the visual field defect (i.e., reference standard). MF were graded by the SDOCT en face images, defined as presence of peripapillary atrophy (PPA), optic disc tilting, or fundus tessellation. The multi-task DL model was developed by ResNet with output of Yes/No GON and Yes/No MF. SDOCT scans were collected in a tertiary eye hospital (Hong Kong SAR, China) for training (80%), tuning (10%), and internal validation (10%). External testing was performed on five independent datasets from eye centres in Hong Kong, the United States, and Singapore, respectively. For GON detection, we compared the model to the average RNFL thickness measurement generated from the SDOCT device. To investigate whether MF can affect the model's performance on GON detection, we conducted subgroup analyses in groups stratified by Yes/No MF. The area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, and accuracy were reported. Results: A total of 8,151 SDOCT volumetric scans from 3,609 eyes were collected. For detecting GON, in the internal validation, the proposed 3D model had significantly higher AUROC (0.949 vs. 0.913, p < 0.001) than average RNFL thickness in discriminating GON from normal. In the external testing, the two approaches had comparable performance. In the subgroup analysis, the multi-task DL model performed significantly better in the group of "no MF" (0.883 vs. 0.965, p-value < 0.001) in one external testing dataset, but no significant difference in internal validation and other external testing datasets. The multi-task DL model's performance to detect MF was also generalizable in all datasets, with the AUROC values ranging from 0.855 to 0.896. Conclusion: The proposed multi-task 3D DL model demonstrated high generalizability in all the datasets and the presence of MF did not affect the accuracy of GON detection generally.

Implant Insertion Procedure of the Port Delivery System With Ranibizumab: Overview and Clinical Pearls.

OBJECTIVE: To describe the Port Delivery System with ranibizumab implant insertion procedure. METHODS: A surgical procedure based on the clinical trial program in patients with retinal diseases. RESULTS: An infusion line is placed in the infero-temporal quadrant; a superotemporal quadrant corneal traction suture is recommended. The superotemporal quadrant peritomy of 6 × 6 mm is executed with gentle, purposeful tissue handling. Generous posterior and lateral sub-Tenon's capsule dissection creates laxity for the subsequent closure. Adequate scleral hemostasis is achieved with wet-field cautery to maintain a clean field. The implant is filled under magnification with a customized formulation of ranibizumab. A precise 3.5-mm-long scleral incision (4 mm posterior and parallel to the limbus) is created to ensure proper implant fit. The exposed pars plana undergoes laser ablation to reduce vitreous hemorrhage risk. A pars plana incision is made, and the implant is inserted perpendicular to the globe and seated flush against the sclera. Complete closure of both the conjunctiva and Tenon's capsule with scleral anchoring and mild tissue overhang at the anterior limbus is performed to reduce conjunctival erosion and retraction risks. CONCLUSION: The procedure is straightforward yet requires precise preoperative and intraoperative preparation and standardized surgical techniques. [Ophthalmic Surg Lasers Imaging Retina. 2022;53:249-256.].

Conjunctiva and Tenon's Capsule Handling in the Port Delivery System with Ranibizumab Implant Insertion Procedure: Surgical Pearls.

OBJECTIVES: To describe conjunctiva and Tenon's capsule handling during the Port Delivery System with ranibizumab (PDS) implant insertion procedure including up-front assessments, planning, and instrumentation, with emphasis placed on the peritomy, scleral dissection, and closure steps. METHODS: Surgical pearls based on experience accumulated in the PDS clinical trial program in patients with retinal diseases. RESULTS: Preoperative preparation, specific instruments, and meticulous techniques are key to optimizing surgical outcomes. Before surgery, assessment of factors that affect conjunctival integrity and an in-office conjunctiva examination are conducted. Gentle, purposeful conjunctiva and Tenon's capsule handling with nontoothed forceps and suturing with a BV needle are recommended to prevent tissue damage. The peritomy is 6 mm by 6 mm, centered around the planned implant location in the superotemporal quadrant. A complete sub-Tenon's capsule dissection is achieved using a wide, robust lateral and posterior dissection technique to free tissue from the sclera and minimize tension. The globe is stabilized during scleral cutdown by grasping the sclera with fine-toothed forceps away from the incision edge to prevent tissue delamination. When closing the peritomy, both the conjunctiva and Tenon's capsule are completely captured and sutured with scleral anchoring at the apex of the peritomy to help prevent conjunctival retraction and erosion. Mitigation and detection of adverse events is critical to successful surgical outcomes. CONCLUSIONS: The PDS implant insertion procedure is straightforward, but it requires planned preoperative preparation, specific instruments, and meticulous techniques. The surgical pearls described here offer insights for optimizing outcomes. [Ophthalmic Surg Lasers Imaging Retina. 2022;53:266-273.].

Management of Key Ocular Adverse Events in Patients Implanted with the Port Delivery System with Ranibizumab.

PURPOSE: To provide strategies for the management of key ocular adverse events (AEs) that may be encountered with the Port Delivery System with ranibizumab (PDS) in practice and provide recommendations that may mitigate such AEs based on clinical trial experiences and considerations from experts in the field. DESIGN: Safety evaluation based on the phase 2 Ladder (NCT02510794) and phase 3 Archway (NCT03677934) trials of the PDS. METHODS: The PDS implant is a permanent, indwelling, and refillable ocular drug delivery system that requires standardized procedural steps for its insertion and refill-exchange procedures, which evolved during the PDS clinical program. We described identified AEs that may arise after implant insertion or refill-exchange procedures, including conjunctival retraction, conjunctival erosion, endophthalmitis, implant dislocation, conjunctival blebs or conjunctival filtering bleb leaks, wound leaks, hypotony, choroidal detachment, vitreous hemorrhage, rhegmatogenous retinal detachment, cataract, and septum dislodgement. RESULTS: Adverse events related to the PDS were well understood, were manageable by trial investigators, and did not prevent patients from achieving optimal outcomes in most cases. CONCLUSIONS: Surgeons using the PDS should be aware of potential ocular AEs and identify them early for optimal management. As with any new surgical procedure, it is important to provide surgeons with appropriate training, ensure adherence to optimal surgical techniques, and continually refine the procedure to mitigate complications and improve outcomes.

Deep Learning for Glaucoma Detection and Identification of Novel Diagnostic Areas in Diverse Real-World Datasets.

Purpose: To develop a three-dimensional (3D) deep learning algorithm to detect glaucoma using spectral-domain optical coherence tomography (SD-OCT) optic nerve head (ONH) cube scans and validate its performance on ethnically diverse real-world datasets and on cropped ONH scans. Methods: In total, 2461 Cirrus SD-OCT ONH scans of 1012 eyes were obtained from the Glaucoma Clinic Imaging Database at the Byers Eye Institute, Stanford University, from March 2010 to December 2017. A 3D deep neural network was trained and tested on this unique raw OCT cube dataset to identify a multimodal definition of glaucoma excluding other concomitant retinal disease and optic neuropathies. A total of 1022 scans of 363 glaucomatous eyes (207 patients) and 542 scans of 291 normal eyes (167 patients) from Stanford were included in training, and 142 scans of 48 glaucomatous eyes (27 patients) and 61 scans of 39 normal eyes (23 patients) were included in the validation set. A total of 3371 scans (Cirrus SD-OCT) from four different countries were used for evaluation of the model: the non overlapping test dataset from Stanford (USA) consisted of 694 scans: 241 scans from 113 normal eyes of 66 patients and 453 scans of 157 glaucomatous eyes of 89 patients. The datasets from Hong Kong (total of 1625 scans; 666 OCT scans from 196 normal eyes of 99 patients and 959 scans of 277 glaucomatous eyes of 155 patients), India (total of 672 scans; 211 scans from 147 normal eyes of 98 patients and 461 scans from 171 glaucomatous eyes of 101 patients), and Nepal (total of 380 scans; 158 scans from 143 normal eyes of 89 patients and 222 scans from 174 glaucomatous eyes of 109 patients) were used for external evaluation. The performance of the model was then evaluated on manually cropped scans from Stanford using a new algorithm called DiagFind. The ONH region was cropped by identifying the appropriate zone of the image in the expected location relative to Bruch's Membrane Opening (BMO) using a commercially available imaging software. Subgroup analyses were performed in groups stratified by eyes, myopia severity of glaucoma, and on a set of glaucoma cases without field defects. Saliency maps were generated to highlight the areas the model used to make a prediction. The model's performance was compared to that of a glaucoma specialist using all available information on a subset of cases. Results: The 3D deep learning system achieved area under the curve (AUC) values of 0.91 (95% CI, 0.90-0.92), 0.80 (95% CI, 0.78-0.82), 0.94 (95% CI, 0.93-0.96), and 0.87 (95% CI, 0.85-0.90) on Stanford, Hong Kong, India, and Nepal datasets, respectively, to detect perimetric glaucoma and AUC values of 0.99 (95% CI, 0.97-1.00), 0.96 (95% CI, 0.93-1.00), and 0.92 (95% CI, 0.89-0.95) on severe, moderate, and mild myopia cases, respectively, and an AUC of 0.77 on cropped scans. The model achieved an AUC value of 0.92 (95% CI, 0.90-0.93) versus that of the human grader with an AUC value of 0.91 on the same subset of scans (\(P=0.99\)). The performance of the model in terms of recall on glaucoma cases without field defects was found to be 0.76 (0.68-0.85). Saliency maps highlighted the lamina cribrosa in glaucomatous eyes versus superficial retina in normal eyes as the regions associated with classification. Conclusions: A 3D convolutional neural network (CNN) trained on SD-OCT ONH cubes can distinguish glaucoma from normal cases in diverse datasets obtained from four different countries. The model trained on additional random cropping data augmentation performed reasonably on manually cropped scans, indicating the importance of lamina cribrosa in glaucoma detection. Translational Relevance: A 3D CNN trained on SD-OCT ONH cubes was developed to detect glaucoma in diverse datasets obtained from four different countries and on cropped scans. The model identified lamina cribrosa as the region associated with glaucoma detection.

Patient Personality and Illness Perceptions in Relation to Follow-Up Appointment Adherence in Neuro-Ophthalmology.

BACKGROUND: Improving patient attendance at medical follow-up visits may have a notable impact on disease and overall health outcomes. Understanding factors contributing to poor attendance is important for identifying at-risk patients and designing interventions to improve clinical outcomes. Our objective was to identify personality and illness perception factors associated with attendance at recommended follow-up visits in a neuro-ophthalmology practice. METHODS: New or established patients (≥18 years) with scheduled neuro-ophthalmology (study) or glaucoma (comparison) appointments at a tertiary care academic medical center completed the Brief Illness Perception Questionnaire and Ten-Item Personality Inventory. Physician recommendations made during the visit were recorded (medications, referrals, follow-up, testing, and procedures). A chart review was performed 18 months after enrollment to assess attendance at follow-up appointment and adherence with other physician recommendations. Multiple variable logistic regression models studied associations between follow-up appointment attendance and demographic factors, appointment factors, and survey responses. RESULTS: Among 152 respondents (97% response rate (152 of 157), aged 19-97 years, 58% female, 34% new, 80 neuro-ophthalmology, 72 glaucoma), neuro-ophthalmology subjects were younger, more likely to be White, non-Hispanic, female and new to the practice than subjects with glaucoma. They reported higher emotional impact, identity, and consequences related to their illness (P = 0.001-0.03). Neuro-ophthalmology physician recommendations included more referrals to other services (17.5% vs 1.4%, P = 0.001, chi-square) and more radiology studies (15% vs 0%, P = 0.001, chi-square), but fewer follow-up visits (75% vs 97%, P < 0.0005, chi-square). Among those with recommended follow-up visits, neuro-ophthalmology subjects had lower rates of on-time appointment attendance (55% vs 77%, P = 0.009, chi-square). In a multiple variable model, on-time follow-up attendance was associated with shorter recommended follow-up interval (≤90 days, P < 0.0005), established (vs new) patient status at enrollment visit (P = 0.04), and glaucoma (P = 0.08), but not subject demographics, illness perception, or personality factors. CONCLUSIONS: Patient demographics, illness perception, and personality traits were not associated with follow-up appointment attendance and therefore unlikely to be useful for identifying patients at risk of being lost to follow-up. New neuro-ophthalmology patients with a follow-up recommended ≥90 days in advance may benefit from targeted interventions to improve follow-up appointment adherence.

Daytime napping is associated with retinal microcirculation: a large population-based study in China.

STUDY OBJECTIVES: To investigate the association between daytime napping and retinal microcirculation. METHODS: This is a cross-sectional study from a prospective population-based cohort. 2,662 participants were recruited after quota sampling. Information on napping was collected through face-to-face interviews. Retinal vascular calibers (RVCs), including central retinal arteriolar equivalent (CRAE), central retinal venular equivalent (CRVE), and arterio-to-venous ratio (AVR), were obtained from fundus photography. Multivariate regression and restricted cubic spline curve were performed to determine the association between RVCs and daytime napping duration. RESULTS: 56.4% participants reported daytime napping regularly. Compared to no nap, daytime nap was related to higher CRAE, with nap duration of 0.5-1 h showing the most significant association. 0.5-1 h daytime nappers displayed an average of 4.18 µm (95% confidence interval [CI] 2.45-5.91, p < 0.001) wider CRAE than non-nappers after adjustment. No significant association was found between CRVE and daytime napping. Moreover, individuals with 0.5-1 h daytime napping had a lower risk for AVR reduction (odds ratio [OR] 0.70, 95% confidence interval [CI] 0.56-0.86, p = 0.001) than non-nappers. Similar association persisted in non-hypertensive population. Restricted cubic spline indicated a J-shaped relationship between AVR reduction and nap duration. CONCLUSION: Retinal microcirculation was positively associated with self-reported 0.5-1 h daytime napping. Better indicators of retinal microcirculation were probably related to nap duration in a J-shaped manner. Also, the possibly beneficial role of 0.5-1 h daytime napping on retinal microcirculation might be independent of clinically diagnosed vascular diseases.

Interpreting Deep Learning Studies in Glaucoma: Unresolved Challenges.

ABSTRACT: Deep learning algorithms as tools for automated image classification have recently experienced rapid growth in imaging-dependent medical specialties, including ophthalmology. However, only a few algorithms tailored to specific health conditions have been able to achieve regulatory approval for autonomous diagnosis. There is now an international effort to establish optimized thresholds for algorithm performance benchmarking in a rapidly evolving artificial intelligence field. This review examines the largest deep learning studies in glaucoma, with special focus on identifying recurrent challenges and limitations within these studies which preclude widespread clinical deployment. We focus on the 3 most common input modalities when diagnosing glaucoma, namely, fundus photographs, spectral domain optical coherence tomography scans, and standard automated perimetry data. We then analyze 3 major challenges present in all studies: defining the algorithm output of glaucoma, determining reliable ground truth datasets, and compiling representative training datasets.

A Multitask Deep-Learning System to Classify Diabetic Macular Edema for Different Optical Coherence Tomography Devices: A Multicenter Analysis.

OBJECTIVE: Diabetic macular edema (DME) is the primary cause of vision loss among individuals with diabetes mellitus (DM). We developed, validated, and tested a deep learning (DL) system for classifying DME using images from three common commercially available optical coherence tomography (OCT) devices. RESEARCH DESIGN AND METHODS: We trained and validated two versions of a multitask convolution neural network (CNN) to classify DME (center-involved DME [CI-DME], non-CI-DME, or absence of DME) using three-dimensional (3D) volume scans and 2D B-scans, respectively. For both 3D and 2D CNNs, we used the residual network (ResNet) as the backbone. For the 3D CNN, we used a 3D version of ResNet-34 with the last fully connected layer removed as the feature extraction module. A total of 73,746 OCT images were used for training and primary validation. External testing was performed using 26,981 images across seven independent data sets from Singapore, Hong Kong, the U.S., China, and Australia. RESULTS: In classifying the presence or absence of DME, the DL system achieved area under the receiver operating characteristic curves (AUROCs) of 0.937 (95% CI 0.920-0.954), 0.958 (0.930-0.977), and 0.965 (0.948-0.977) for the primary data set obtained from CIRRUS, SPECTRALIS, and Triton OCTs, respectively, in addition to AUROCs >0.906 for the external data sets. For further classification of the CI-DME and non-CI-DME subgroups, the AUROCs were 0.968 (0.940-0.995), 0.951 (0.898-0.982), and 0.975 (0.947-0.991) for the primary data set and >0.894 for the external data sets. CONCLUSIONS: We demonstrated excellent performance with a DL system for the automated classification of DME, highlighting its potential as a promising second-line screening tool for patients with DM, which may potentially create a more effective triaging mechanism to eye clinics.

Smartphone-based Ophthalmic Imaging Compared With Spectral-domain Optical Coherence Tomography Assessment of Vertical Cup-to-disc Ratio Among Adults in Southwestern Uganda.

PRECIS: Using optical coherence tomography (OCT) measurements as a reference standard for vertical cup-to-disc ratio (vCDR), a smartphone-based ophthalmic camera has a sensitivity of 67.7% and specificity of 96.7% to detect a vCDR>0.5. PURPOSE: The purpose of this study was to assess the performance of a smartphone-based ophthalmic camera system using an Apple iPhone 6S and an adapter, Paxos Scope, to obtain adequate dilated fundus photos to measure clinically useful vCDR cutoffs. PATIENTS AND METHODS: Adult patients from a government tertiary level eye hospital in Southwestern Uganda were prospectively recruited from January to April 2019. All patients experienced a comprehensive eye examination, dilated posterior segment indirect ophthalmoscope imaging with the Paxos Scope, and spectral-domain OCT imaging with a Cirrus HD-OCT to measure vCDR. Patients' eyes excluded had media opacities or existing disease precluding a view of the fundus. Fundus images underwent a single masked review to assign vCDR at increments of 0.1. Descriptive statistics, parametric and χ2 tests for significance, repeated measures correlation, κ, receiver operating characteristics curve, and Bland-Altman were used to assess the data. RESULTS: Among 467 (consecutive) individuals, fundus photographs acquired with the Paxos Scope demonstrated a 67.7% [95% confidence interval (CI), 63.0-72.0] sensitivity and 96.7% (95% CI, 94.2-98.3) specificity to detect a vCDR>0.5, using OCT as the reference standard. A total of 138 eyes were excluded due to poor imaging acquisition, such as dense cataract, rendering 796 eyes for analysis. The vCDR from graded Paxos Scope images and OCT correlated well with repeated measures correlation of 0.82 (95% CI, 0.77-0.86, P<0.001) and agreement, dichotomized as >0.5 or ≤0.5, was 80.9% (κ=0.63±0.034, P<0.001). Among glaucoma and glaucoma suspects (85 eyes), the sensitivity and specificity dichotomized using vCDR>0.5 were 97.5% (95% CI, 91.3-99.7) and 80.0% (95% CI, 28.4-99.5), respectively. The area under the receiver operating characteristics curve was 0.92 (95% CI, 0.89-0.94) for all eyes and 0.98 (95% CI, 0.78-1.0) for glaucoma and glaucoma suspects. CONCLUSIONS: The Paxos Scope produced images that can be reliably used to estimate vCDR, which is closely aligned with the automated algorithm from the OCT optic disc cube scan. The low-cost, ready-to-integrate adapter, and minimal training requirements make it a viable option for population-based screening in low-resource settings.

Automatic Identification of Referral-Warranted Diabetic Retinopathy Using Deep Learning on Mobile Phone Images.

Purpose: To evaluate the performance of a deep learning algorithm in the detection of referral-warranted diabetic retinopathy (RDR) on low-resolution fundus images acquired with a smartphone and indirect ophthalmoscope lens adapter. Methods: An automated deep learning algorithm trained on 92,364 traditional fundus camera images was tested on a dataset of smartphone fundus images from 103 eyes acquired from two previously published studies. Images were extracted from live video screenshots from fundus examinations using a commercially available lens adapter and exported as a screenshot from live video clips filmed at 1080p resolution. Each image was graded twice by a board-certified ophthalmologist and compared to the output of the algorithm, which classified each image as having RDR (moderate nonproliferative DR or worse) or no RDR. Results: In spite of the presence of multiple artifacts (lens glare, lens particulates/smudging, user hands over the objective lens) and low-resolution images achieved by users of various levels of medical training, the algorithm achieved a 0.89 (95% confidence interval [CI] 0.83-0.95) area under the curve with an 89% sensitivity (95% CI 81%-100%) and 83% specificity (95% CI 77%-89%) for detecting RDR on mobile phone acquired fundus photos. Conclusions: The fully data-driven artificial intelligence-based grading algorithm herein can be used to screen fundus photos taken from mobile devices and identify with high reliability which cases should be referred to an ophthalmologist for further evaluation and treatment. Translational Relevance: The implementation of this algorithm on a global basis could drastically reduce the rate of vision loss attributed to DR.