Ocular Fundus Photography Research Articles

The application of artificial intelligence in diabetic retinopathy: progress and prospects

In recent years, artificial intelligence (AI), especially deep learning models, has increasingly been integrated into diagnosing and treating diabetic retinopathy (DR). From delving into the singular realm of ocular fundus photography to the gradual development of proteomics and other molecular approaches, from machine learning (ML) to deep learning (DL), the journey has seen a transition from a binary diagnosis of “presence or absence” to the capability of discerning the progression and severity of DR based on images from various stages of the disease course. Since the FDA approval of IDx-DR in 2018, a plethora of AI models has mushroomed, gradually gaining recognition through a myriad of clinical trials and validations. AI has greatly improved early DR detection, and we’re nearing the use of AI in telemedicine to tackle medical resource shortages and health inequities in various areas. This comprehensive review meticulously analyzes the literature and clinical trials of recent years, highlighting key AI models for DR diagnosis and treatment, including their theoretical bases, features, applicability, and addressing current challenges like bias, transparency, and ethics. It also presents a prospective outlook on the future development in this domain.

A Deep Learning Approach for Accurate Discrimination Between Optic Disc Drusen and Papilledema on Fundus Photographs.

Optic disc drusen (ODD) represent an important differential diagnosis of papilledema caused by intracranial hypertension, but their distinction may be difficult in clinical practice. The aim of this study was to train, validate, and test a dedicated deep learning system (DLS) for binary classification of ODD vs papilledema (including various subgroups within each category), on conventional mydriatic digital ocular fundus photographs collected in a large international multiethnic population. This retrospective study included 4,508 color fundus images in 2,180 patients from 30 neuro-ophthalmology centers (19 countries) participating in the Brain and Optic Nerve Study with Artificial Intelligence (BONSAI) Group. For training and internal validation, we used 857 ODD images and 3,230 papilledema images, in 1,959 patients. External testing was performed on an independent data set (221 patients), including 207 images with ODD (96 visible and 111 buried), provided by 3 centers of the Optic Disc Drusen Studies Consortium, and 214 images of papilledema (92 mild-to-moderate and 122 severe) from a previously validated study. The DLS could accurately distinguish between all ODD and papilledema (all severities included): area under the receiver operating characteristic curve (AUC) 0.97 (95% confidence interval [CI], 0.96-0.98), accuracy 90.5% (95% CI, 88.0%-92.9%), sensitivity 86.0% (95% CI, 82.1%-90.1%), and specificity 94.9% (95% CI, 92.3%-97.6%). The performance of the DLS remained high for discrimination of buried ODD from mild-to-moderate papilledema: AUC 0.93 (95% CI, 0.90-0.96), accuracy 84.2% (95% CI, 80.2%-88.6%), sensitivity 78.4% (95% CI, 72.2%-84.7%), and specificity 91.3% (95% CI, 87.0%-96.4%). A dedicated DLS can accurately distinguish between ODD and papilledema caused by intracranial hypertension, even when considering buried ODD vs mild-to-moderate papilledema.

The Role of Artificial Intelligence in Predicting Optic Neuritis Subtypes From Ocular Fundus Photographs.

Optic neuritis (ON) is a complex clinical syndrome that has diverse etiologies and treatments based on its subtypes. Notably, ON associated with multiple sclerosis (MS ON) has a good prognosis for recovery irrespective of treatment, whereas ON associated with other conditions including neuromyelitis optica spectrum disorders or myelin oligodendrocyte glycoprotein antibody-associated disease is often associated with less favorable outcomes. Delay in treatment of these non-MS ON subtypes can lead to irreversible vision loss. It is important to distinguish MS ON from other ON subtypes early, to guide appropriate management. Yet, identifying ON and differentiating subtypes can be challenging as MRI and serological antibody test results are not always readily available in the acute setting. The purpose of this study is to develop a deep learning artificial intelligence (AI) algorithm to predict subtype based on fundus photographs, to aid the diagnostic evaluation of patients with suspected ON. This was a retrospective study of patients with ON seen at our institution between 2007 and 2022. Fundus photographs (1,599) were retrospectively collected from a total of 321 patients classified into 2 groups: MS ON (262 patients; 1,114 photographs) and non-MS ON (59 patients; 485 photographs). The dataset was divided into training and holdout test sets with an 80%/20% ratio, using stratified sampling to ensure equal representation of MS ON and non-MS ON patients in both sets. Model hyperparameters were tuned using 5-fold cross-validation on the training dataset. The overall performance and generalizability of the model was subsequently evaluated on the holdout test set. The receiver operating characteristic (ROC) curve for the developed model, evaluated on the holdout test dataset, yielded an area under the ROC curve of 0.83 (95% confidence interval [CI], 0.72-0.92). The model attained an accuracy of 76.2% (95% CI, 68.4-83.1), a sensitivity of 74.2% (95% CI, 55.9-87.4) and a specificity of 76.9% (95% CI, 67.6-85.0) in classifying images as non-MS-related ON. This study provides preliminary evidence supporting a role for AI in differentiating non-MS ON subtypes from MS ON. Future work will aim to increase the size of the dataset and explore the role of combining clinical and paraclinical measures to refine deep learning models over time.

Eye stroke protocol in in the emergency department

BackgroundVisual outcomes of acute central and branch retinal artery occlusions (CRAO/BRAO) are poor and acute treatment options are limited by delayed diagnosis. In the hyper-acute setting, the ocular fundus may appear “normal”, making recognition challenging, but is facilitated by retinal optical coherence tomography (OCT), which is seldom available in emergency departments (ED). We evaluated the use of non-mydriatic ocular fundus photographs (NMFP) combined with OCT to facilitate ultra-rapid remote diagnosis and stroke alert for patients with acute vision loss presenting to the ED. MethodsProspective evaluation of all CRAO/BRAO between 06/06/2023-06/06/2024 who had NMFP-OCT in our general ED affiliated with a stroke center. ResultsOver 1 year, 22 patients were diagnosed with CRAO, 4 with BRAO. Five patients presented within 4.5 hours of vision loss onset, 6 within 4.5 to ≤12 hours and 15 within >12 to 24 hours. On average, NMFP-OCT was performed within 141 minutes of presentation to the ED (range 27- 422 minutes). Diagnosis of acute RAO was made remotely with NMFP-OCT within 4.5 hours in 4 patients, 2 of whom received intravenous thrombolysis. Of the 9 patients with NMFP-OCT within 12 hours of symptom onset, 5 patients had subtle retinal whitening on color fundus photograph, but all had OCT inner retinal hyper-reflectivity/edema. ConclusionImplementation of NMFP-OCT in a general ED enables rapid remote diagnosis of CRAO/BRAO and facilitates initiation of an eye stroke protocol in acute patients. OCT complements color fundus photography and provides greater diagnostic accuracy in hyperacute cases with near-normal appearing ocular fundi.

The BONSAI (Brain and Optic Nerve Study with Artificial Intelligence) deep learning system can accurately identify pediatric papilledema on standard ocular fundus photographs

BackgroundPediatric papilledema often reflects an underlying severe neurologic disorder and may be difficult to appreciate, especially in young children. Ocular fundus photographs are easy to obtain even in young children and in nonophthalmology settings. The aim of our study was to ascertain whether an improved deep-learning system (DLS), previously validated in adults, can accurately identify papilledema and other optic disk abnormalities in children. MethodsThe DLS was tested on mydriatic fundus photographs obtained in a multiethnic pediatric population (<17 years) from three centers (Atlanta-USA; Bucharest-Romania; Singapore). The DLS’s multiclass classification accuracy (ie, normal optic disk, papilledema, disks with other abnormality) was calculated, and the DLS’s performance to specifically detect papilledema and normal disks was evaluated in a one-vs-rest strategy using the AUC, sensitivity and specificity, with reference to expert neuro-ophthalmologists. ResultsExternal testing was performed on 898 fundus photographs:447 patients; mean age, 10.33 (231 patients ≤10 years of age; 216, 11-16 years); 558 normal disks, 254 papilledema, 86 other disk abnormalities. Overall multiclass accuracy of the DLS was 89.6% (range, 87.8%-91.6%). The DLS successfully distinguished “normal” from “abnormal” optic disks (AUC 0.99 [0.98-0.99]; sensitivity, 87.3% [84.9%-89.8%]; specificity, 98.5% [97.6%-99.6%]), and “papilledema” from “normal and other” (AUC 0.99 [0.98-1.0]; sensitivity, 98.0% [96.8%-99.4%]; specificity, 94.1% (92.4%-95.9%)]. ConclusionsOur DLS reliably distinguished papilledema from normal optic disks and other disk abnormalities in children, suggesting it could be utilized as a diagnostic aid for the assessment of optic nerve head appearance in the pediatric age group.

Artificial intelligence and neuro‐ophthalmology

Deep‐learning (DL) methods have been successfully applied to Neuro‐ophthalmology and in particular to interpretation of images of retina and the optic nerve for detection of life‐ and/or vision‐treating cerebral conditions. DL systems are nowadays able to accurately detect papilledema related to raised intracranial pressure, (1) as well as other optic neuropathies, on standard retinal photographs alone. In addition to detection of specific retinal features and abnormalities, more recent DL methods suggest that neurodegenerative conditions (Alzheimer's disease, Parkinson's disease), affecting the ocular structures, can be also detected on trivial retinal photographs.(2).Ophthalmologists and neuro‐ophthalmologists are in a privileged position to detect cerebral conditions affecting the eyes, using novel, cutting‐edge DL technologies, which may transform our clinical practice in the next years.(3) Such algorithms may be implemented in the future, if further validated in real‐life situations, for the triaging of patients who are seen in non‐ophthalmic set‐ups (emergency rooms, neurology offices, etc.). This presentation will share the current hopes, doubts, successes and failures of assistive AI applied to Neuro‐ophthalmic conditions.References. Milea D, Najjar RP, Zhubo J, Ting D, Vasseneix C, Xu X, Aghsaei Fard M, Fonseca P, Vanikieti K, Lagrèze WA, La Morgia C, Cheung CY, Hamann S, Chiquet C, Sanda N, Yang H, Mejico LJ, Rougier M‐B, Kho R, Thi Ha Chau T, Singhal S, Gohier P, Clermont‐Vignal C, Cheng C‐Y, Jonas JB, Yu‐Wai‐Man P, Fraser CL, Chen JJ, Ambika S, Miller NR, Liu Y, Newman NJ, Wong TY, Biousse V for the BONSAI Group. Artificial Intelligence to Detect Papilledema from Ocular Fundus Photographs. N Engl J Med. 2020 Apr 30;382(18):1687–1695. Cheung CY, Ran A, Wang S, Chan VTT, Sham K, Hilal S, Venketasubramanian N, Cheng CY, Sabanayagam C, Tham YC, Schmetterer L, McKay GJ, Williams MA, 12, Wong A, Au L, Lu A, Yam JC, 1 Tham CC, Chen JJ, Dumitrascu OM, Pheng‐Ann H, Kwok T, Mok V **, Milea D**, Chen LH, ** Wong TY**. A deep‐learning screening model for Alzheimer's disease based on retinal photographs. Lancet Digital Medicine, 2022 Nov;4(11):e806‐e815. Sathianvichitr K, Lamoureux O, Nakada S, Tang Z, Schmetterer L, Chen C, Cheung CY, Najjar RP, Milea D. Through the eyes into the brain, using artificial intelligence. Ann Acad Med Singap. 2023 Feb;52(2):88–95.

Application of a Deep Learning System to Detect Papilledema on Nonmydriatic Ocular Fundus Photographs in an Emergency Department

The Fundus photography vs Ophthalmoscopy Trial Outcomes in the Emergency Department (FOTO-ED) studies showed that ED providers poorly recognized funduscopic findings in patients in the ED. We tested a modified version of the Brain and Optic Nerve Study Artificial Intelligence (BONSAI) deep learning system on nonmydriatic fundus photographs from the FOTO-ED studies to determine if the deep learning system could have improved the detection of papilledema had it been available to ED providers as a real-time diagnostic aid. Retrospective secondary analysis of a cohort of patients included in the FOTO-ED studies. The testing data set included 1608 photographs obtained from 828 patients in the FOTO-ED studies. Photographs were reclassified according to the optic disc classification system used by the deep learning system ("normal optic discs," "papilledema," and "other optic disc abnormalities"). The system's performance was evaluated by calculating the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity using a 1-vs-rest strategy, with reference to expert neuro-ophthalmologists. The BONSAI deep learning system successfully distinguished normal from abnormal optic discs (AUC 0.92 [95% confidence interval {CI} 0.90-0.93]; sensitivity 75.6% [73.7%-77.5%] and specificity 89.6% [86.3%-92.8%]), and papilledema from normal and others (AUC 0.97 [0.95-0.99]; sensitivity 84.0% [75.0%-92.6%] and specificity 98.9% [98.5%-99.4%]). Six patients with missed papilledema in 1 eye were correctly identified by the deep learning system as having papilledema in the other eye. The BONSAI deep learning system was able to reliably identify papilledema and normal optic discs on nonmydriatic photographs obtained in the FOTO-ED studies. Our deep learning system has excellent potential as a diagnostic aid in EDs and non-ophthalmology clinics equipped with nonmydriatic fundus cameras. NOTE: Publication of this article is sponsored by the American Ophthalmological Society.

Study of changes in choroidal thickness with severity of diabetic retinopathy and diabetic macular edema in type 2 diabetic patients

: Diabetic retinopathy (DR) is one of the long-term microvascular complications of Diabetes mellitus. Chronic hyperglycemia cause microvascular abnormalities to both retina and choroid. Optical coherence tomography (OCT) is a non-invasive fundus imaging modality, which plays a vital role in revealing the pathogenesis and development of retinal–choroidal diseases.This prospective observational study included 128 eyes of 64 subjects diagnosed with type 2 diabetes and out of 128 eyes, 113 (88%) eyes were found to have DR. Collected data included age, gender, duration of diabetes, glycemic control, comprehensive ocular examination, fundus photography, and CT measurement on OCT.Mean age in the study group was 53.71 ± 9.37 years (45–70 years). Out of 113 eyes, 19 (17%) eyes were diagnosed as mild NPDR, 21 (19%) eyes had moderate NPDR, 36 (31%) eyes had severe NPDR and 37 (33%) eyes had PDR. The average SFCT in MILD NPDR was 310 ± 14.70 μm at 95% CI (1.86), MODERATE NPDR was 316 ± 17.97 μm at 95% CI (2.70), SEVERE NPDR was 326.02 ± 14.05 μm at 95% CI (4.59) and PDR was 298.55 ± 18.75μm at 95% CI. (1.41). The presence of DME significantly affects average SFCT. It was observed that choroidal thickness tends to increase as the severity DR with DME progress.The average SFCT was thicker in patients with increase in severity of DR, and then SFCT decreased in patients with PDR. As the severity of NPDR increased from mild to moderate to severe NPDR, there was increase in sub foveal choroidal thickness. (p&amp;#60;0.05). PDR showed decrease in sub foveal choroidal thickness in comparison to NPDR which is highly statistically significant. p=0.01 (p&amp;#60;0.05). The presence of DME significantly affects average SFCT (p&amp;#60;0.05). It was observed that choroidal thickness tends to increase as the severity DR with DME progress. Thus, Choroidal thickness measurement can help in assessment of DR pathogenesis.

Retinal vessel caliber and tortuosity and prediction of 5-year incidence of hypertension.

With arterial hypertension as a global risk factor for cerebrovascular and cardiovascular diseases, we examined whether retinal blood vessel caliber and tortuosity assessed by a vessel-constraint network model can predict the incidence of hypertension. The community-based prospective study included 9230 individuals who were followed for 5 years. Ocular fundus photographs taken at baseline were analyzed by a vessel-constraint network model. Within the 5-year follow-up, 1279 (18.8%) and 474 (7.0%) participants out of 6813 individuals free of hypertension at baseline developed hypertension and severe hypertension, respectively. In multivariable analysis, a higher incidence of hypertension was related to a narrower retinal arteriolar diameter ( P < 0.001), wider venular diameter ( P = 0.005), and a smaller arteriole-to-venule diameter ratio ( P < 0.001) at baseline. Individuals with the 5% narrowest arteriole or the 5% widest venule diameter had a 17.1-fold [95% confidence interval (CI):7.9, 37.2] or 2.3-fold (95% CI: 1.4, 3.7) increased risk for developing hypertension, as compared with those with the 5% widest arteriole or the 5% narrowest venule. The area under the receiver operator characteristic curve for predicting the 5-year incidence of hypertension and severe hypertension was 0.791 (95% CI: 0.778, 0.804) and 0.839 (95% CI: 0.821, 0.856), respectively. Although the venular tortuosity was positively associated with the presence of hypertension at baseline ( P = 0.01), neither arteriolar tortuosity nor venular tortuosity was associated with incident hypertension (both P ≥ 0.10). Narrower retinal arterioles and wider venules indicate an increased risk for incident hypertension within 5 years, while tortuous retinal venules are associated with the presence rather than the incidence of hypertension. The automatic assessment of retinal vessel features performed well in identifying individuals at risk of developing hypertension.

Visual impairment and blindness due to retinitis pigmentosa in India: 15-year follow-up of the Andhra Pradesh Eye Disease Study cohort.

To assess the incidence, visual impairment, and blindness due to retinitis pigmentosa (RP) in a rural southern Indian cohort. This is a population-based longitudinal cohort study of participants with RP from the Andhra Pradesh Eye Disease Study (APEDS) cohorts I and III, respectively. The study included participants with RP of APEDS I who were followed until APEDS III. Their demographic data along with ocular features, fundus photographs, and visual fields (Humphrey) were collected. Descriptive statistics using mean ± standard deviation with interquartile range (IQR) were calculated. The main outcome measures were RP incidence, visual impairment, and blindness as per the World Health Organization (WHO) definitions. At baseline (APEDS I), 7771 participants residing in three rural areas were examined. There were nine participants with RP with a mean age at baseline of 47.33 ± 10.89 years (IQR: 39-55). There was a male preponderance (6:3), and the mean best-corrected visual acuity (BCVA) of 18 eyes from nine participants with RP was 1.2 ± 0.72 logarithm of minimum angle of resolution (logMAR; IQR: 0.7-1.6). Over a mean follow-up duration of 15 years, 5395/7771 (69.4%) were re-examined, which included seven RP participants from APEDS 1. Additionally, two new participants with RP were identified; so, the overall incidence was 370/ million in 15 years (24.7/million per year). The mean BCVA of 14 eyes of seven participants with RP who were re-examined in APEDS III was 2.17 ± 0.56 logMAR (IQR: 1.8-2.6), and five of these seven participants with RP developed incident blindness during the follow-up period. RP is a prevalent disease in southern India that warrants appropriate strategies to prevent this condition.

A Deep Learning System for Automated Quality Evaluation of Optic Disc Photographs in Neuro-Ophthalmic Disorders

The quality of ocular fundus photographs can affect the accuracy of the morphologic assessment of the optic nerve head (ONH), either by humans or by deep learning systems (DLS). In order to automatically identify ONH photographs of optimal quality, we have developed, trained, and tested a DLS, using an international, multicentre, multi-ethnic dataset of 5015 ocular fundus photographs from 31 centres in 20 countries participating to the Brain and Optic Nerve Study with Artificial Intelligence (BONSAI). The reference standard in image quality was established by three experts who independently classified photographs as of "good", "borderline", or "poor" quality. The DLS was trained on 4208 fundus photographs and tested on an independent external dataset of 807 photographs, using a multi-class model, evaluated with a one-vs-rest classification strategy. In the external-testing dataset, the DLS could identify with excellent performance "good" quality photographs (AUC = 0.93 (95% CI, 0.91-0.95), accuracy = 91.4% (95% CI, 90.0-92.9%), sensitivity = 93.8% (95% CI, 92.5-95.2%), specificity = 75.9% (95% CI, 69.7-82.1%) and "poor" quality photographs (AUC = 1.00 (95% CI, 0.99-1.00), accuracy = 99.1% (95% CI, 98.6-99.6%), sensitivity = 81.5% (95% CI, 70.6-93.8%), specificity = 99.7% (95% CI, 99.6-100.0%). "Borderline" quality images were also accurately classified (AUC = 0.90 (95% CI, 0.88-0.93), accuracy = 90.6% (95% CI, 89.1-92.2%), sensitivity = 65.4% (95% CI, 56.6-72.9%), specificity = 93.4% (95% CI, 92.1-94.8%). The overall accuracy to distinguish among the three classes was 90.6% (95% CI, 89.1-92.1%), suggesting that this DLS could select optimal quality fundus photographs in patients with neuro-ophthalmic and neurological disorders affecting the ONH.

A narrative review of retinal vascular parameters and the applications (Part II): Diagnosis in stroke.

The retina, as an external extension of the diencephalon, can be directly, noninvasively observed by ocular fundus photography. Therefore, it offers a convenient and feasible mode to study nervous system diseases. Caliber, tortuosity, and fractal dimension, as three commonly used retinal vascular parameters, are not only the reflection of structural changes in the retinal microcirculation but also capture the branching pattern or density changes of the retinal microvascular network. Therefore, it contributes to better reflecting the subclinical pathological changes (e.g., lacunar stroke and small cerebral vascular disease) and predicting the risk of incident stroke and recurrent stroke.

Efficacy of a Deep Learning System for Screening Myopic Maculopathy Based on Color Fundus Photographs.

The maculopathy in highly myopic eyes is complex. Its clinical diagnosis is a huge workload and subjective. To simply and quickly classify pathologic myopia (PM), a deep learning algorithm was developed and assessed to screen myopic maculopathy lesions based on color fundus photographs. This study included 10,347 ocular fundus photographs from 7606 participants. Of these photographs, 8210 were used for training and validation, and 2137 for external testing. A deep learning algorithm was trained, validated, and externally tested to screen myopic maculopathy which was classified into four categories: normal or mild tessellated fundus, severe tessellated fundus, early-stage PM, and advanced-stage PM. The area under the precision-recall curve, the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, accuracy, and Cohen's kappa were calculated and compared with those of retina specialists. In the validation data set, the model detected normal or mild tessellated fundus, severe tessellated fundus, early-stage PM, and advanced-stage PM with AUCs of 0.98, 0.95, 0.99, and 1.00, respectively; while in the external-testing data set of 2137 photographs, the model had AUCs of 0.99, 0.96, 0.98, and 1.00, respectively. We developed a deep learning model for detection and classification of myopic maculopathy based on fundus photographs. Our model achieved high sensitivities, specificities, and reliable Cohen's kappa, compared with those of attending ophthalmologists.

Ophthalmologic problems correlates with cognitive impairment in patients with Parkinson's disease.

ObjectiveVisual impairment is a common non-motor symptom (NMS) in patients with Parkinson's disease (PD) and its implications for cognitive impairment remain controversial. We wished to survey the prevalence of visual impairment in Chinese Parkinson's patients based on the Visual Impairment in Parkinson's Disease Questionnaire (VIPD-Q), identify the pathogens that lead to visual impairment, and develop a predictive model for cognitive impairment risk in Parkinson's based on ophthalmic parameters.MethodsA total of 205 patients with Parkinson's disease and 200 age-matched controls completed the VIPD-Q and underwent neuro-ophthalmologic examinations, including ocular fundus photography and optical coherence tomography. We conducted nomogram analysis and the predictive model was summarized using the multivariate logistic and LASSO regression and verified via bootstrap validation.ResultsOne or more ophthalmologic symptoms were present in 57% of patients with Parkinson's disease, compared with 14% of the controls (χ2-test; p < 0.001). The visual impairment questionnaire showed good sensitivity and specificity (area under the curve [AUC] = 0.918, p < 0.001) and a strong correlation with MoCA scores (Pearson r = −0.4652, p < 0.001). Comparing visual impairment scores between pre- and post-deep brain stimulation groups showed that DBS improved visual function (U-test, p < 0.001). The thickness of the retinal nerve fiber layer and vessel percentage area predicted cognitive impairment in PD.InterpretationThe study findings provide novel mechanistic insights into visual impairment and cognitive decline in Parkinson's disease. The results inform an effective tool for predicting cognitive deterioration in Parkinson's based on ophthalmic parameters.

Valérie Biousse: a leader in neuro-ophthalmology

Valérie Biousse: a leader in neuro-ophthalmology

Efficiency and Complication of 577-nm Laser Membranotomy for the Treatment of Retinal Sub-Inner Limiting Membrane Hemorrhage.

We investigated the clinical efficiency and complications of treatment of retinal sub-inner limiting membrane (sub-ILM) hemorrhage by 577-nm semiconductive laser membranotomy. The clinical features, ocular fundus photography, and SD-OCT image of patients who received 577-nm laser membranotomy for sub-ILM hemorrhage were assessed from January 2017 to April 2022 in this retrospective case-series study. A total of 19 patients (19 eyes) were treated for sub-ILM hemorrhage of the macula, in which eight were women and 11 were men. The age was 15-83 years (with an average age of 49.05 ± 19.41 years old). The right eye was affected in 12 patients, the left eye in seven patients. The follow-up period after laser treatment was from 0.5 to 9 months (with an average follow-up time of 3.25 ± 2.45 months). Treatment was not successful in one patient, and 577-nm laser membranotomy was successful in 18 patients (equaling a success rate of 94.74%). The best corrected visual acuity (BCVA) before laser treatment was from figure count to 40/200, and the BCVA after laser treatment was from 20/2000 to 20/20. Complications after laser treatment comprised macular hole (one patient), macular epi-membrane (one patient), vitreous hemorrhage without absorption (two patients), and sub-ILM cavity (12 patients). The 577-nm laser is effective in treating sub-ILM hemorrhage and has a high success rate. Posttreatment complications should be monitored, and vitrectomy was needed with long-lasting vitreous hemorrhage and macular hole.

Deep Learning Can Accurately Distinguish Between True Papilledema and Optic Disc Drusen On Ocular Fundus Photographs (P17-12.001)

Deep Learning Can Accurately Distinguish Between True Papilledema and Optic Disc Drusen On Ocular Fundus Photographs (P17-12.001)

Artificial Intelligence for Screening of Multiple Retinal and Optic Nerve Diseases

The lack of experienced ophthalmologists limits the early diagnosis of retinal diseases. Artificial intelligence can be an efficient real-time way for screening retinal diseases. To develop and prospectively validate a deep learning (DL) algorithm that, based on ocular fundus images, recognizes numerous retinal diseases simultaneously in clinical practice. This multicenter, diagnostic study at 65 public medical screening centers and hospitals in 19 Chinese provinces included individuals attending annual routine medical examinations and participants of population-based and community-based studies. Based on 120 002 ocular fundus photographs, the Retinal Artificial Intelligence Diagnosis System (RAIDS) was developed to identify 10 retinal diseases. RAIDS was validated in a prospective collected data set, and the performance between RAIDS and ophthalmologists was compared in the data sets of the population-based Beijing Eye Study and the community-based Kailuan Eye Study. The performance of each classifier included sensitivity, specificity, accuracy, F1 score, and Cohen κ score. In the prospective validation data set of 208 758 images collected from 110 784 individuals (median [range] age, 42 [8-87] years; 115 443 [55.3%] female), RAIDS achieved a sensitivity of 89.8% (95% CI, 89.5%-90.1%) to detect any of 10 retinal diseases. RAIDS differentiated 10 retinal diseases with accuracies ranging from 95.3% to 99.9%, without marked differences between medical screening centers and geographical regions in China. Compared with retinal specialists, RAIDS achieved a higher sensitivity for detection of any retinal abnormality (RAIDS, 91.7% [95% CI, 90.6%-92.8%]; certified ophthalmologists, 83.7% [95% CI, 82.1%-85.1%]; junior retinal specialists, 86.4% [95% CI, 84.9%-87.7%]; and senior retinal specialists, 88.5% [95% CI, 87.1%-89.8%]). RAIDS reached a superior or similar diagnostic sensitivity compared with senior retinal specialists in the detection of 7 of 10 retinal diseases (ie, referral diabetic retinopathy, referral possible glaucoma, macular hole, epiretinal macular membrane, hypertensive retinopathy, myelinated fibers, and retinitis pigmentosa). It achieved a performance comparable with the performance by certified ophthalmologists in 2 diseases (ie, age-related macular degeneration and retinal vein occlusion). Compared with ophthalmologists, RAIDS needed 96% to 97% less time for the image assessment. In this diagnostic study, the DL system was associated with accurately distinguishing 10 retinal diseases in real time. This technology may help overcome the lack of experienced ophthalmologists in underdeveloped areas.

Nonmydriatic Retinal Photography in the Outpatient Neurology Resident Clinic.

The funduscopic examination is an essential component of the neurologic examination. However, examination of the ocular fundus with a direct ophthalmoscope is often difficult. Nonmydriatic ocular fundus photography allows direct visualization of the ocular fundus with high-quality photographs. We used nonmydriatic ocular fundus photography to improve patient care and funduscopy skills of residents in the Neurology Resident Clinic. At the time of triage, funduscopic photographs of all new neurology resident clinic patients were taken. The images were imported into the hospital's imaging software. The residents completed a full neurologic examination, including a funduscopic examination with a handheld ophthalmoscope. At the time of staffing the patients with the attendings, the residents received immediate feedback and teaching on retina photograph evaluation. A total of 255 patients were enrolled. Of those, 230 (90%) had at least one high-quality funduscopic photograph. Retinal photographs were normal in 161 (70%). Out of the 69 abnormal photographs, only 7% of abnormalities were detected by the residents. Ninety-three percent of residents found the retinal photographs useful. Nonmydriatic ocular fundus photography improved the care in patients presenting to a Neurology resident clinic and facilitated residents in recognizing funduscopic findings. Its benefits are clear when one considers (1) the high risk of negative patient outcomes and possible medicolegal consequences due to missed findings, (2) the ease of incorporating retinal photographs into the patients' medical records, and (3) the benefit of improving resident education in regard to the ophthalmologic examination.

Multimodal Imaging-Based Phenotyping of a Singaporean Hospital-Based Cohort of High Myopia Patients

Purpose: To assess the effect of axial length (AL) on the prevalence of pathologic myopia (PM) and associated myopic features in a Singaporean hospital-based cohort of patient with high myopia (HM).Methods: In total, 923 HM eyes from 495 individuals were recruited from the Myopic and Pathologic Eyes in Singapore (MyoPES) cohort and underwent ocular biometry, fundus photography, fundus autofluorescence, and swept-source optical coherence tomography (SS-OCT). Images were analyzed for the presence of myopic macular degeneration (MMD), myopic choroidal neovascularization (mCNV), myopic traction maculopathy (MTM), peripapillary atrophy (PPA), myopic tilted disc, posterior staphyloma (PS), dome-shaped macula (DSM), vitremacular adhesions (VMA), and the epiretinal membrane (ERM). Eyes were stratified into quartiles based on ALs to determine cut-off values to perform comparisons between shorter-length and longer-length groups. A χ2-test was done to determine the difference in the prevalence of pathologies between groups.Results: Overall, mean AL was 29.2 ± 2.2 mm (range 25.0–36.7 mm). Myopic macular degeneration, PPA, myopic tilted disc, and ERM have AL threshold of ≥27.5 mm, whereas MTM has an AL threshold of ≥29.0 mm. We found that there was a significantly higher prevalence of MMD (88.2 vs. 49.4%; p < 0.001), PPA (98.1 vs. 80.1%; p < 0.001), myopic tilted disc (72.7 vs. 50.2%; p < 0.001), and ERM (81.4 vs. 17.3%; p = 0.003) in eyes with AL ≥ 27.5 mm vs. eyes without AL <27.5 mm. Prevalence of MTM (34.7 vs. 32.1%; p < 0.001), mCNV (17.4 vs. 12.1%; p = 0.03), PS (43.4 vs. 34.7%; p = 0.012), DSM (21.3 vs. 13.2%; p = 0.002), and VMA (5.9 vs. 2.6%; p = 0.014) in eyes with AL ≥ 29.0 mm compared with AL < 29.0 mm.Conclusion: Our study describes the overall prevalence of PM and related pathologies among patients with HM in our hospital-based cohort. Longer eyes even among HM eyes had a significantly higher prevalence of PM-associated pathologies studied. This supports the premise that eyes with longer AL, even among HM eyes may be at greater risk of vision-threatening changes and therefore merit regular follow-up.