Generalizability of Deep Neural Networks for Vertical Cup-to-Disc Ratio Estimation in Ultra-Widefield and Smartphone-Based Fundus Images.

Purpose: To develop and validate a deep learning system (DLS) for estimation of vertical cup-to-disc ratio (vCDR) in ultra-widefield (UWF) and smartphone-based fundus images. Methods: A DLS consisting of two sequential convolutional neural networks (CNNs) to delineate optic disc (OD) and optic cup (OC) boundaries was developed using 800 standard fundus images from the public REFUGE data set. The CNNs were tested on 400 test images from the REFUGE data set and 296 UWF and 300 smartphone-based images from a teleophthalmology clinic. vCDRs derived from the delineated OD/OC boundaries were compared with optometrists' annotations using mean absolute error (MAE). Subgroup analysis was conducted to study the impact of peripapillary atrophy (PPA), and correlation study was performed to investigate potential correlations between sectoral CDR (sCDR) and retinal nerve fiber layer (RNFL) thickness. Results: The system achieved MAEs of 0.040 (95% CI, 0.037-0.043) in the REFUGE test images, 0.068 (95% CI, 0.061-0.075) in the UWF images, and 0.084 (95% CI, 0.075-0.092) in the smartphone-based images. There was no statistical significance in differences between PPA and non-PPA images. Weak correlation (r = -0.4046, P < 0.05) between sCDR and RNFL thickness was found only in the superior sector. Conclusions: We developed a deep learning system that estimates vCDR from standard, UWF, and smartphone-based images. We also described anatomic peripapillary adversarial lesion and its potential impact on OD/OC delineation. Translational Relevance: Artificial intelligence can estimate vCDR from different types of fundus images and may be used as a general and interpretable screening tool to improve community reach for diagnosis and management of glaucoma.

PREDICTORS AND IMPORTANCE OF COMPLETE POLYPOIDAL LESION REGRESSION IN THE EVEREST II STUDY: Predictors of Polyp Regression in Polypoidal Choroidal Vasculopathy.

PURPOSE: To evaluate the predictors of complete polypoidal lesion regression (CPREG) in polypoidal choroidal vasculopathy. METHODS: Post hoc analysis of EVEREST II-a 24-month, multicenter, randomized, controlled clinical trial of 322 patients with polypoidal choroidal vasculopathy, randomized to receive ranibizumab with or without photodynamic therapy. Images of indocyanine green angiography (ICGA) were graded by a central reading center. Multiple logistic regression analysis with significant baseline predictors then was conducted to assess adjusted odds ratios for CPREG at month (M) 12. RESULTS: Baseline ICGA characteristics were comparable between the treatment groups. Patients treated with combination therapy had higher odds of achieving CPREG at M12 (adjusted odds ratio = 4.64; 95% confidence interval, 2.85-7.55; P < 0.001) compared with those in the monotherapy group. Absence of polypoidal lesion pulsation on ICGA was also associated with CPREG at M12 (adjusted odds ratio = 2.62; 95% confidence interval, 1.32-5.21; P = 0.006). The presence of CPREG at M3 had higher odds of maintaining CPREG at M12 (adjusted odds ratio = 6.60; 95% confidence interval, 3.77-11.57; P < 0.001) compared with those with persistent polypoidal lesions. CONCLUSION: At M12, treatment with combination therapy was associated with higher probability of achieving CPREG than with ranibizumab monotherapy. The results contribute to the further understanding of the response of polypoidal lesions to treatment.

Comparing efficacy of reduced-fluence and standard-fluence photodynamic therapy in the treatment of polypoidal choroidal vasculopathy.

BACKGROUND: The EVEREST II study reported superior polyp closure rates and visual outcomes using combination standard photodynamic therapy (PDT) with intravitreal ranibizumab in the treatment of polypoidal choroidal vasculopathy (PCV). The optimal PDT protocol remains controversial and it is postulated that less intensive PDT strategies may reduce complications. We aimed to compare the efficacy of reduced and standard-fluence PDT. METHODS: Case-control review of 38 consecutive PDT-naïve macular PCV patients who underwent verteporfin PDT using one of two PDT regimens at a tertiary referral centre in an Asian population. Comparison of outcomes between standard-fluence PDT (light dose, 50 J/cm2; dose rate, 600 mW/cm2; wavelength, 689 nm PDT applied to the treatment eye for 83 s) and reduced-fluence PDT (light dose, 25 J/cm2; dose rate, 600 mW/cm2; wavelength, 689 nm PDT applied to the treatment eye for 42 s). Primary outcome measure was best corrected LogMAR visual acuity (VA). Secondary outcome measures included OCT measurements such as central retinal thickness (CRT), height of subfoveal sub-retinal fluid (SRF), central choroid thickness (CCT), mean number of PDT treatments needed, mean number of anti-VEGF injections needed, polyp closure and recurrence rates. RESULTS: Of these 38 eyes of 38 patients, an equal number of eyes (19 in each arm) were treated with standard-fluence and reduced-fluence PDT. Mean letter gain at 12 months for the standard-fluence group was 6.0 compared to 4.3 letters for the reduced-fluence group (p = 0.61). Similar results were observed at all time points. There was no statistically significant difference between the retinal and choroidal anatomical OCT outcomes, rates of polyp closure and recurrences between the two PDT regimens. CONCLUSIONS: Reduced-fluence PDT was comparable to standard-fluence PDT in the treatment of PCV in terms of visual gains, clinical and anatomical OCT outcomes.

EVEREST Report 5: Clinical Outcomes and Treatment Response of Polypoidal Choroidal Vasculopathy Subtypes in a Multicenter, Randomized Controlled Trial.

Purpose: The purpose of this study was to describe the characteristics of polypoidal choroidal vasculopathy (PCV) subtypes among patients from a multicenter randomized controlled trial and to determine the impact of PCV subtypes on clinical outcomes. Methods: This was a prospective cohort study of 61 patients with macular PCV from the EVEREST study. Indocyanine green (ICGA) and fluorescein angiography (FA) obtained using standardized imaging protocols were graded to classify PCV into three subtypes. Type A PCV had polyps with interconnecting channels, type B had polyps with branching vascular networks, but no significant leakage on FA, and type C had polyps with branching vascular networks and leakage on FA. The best-corrected visual acuity (BCVA) and proportion of patients with BCVA ≥ 20/40 were compared among the three PCV subtypes. Results: Of the 61 patients, 54 were gradable for PCV subtype. Among these, 8 had type A PCV (14.8%), 27 had type B (50%), and 19 had type C (35.2%). At baseline, BCVA was 67.1 letters for type A, 58.7 for type B, and 43.5 for type C (P < 0.001). At 6 months, BCVA was highest among patients with type A compared with types B and C (80.1 letters versus 67.2 versus 50.4, respectively; P < 0.001). Type A PCV gained 13 letters compared with 8.5 (type B) and 6.9 (type C). BCVA ≥ 20/40 was highest for type A compared with types B and C (100% vs. 51.9% vs. 10.5%; P < 0.001). On performing ANCOVA, PCV subtype and baseline BCVA significantly affected final BCVA. Conclusions: The visual outcome following treatment varies with PCV subtype classification. The distinction in clinical outcomes between the PCV subtypes is observed in the initial months following the start of treatment.

HDL-cholesterol levels and risk of age-related macular degeneration: a multiethnic genetic study using Mendelian randomization.

Background: Dyslipidemia, particularly high-density lipoprotein cholesterol (HDL-C), has recently been implicated in the pathogenesis of age-related macular degeneration (AMD), the leading cause of vision loss. However, epidemiological studies have yielded conflicting results. Methods: We investigated the causal role of plasma lipid levels in AMD in multiethnic populations comprising 16 144 advanced AMD cases and 17 832 controls of European descent, together with 2219 cases and 5275 controls of Asian descent, using Mendelian randomization in three models. Model 1 is a conventional meta-analysis which does not account for pleiotropy of instrumental variable (IV) effects. Model 2 is a univariate, inverse variance weighted regression analysis that accounts for potential unbalanced pleiotropy using MR-Egger method. Finally, Model 3 is a multivariate regression analysis that addresses pleiotropy by MR-Egger method and by adjusting for effects on other lipid traits. Results: A 1 standard deviation (SD) higher HDL-cholesterol level was associated with an odds ratio (OR) for AMD of 1.17 (95% confidence interval: 1.07-1.29) in Europeans (P = 6.88 × 10-4) and of 1.58 (1.24-2.00) in Asians (P = 2.92 × 10-4) in Model 3. The corresponding OR estimates were 1.30 (1.09-1.55) in Europeans (P = 3.18 × 10-3) and 1.42 (1.11-1.80) in Asians (P = 4.42 × 10-3) in Model 1, and 1.21 (1.11-1.31) in Europeans (P = 3.12 × 10-5) and 1.51 (1.20-1.91) in Asians (P = 7.61 × 10-4) in Model 2. Conversely, neither LDL-C (Europeans: OR = 0.96, P = 0.272; Asians: OR = 1.02, P = 0.874; Model 3) nor triglyceride levels (Europeans: OR = 0.91, P = 0.102; Asians: OR = 1.06, P = 0.613) were associated with AMD. We also assessed the association between lipid levels and polypoidal choroidal vasculopathy (PCV) in Asians, a subtype of AMD, and found a similar trend for association of PCV with HDL-C levels. Conclusions: Our study shows that high levels of plasma HDL-C are causally associated with an increased risk for advanced AMD in European and Asian populations, implying that strategies reducing HDL-C levels may be useful to prevent and treat AMD.

Shared genetic variants for polypoidal choroidal vasculopathy and typical neovascular age-related macular degeneration in East Asians.

Polypoidal choroidal vasculopathy (PCV), a subtype of age-related macular degeneration (AMD) more frequently seen in East Asians, has both common and distinct clinical manifestations with typical neovascular AMD (tAMD). We aim to examine the extent to which common genetic variants are shared between these two subtypes. We performed the meta-analysis of association in a total of 1062 PCV patients, 1157 tAMD patients and 5275 controls of East Asian descent from the Genetics of AMD in Asians Consortium at the 34 known AMD loci. A total of eight loci were significantly associated with PCV, including age-related maculopathy susceptibility 2 (ARMS2)-HtrA serine peptidase 1 (HTRA1), complement factor H (CFH), C2-CFB-SKIV2L, CETP, VEGFA, ADAMTS9-AS2 and TGFBR1 (P<5 × 10-4) from the single-nucleotide polymorphism-based test and COL4A3 from the gene-based tests (Pgene=2.02 × 10-4). PCV and tAMD are genetically highly correlated (rg=0.69, P=4.68 × 10-3), with AMD known loci accounting for up to 36% variation. Weaker association for PCV was observed at ARMS2-HTRA1 (Pdif=4.39 × 10-4) and KMT2E-SRPK2(Pdif=4.43 × 10-3), compared with tAMD. Variants at CFH, CETP and VEGFA exhibited different association signals in East Asians, in contrast to those in European individuals. Our data suggest a substantially shared genetic susceptibility for PCV and tAMD, while also highlight the unique associations for PCV, which is useful in understanding the pathogenesis of PCV.

Systemic, Ocular and Genetic Risk Factors for Age-related Macular Degeneration and Polypoidal Choroidal Vasculopathy in Singaporeans.

To examine the association of systemic, ocular and genetic risk factors in neovascular age-related macular degeneration (nAMD) in a large cohort of Asian patients, and to further compare risk factors between those with typical AMD and polypoidal choroidal vasculoapthy (PCV) subtypes. We recruited 456 cases and 1,824 controls matched for age, gender and ethnicity. Data on systemic and ocular risk factors were collected on questionnaires. In a subgroup of subjects, we included genetic data on four AMD-associated single nucleotide polymorphisms (SNPs). Risk factors for nAMD and subtypes were analyzed. Systemic risk factors for nAMD included older age, male gender, higher BMI and higher HDL-cholesterol. Ocular risk factors included pseudophakic and shorter axial length. Risk factors common to both typical AMD and PCV subtypes included age, BMI and HDL-cholesterol. Shorter axial length was only associated with PCV, while male gender and pseudophakia were only associated with typical AMD. In the subgroup with genotype data, ARMS2 rs10490924 and CFH rs800292 were associated with nAMD. None of the risk factors were significantly different between PCV and typical AMD. Systemic, ocular and genetic risk factors were largely similar for typical AMD and PCV subtypes in this Asian population based in Singapore.

Fundus Photography in the 21st Century--A Review of Recent Technological Advances and Their Implications for Worldwide Healthcare.

BACKGROUND: The introduction of fundus photography has impacted retinal imaging and retinal screening programs significantly. LITERATURE REVIEW: Fundus cameras play a vital role in addressing the cause of preventive blindness. More attention is being turned to developing countries, where infrastructure and access to healthcare are limited. One of the major limitations for tele-ophthalmology is restricted access to the office-based fundus camera. RESULTS: Recent advances in access to telecommunications coupled with introduction of portable cameras and smartphone-based fundus imaging systems have resulted in an exponential surge in available technologies for portable fundus photography. Retinal cameras in the near future would have to cater to these needs by featuring a low-cost, portable design with automated controls and digitalized images with Web-based transfer. CONCLUSIONS: In this review, we aim to highlight the advances of fundus photography for retinal screening as well as discuss the advantages, disadvantages, and implications of the various technologies that are currently available.

IMPROVED SPECIFICITY OF POLYPOIDAL CHOROIDAL VASCULOPATHY DIAGNOSIS USING A MODIFIED EVEREST CRITERIA.

PURPOSE: To evaluate the performance of polypoidal choroidal vasculopathy (PCV) diagnosis using fundus camera-based indocyanine green angiography, comparing a single sign of "subretinal focal hyperfluorescence" on indocyanine green angiography with a modification of the EVEREST criteria. METHODS: Color fundus photograph, flash fundus camera-based fluorescein angiography, and indocyanine green angiography of 241 eyes of 230 consecutive patients with exudative maculopathy due to PCV or typical age-related macular degeneration were graded independently by 2 retinal specialists using a modified EVEREST criteria, which requires the presence of subretinal focal hyperfluorescence plus any 1 of 5 additional criteria. Discordant cases were adjudicated by a senior retinal specialist to arrive at the final diagnosis. Sensitivity, specificity, and area under the receiver operating curve of subretinal focal hyperfluorescence versus the EVEREST criteria and combinations of individual EVEREST criteria were compared. RESULTS: Among the 241 eyes with exudative maculopathy, 131 eyes had PCV and 110 eyes had typical age-related macular degeneration. Using a single sign of subretinal focal hyperfluorescence alone for the diagnosis of PCV, sensitivity was 85.3% and specificity was 80.9%, with an area under the receiver operating curve of 83.1%. When applying the EVEREST definition, sensitivity was reduced to 78.4% but specificity improved to 87.1% with a similar area under the receiver operating curve of 82.8%. The frequency of individual criteria was highly variable, with stereo nodular appearance (73.7%) and orange nodule (55.0%) being the most common and branching vascular network, massive hemorrhage, and hypofluorescent halo in the presence of subretinal focal hyperfluorescence being less common (21.5%-28.1%). CONCLUSION: The EVEREST criteria have a higher specificity for the diagnosis of PCV than subretinal focal hyperfluorescence alone and may be applied to flash fundus camera-based indocyanine green angiography in a clinical setting. Stereo nodular appearance is the most important additional criterion.

Automatic Segmentation of Polypoidal Choroidal Vasculopathy from Indocyanine Green Angiography Using Spatial and Temporal Patterns.

PURPOSE: To develop a computer-aided diagnostic tool for automated detection and quantification of polypoidal regions in indocyanine green angiography (ICGA) images. METHODS: The ICGA sequences of 59 polypoidal choroidal vasculopathy (PCV) treatment-naïve patients from five Asian countries (Hong Kong, Singapore, South Korea, Taiwan, and Thailand) were provided by the EVEREST study. The ground truth was provided by the reading center for the presence of polypoidal regions. The proposed detection algorithm used both temporal and spatial features to characterize the severity of polypoidal lesions in ICGA sequences. Leave-one-out cross validation was carried out so that each patient was used once as the validation sample. For each patient, a fixed detection threshold of 0.5 on the severity was applied to obtain sensitivity, specificity, and balanced accuracy with respect to the ground truth. RESULTS: Our system achieved an average accuracy of 0.9126 (sensitivity = 0.9125, specificity = 0.9127) for detection of polyps in the 59 ICGA sequences. Among the total of 222 features extracted from ICGA sequence, the spatial variances exhibited best discriminative power in distinguishing between polyp and nonpolyp regions. The results also indicated the importance of combining spatial and temporal features to further improve detection accuracy. CONCLUSIONS: The developed software provided a means of detecting and quantifying polypoidal regions in ICGA images for the first time. TRANSLATIONAL RELEVANCE: This preliminary study demonstrated a computer-aided diagnostic tool, which enables objective evaluation of PCV and its progression. Ophthalmologists can easily visualize the polypoidal regions and obtain quantitative information about polyps by using the proposed system.

EVEREST study report 2: imaging and grading protocol, and baseline characteristics of a randomised controlled trial of polypoidal choroidal vasculopathy.

PURPOSE: To describe the imaging standards, grading protocol and baseline characteristics of polypoidal choroidal vasculopathy (PCV) from the EVEREST study. METHODS: In a prospective, multicentre study, confocal scanning laser ophthalmoscope indocyanine green angiography (ICGA) was performed using a standardised imaging protocol. All images were graded using standardised, calibrated equipment by fellowship-trained ophthalmologists at the Central Reading Center. RESULTS: Sixty-one patients with PCV were included in the study. ICGA characteristics included: nodular appearance stereoscopically (56 eyes, 91.8%), hypofluorescent halo (42, 68.9%), abnormal vascular network (54, 88.5%) and pulsation of the polyps (4, 6.6%). Colour fundus photography revealed orange subretinal nodules (34, 55.7%) and massive submacular haemorrhage (8, 13.1%). The mean area of the PCV lesion was 3.11 mm(2) (range, 0.2-10.7 mm(2)). The vascular channels filled within 7.3-32.0 s (mean: 17.9 s) while the mean filling time for polyps was 21.9 s (range, 7.3-40.4 s). Patients with massive submacular haemorrhage were less likely to have abnormal vascular channels seen on ICGA (28.6% vs 83.3% for those without massive haemorrhage, p=0.001). CONCLUSIONS: The imaging and grading protocols and baseline characteristics of a multicentre, randomised controlled trial of PCV are described in detail, and may serve as reference for future randomised, controlled trials on PCV. CLINICAL TRIAL NUMBER: This work was supported by Novartis Pharma AG, Basel, Switzerland grant number NCT00674323 (clinicaltrials.gov).

A novel classification of the vascular patterns of polypoidal choroidal vasculopathy and its relation to clinical outcomes.

PURPOSE: To propose a novel classification system for polypoidal choroidal vasculopathy (PCV), and compare the clinical outcomes among PCV subtypes. METHODS: Consecutive treatment-naive patients with symptomatic PCV were managed over 5 years. PCV subtypes were classified based on indocyanine green angiography (ICGA) and fluorescein angiography (FA) characteristics. RESULTS: Among 107 patients, 3 PCV subtypes were seen: Type A (interconnecting channels on ICGA) -22.4%; Type B (branching vascular network with no leakage) -24.3%; Type C (branching vascular network with late leakage on FA) -53.3%. The proportion of patients with best-corrected visual acuity (BCVA) ≥20/40 was highest in Type A, intermediate in Type B and lowest in Type C at all time points (80% vs 66.7% vs 7.7% at 5 years, p<0.001). The highest rate of moderate visual loss (loss of ≥3 lines) occurred in Type C PCV (57.7% vs 0% for Types B and A at 5 years, p<0.001). Risk factors for poor visual outcomes were PCV subtype (OR 53.7, p<0.001 for Type C and OR 13.7, p=0.023 for Type B compared to Type A) and age (OR 1.06, 95% CI 1.002 to 1.125, p=0.044). CONCLUSIONS: The PCV subtype seen on initial presentation affects the long-term visual outcomes over a 5-year period.

Retinal vascular parameter variations in patients with human immunodeficiency virus.

PURPOSE: To compare the retinal vascular parameters in patients with human immunodeficiency virus (HIV) infection with normal controls, and to determine the relationship between retinal vascular parameters and HIV-related blood biomarkers (CD4(+) T-lymphocytes count, presence of HIV RNA). METHODS: Case-control study of eighty-five patients with HIV on follow-up at the Communicable Disease Center, Singapore, and 251 age-, sex-, and race-matched normal healthy controls (case: control matching ≈ 1:3) selected from the Singapore Epidemiology of Eye Disease program were included in this study. Standardized retinal photographs were taken from patients and controls. Trained technicians measured quantitative retinal vascular parameters (retinal vascular caliber, branching angle, tortuosity, and fractal dimension) with a semiautomated computer-based program following a standardized protocol. RESULTS: HIV-patients had more tortuous arterioles (0.77 × [10(4)] vs. 0.59 × [10(4)], P < 0.001) and venules (0.90 × [10(4)] vs. 0.74 × [10(4)], P < 0.001), compared with healthy normal subjects. Amongst the HIV-patients, increasing HIV viral loads were associated with decreased retinal arteriolar caliber (P trend = 0.009) and decreased arteriolar-venular ratio (P trend = 0.025). CONCLUSIONS: Our study showed that patients with HIV have significant variations in retinal vasculature. Retinal vascular imaging may offer further insight into the pathophysiology behind HIV-related vascular disease in future.

A pilot trial of tele-ophthalmology for diagnosis of chronic blurred vision.

We evaluated the accuracy of tele-ophthalmology in diagnosing the major causes of chronic blurring of vision. Thirty consecutive patients attending a primary eye-care facility in Singapore (the Ang Mo Kio Polyclinic, AMKP) with the symptom of chronic blurred vision were recruited. An ophthalmic technician was trained to perform Snellen acuity; auto-refraction; intraocular pressure measurement; red-colour perimetry; video recordings of extraocular movement, cover tests and pupillary reactions; and anterior segment and fundus photography. Digital information was transmitted to a tertiary hospital in Singapore (the Tan Tock Seng Hospital) via a tele-ophthalmology system for teleconsultation with an ophthalmologist. The diagnoses were compared with face-to-face consultation by another ophthalmologist at the AMKP. A user experience questionnaire was administered at the end of the consultation. Using face-to-face consultation as the gold standard, tele-ophthalmology achieved 100% sensitivity and specificity in diagnosing media opacity (n = 29), maculopathy (n = 23) and keratopathy (n = 30) of any type; and 100% sensitivity and 92% specificity in diagnosing optic neuropathy of any type (n = 24). The majority of the patients (97%) were satisfied with the tele-ophthalmology workflow and consultation. The tele-ophthalmology system was able to detect causes of chronic blurred vision accurately. It has the potential to deliver high-accuracy diagnostic eye support to remote areas if suitably trained ophthalmic technicians are available.

Asian age-related macular degeneration phenotyping study: rationale, design and protocol of a prospective cohort study.

BACKGROUND: Current knowledge of the phenotypic characteristics (e.g. clinical features, risk factors, natural history and treatment response) of age-related macular degeneration (AMD) in Asians remains limited. This report summarizes the rationale and study design of a prospective observational study of Asian neovascular AMD, including polypoidal choroidovasculopathy variant. DESIGN: The Asian AMD phenotyping study is a prospective, observational clinical study of Asian patients with neovascular AMD or polypoidal choroidovasculopathy in three tertiary eye centres in Singapore. PARTICIPANTS: The study aims to recruit 500 consecutive patients from the retinal clinics of three tertiary eye centres in Singapore. METHODS: Standardized examination procedures include interviews, a comprehensive eye examination, digital photography of the retina, fundus fluorescein and indocyanine green angiography and spectral domain optical coherence tomography using a standardized protocol. Blood samples were collected for biochemical analyses and stored for genetic and proteomic studies. MAIN OUTCOME MEASURES: The aim of the study was to build a comprehensive database of clinical, angiographic, functional and natural history data of Asian AMD over a 12-month follow-up period. RESULTS: This article discusses the methodology and design of this prospective multi-centred study. CONCLUSION: This study will provide in-depth longitudinal data of the evolution of clinical features, risk factors, natural history and treatment pattern and response of Asian AMD and polypoidal choroidovasculopathy, allowing unique insights into pathogenesis and the design of new treatment strategies.