Search

Abstract Central serous chorioretinopathy (CSC) is the fourth most common retinopathy and can reduce quality of life. CSC is assessed using optical coherence tomography (OCT), but deep learning systems have not been used to classify CSC subtypes. This study aimed to build a deep learning system model to distinguish CSC subtypes using a convolutional neural network (CNN). We enrolled 435 patients with CSC from a single tertiary center between January 2015 and January 2020. Data from spectral domain OCT (SD-OCT) images of the patients were analyzed using a deep CNN. Five-fold cross-validation was employed to evaluate the model’s ability to discriminate acute, non-resolving, inactive, and chronic atrophic CSC. We compared the performances of the proposed model, Resnet-50, Inception-V3, and eight ophthalmologists. Overall, 3209 SD-OCT images were included. The proposed model showed an average cross-validation accuracy of 70.0% (95% confidence interval [CI], 0.676–0.718) and the highest test accuracy was 73.5%. Additional evaluation in an independent set of 104 patients demonstrated the reliable performance of the proposed model (accuracy: 76.8%). Our model could classify CSC subtypes with high accuracy. Thus, automated deep learning systems could be useful in the classification and management of CSC.

Abstract To describe the eyes with vitreous floaters and to analyze the development of acute symptomatic posterior vitreous detachment (PVD). A retrospective review of medical records was performed on patients with the vitreous floater developed for the first time of their life. Peripapillary vitreous opacity (pVO) was searched in Ultra-wide field (UWF) scanning laser ophthalmoscopy and PVD stage was assessed through spectral-domain optical coherence tomography (SD-OCT). 196 patients (55 males and 141 females), who were 58.4 (± 9.1) years old, visited a retinal clinic 9.4 (± 9.1) days after they experienced vitreous floaters. In 196 eyes, pVO was noticed in 122 eyes (62.2%) at UWF. In 106 eyes where SD-OCT data were available, PVD was noticed in 100 eyes (94.3%). Symptomatic eyes showed more advanced stage of PVD (p

Abstract This cross-sectional study aimed to build a deep learning model for detecting neovascular age-related macular degeneration (AMD) and to distinguish retinal angiomatous proliferation (RAP) from polypoidal choroidal vasculopathy (PCV) using a convolutional neural network (CNN). Patients from a single tertiary center were enrolled from January 2014 to January 2020. Spectral-domain optical coherence tomography (SD-OCT) images of patients with RAP or PCV and a control group were analyzed with a deep CNN. Sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUROC) were used to evaluate the model’s ability to distinguish RAP from PCV. The performances of the new model, the VGG-16, Resnet-50, Inception, and eight ophthalmologists were compared. A total of 3951 SD-OCT images from 314 participants (229 AMD, 85 normal controls) were analyzed. In distinguishing the PCV and RAP cases, the proposed model showed an accuracy, sensitivity, and specificity of 89.1%, 89.4%, and 88.8%, respectively, with an AUROC of 95.3% (95% CI 0.727–0.852). The proposed model showed better diagnostic performance than VGG-16, Resnet-50, and Inception-V3 and comparable performance with the eight ophthalmologists. The novel model performed well when distinguishing between PCV and RAP. Thus, automated deep learning systems may support ophthalmologists in distinguishing RAP from PCV.

Abstract This study aimed to compare the surgical outcomes of pars plana vitrectomy (PPV) with and without air tamponade in patients with idiopathic epiretinal membrane (iERM). We prospectively enrolled 145 patients with iERM who underwent a 25-gauge transconjunctival sutureless PPV. Patients were assigned to either the air tamponade (air) group (79 eyes) or balanced salt solution (BSS; no tamponade) group (66 eyes). The central macular thickness (CMT), peripapillary retinal nerve fiber layer (pRNFL) thickness, and best-corrected visual acuity (BCVA) were compared for two years. At baseline, there were no significant differences between the two groups. CMT and BCVA were not significantly different between the groups for 2 years. However, the air group had a significantly lower thickness in the superior temporal pRNFL sector at 1 month (p = 0.01) and in the inferior temporal and superior temporal pRNFL sectors at 3 months (p = 0.02 for both). There were no significant differences between both groups in all the pRNFL sectors from 6 months to 2 years. The outcomes of PPV with air tamponade and that with no tamponade appear to be equivalent. This shows that air tamponade may not be an imperative procedure for iERM surgery and has no additional benefit.

Abstract The purpose is to evaluate the effects of multiple intravitreal ranibizumab (IVR) and aflibercept (IVA) injections on peripapillary retinal nerve fiber layer (RNFL) thickness in patients with exudative age-related macular degeneration (AMD). This retrospective, observational, consecutive case series study enrolled patients newly diagnosed with monocular exudative AMD from January 2014 to October 2019 who were administered IVR or IVA injections. Normal fellow eyes were included as controls. Medical records and spectral domain optical coherence tomography results were reviewed at baseline and at 3, 6, and 12 months after injection. No statistically significant differences in peripapillary RNFL thickness and intraocular pressure were observed between the treated and fellow eyes in the two groups. The global RNFL thicknesses for the treated eyes decreased significantly after 12 months compared with baseline, but no significant difference was observed in any of the six examined sectors (temporal, superior temporal, superior nasal, nasal, inferior nasal, and inferior temporal). At 12 months, the central macular thickness of the treated eyes decreased significantly. Multiple IVR and IVA injections are apparently safe considering peripapillary RNFL damage in patients with exudative AMD. The decreased RNFL thickness of the global sector was presumably due to anatomical improvement of macular lesions.

Background/aims Evaluation of the retinal nerve fibre layer (RNFL) is important for identifying glaucomatous damage. Ultrawide-field fundus photography (UWP) imaging is increasingly used in the ophthalmological field; however, it is unknown whether it can be used for detecting RNFL defects (RNFLDs). We investigated whether RNFLD can be detected with UWP images and compared the clinical effectiveness of three types of images for detecting RNFLD: conventional red-free RNFL photography (RFP), non-mydriatic UWP and digitally converted green separation of non-mydriatic UWP (G-UWP). Methods Eyes with glaucoma or glaucoma suspect and normal control eyes meeting the eligibility criteria were consecutively enrolled from September 2019 to April 2020. Their conventional RFP, non-mydriatic UWP and G-UWP images were assessed for detecting RNFLD to evaluate the sensitivity and specificity for detecting RNFLD. Results Three image sets of 196 participants (84 normal control, 25 glaucoma suspect and 87 glaucoma) were obtained. The sensitivity of G-UWP (94.6%; 95% CI 88.7 to 98.0) and RFP (92.9%; 95% CI 86.4 to 96.9) was higher than that of UWP (82.1%; 95% CI 73.8 to 88.7; p<0.05). The sensitivities of G-UWP and RFP are comparable. The specificity of G-UWP (78.6%; 95% CI 68.3 to 86.8) and UWP (75.0%; 95% CI 64.4 to 83.8) was comparable, but both were lower than that of RFP (98.8%; 95% CI 93.5 to 100.0; p<0.05). Conclusion Non-mydriatic UWP images can be used to detect RNFLD. Non-mydriatic G-UWP showed comparable sensitivity but lower specificity to conventional RFP. Non-mydriatic G-UWP could be used as a convenient and useful diagnostic tool for screening glaucoma in clinical settings. [ABSTRACT FROM AUTHOR]

Objective: To compare the effectiveness of the Quickert suture (Q-suture) with that of the lateral tarsoligamentous sling in treating lower eyelid involutional entropion in Asians., Method: We reviewed the charts of all patients who underwent repair of an involutional lower eyelid entropion in Hangil Eye Hospital between September 2011 and March 2016. A single surgeon (S. C. Kim) performed the procedure in every case. Patients underwent either Q-suturing or lateral tarsoligamentous sling. We analyzed the results using Mann-Whitney and Chi-square tests, as well as Kaplan-Meier survival analysis. A p-value of <0.05 was considered to indicate significance., Results: A total of 50 patients (61 eyes) were included in the study. Of the 50 patients, 24 (30 eyelids) received Q-suture and 26 patients (31 eyelids) underwent lateral tarsoligamentous sling. In all patients who underwent lateral tarsoligamentous sling, the entropion was successfully corrected and did not recur. Furthermore, none of the patients experienced any significant postoperative complications. In contrast, entropion recurred in 16 (36.67%) of the patients who underwent Q-suture (mean of 10.88 ± 9.02 months after surgery). The recurrence rate was not significantly correlated with age (p = 0.093); it was higher in men than in women, although this was not a significant difference (46.67% vs 26.6.7%; p = 0.175 by Chi-squared test). Kaplan-Meier survival analysis showed that the recurrence rate after Q-suture differed significantly from that after lateral tarsoligamentous sling (36.67% vs 0%, respectively; p = 0.007)., Conclusion: This study suggests that lateral tarsoligamentous sling is an effective surgical method for correcting lower eyelid involutional entropion; the technique is not associated with recurrence or significant complications in Asians., (Copyright © 2017 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.)