Categorization of Meibomian Gland Dysfunction Using Lipid Layer Thickness and Meibomian Gland Dropout in Dry Eye Patients: A Retrospective Study.

PURPOSE: In the present study, we determined the prevalence of obstructive meibomian gland dysfunction (MGD), hyposecretory MGD, grossly normal MG, and hypersecretory MGD in patients with dry eye syndrome using lipid layer thickness (LLT) and MG dropout. METHODS: Eighty-eight patients with dry eye syndrome were included in the study. Patients were categorized into four groups according to the LLT and weighted total meiboscore. The proportion of patients in each group was calculated. The age, sex, Ocular Surface Disease Index, LLT, Schirmer, tear film breakup time, cornea stain, weighted total meiboscore, expressibility, and quality of meibum were compared between the four groups. RESULTS: Fifteen eyes (17.0%) had obstructive MGD, two eyes (2.3%) had hyposecretory MGD, 40 eyes (45.5%) had grossly normal MG, and 17 eyes (19.3%) had hypersecretory MGD. The obstructive MGD group was younger than the grossly normal MG group. In obstructive MGD, the ratio of men to women was higher than that of the other groups. However, Ocular Surface Disease Index, Schirmer, tear film breakup time, and corneal stain did not show statistically significant differences between the four groups. The meibum expressibility of the hyposecretoy MGD group was worse than those of the other groups. The meibum expressibility of the hyposecretoy MGD group was poor than those of the obstructive and hypersecretory MGD group. CONCLUSIONS: This categorization was expected to help determine the best treatment method for dry eye syndrome, according to the MG status.

QUANTITATIVE ANALYSIS OF CHOROIDAL MORPHOLOGY USING MULTIMODAL IMAGING IN ACUTE AND PERSISTENT CENTRAL SEROUS CHORIORETINOPATHY.

PURPOSE: To analyze quantitative differences in choroidal morphology between acute and persistent central serous chorioretinopathy using multimodal images. METHODS: Ultra-widefield indocyanine green angiography (UWICGA) and optical coherence tomography images of 72 eyes of 72 patients with acute (32 eyes) and persistent (40 eyes) central serous chorioretinopathy were collected. Choroidal thickness, area, vessel density, symmetry, and intervortex anastomosis were assessed. RESULTS: The choroidal area on optical coherence tomography B-scan images was smaller and the choroidal vessel density on UWICGA images was lower in the persistent group ( P < 0.001 and P = 0.028, respectively). Choroidal vessel density on UWICGA showed positive correlation with that of vortex ampullae (all P ≤ 0.046). The constitution of the intervortex anastomosis and dominant vessels in the macular area showed differences between the groups ( P = 0.014 and P = 0.010, respectively), with greater inferonasal vessel participation in the anastomosis and combined superotemporal and inferotemporal vessels as dominant vessels in the persistent groups. CONCLUSION: Acute and persistent central serous chorioretinopathy differed in subfoveal choroidal area, choroidal vessel density, and intervortex anastomosis constitution on UWICGA images. Choroidal vessel density on UWICGA images correlated with that of vortex ampullae. These findings enhance our understanding of the pathophysiology of central serous chorioretinopathy subtypes.

Choroidal Morphology on Ultra-Widefield Indocyanine Green Angiography and Response to Aflibercept in Pachychoroid Neovasculopathy.

Purpose: This study assessed the relationship between the choroidal morphology and short-term response to aflibercept treatment in pachychoroid neovasculopathy (PNV). Methods: This was a retrospective case-control study. Ultra-widefield indocyanine green angiography (UWICGA) and optical coherence tomography (OCT) images of 90 PNV eyes of 90 patients treated with aflibercept were enrolled. Responsiveness to aflibercept was defined as a complete resolution of sub- or intra-retinal fluid after three loading doses (50 dry and 40 non-dry eyes). Subfoveal choroidal thickness (SFCT) was measured on OCT images, and choroidal vessel density (CVD), CVD asymmetry, intervortex anastomosis, and choroidal vascular hyperpermeability (CVH) were assessed on UWICGA images. Results: CVD on UWICGA differed between groups in terms of the total area (0.323 ± 0.034 in dry vs. 0.286 ± 0.038 in non-dry, p < 0.001) and area of each quadrant (superotemporal: 0.317 ± 0.040 vs. 0.283 ± 0.040, superonasal: 0.334 ± 0.040 vs. 0.293 ± 0.045, inferonasal: 0.306 ± 0.051 vs. 0.278 ± 0.052, inferotemporal: 0.334 ± 0.047 vs. 0.290 ± 0.046; all p ≤ 0.010). The CVH grade differed between groups (mean 1.480 ± 0.735 vs. 1.875 ± 0.822, p = 0.013). ST and IT intervortex anastomoses were common in the dry group, while SN, ST, and IT were most common in the non-dry group (p = 0.001). Conclusions: A poor short-term response to aflibercept treatment in PNV eyes was associated with a lower Haller vessel density, higher CVH grade, and intervortex anastomosis involving more quadrants on UWICGA.

Automated segmentation of ultra-widefield fluorescein angiography of diabetic retinopathy using deep learning.

BACKGROUND/AIMS: Retinal capillary non-perfusion (NP) and neovascularisation (NV) are two of the most important angiographic changes in diabetic retinopathy (DR). This study investigated the feasibility of using deep learning (DL) models to automatically segment NP and NV on ultra-widefield fluorescein angiography (UWFA) images from patients with DR. METHODS: Retrospective cross-sectional chart review study. In total, 951 UWFA images were collected from patients with severe non-proliferative DR (NPDR) or proliferative DR (PDR). Each image was segmented and labelled for NP, NV, disc, background and outside areas. Using the labelled images, DL models were trained and validated (80%) using convolutional neural networks (CNNs) for automated segmentation and tested (20%) on test sets. Accuracy of each model and each label were assessed. RESULTS: The best accuracy from CNN models for each label was 0.8208, 0.8338, 0.9801, 0.9253 and 0.9766 for NP, NV, disc, background and outside areas, respectively. The best Intersection over Union for each label was 0.6806, 0.5675, 0.7107, 0.8551 and 0.924 and mean mean boundary F1 score (BF score) was 0.6702, 0.8742, 0.9092, 0.8103 and 0.9006, respectively. CONCLUSIONS: DL models can detect NV and NP as well as disc and outer margins on UWFA with good performance. This automated segmentation of important UWFA features will aid physicians in DR clinics and in overcoming grader subjectivity.