Deep learning algorithms have been used to detect diabetic retinopathy (DR) with specialist-level accuracy. This study aims to validate one such algorithm on a large-scale clinical population, and compare the algorithm performance with that of human graders. A total of 25,326 gradable retinal images of patients with diabetes from the community-based, nationwide screening program of DR in Thailand were analyzed for DR severity and referable diabetic macular edema (DME). Grades adjudicated by a panel of international retinal specialists served as the reference standard. Relative to human graders, for detecting referable DR (moderate NPDR or worse), the deep learning algorithm had significantly higher sensitivity (0.97 vs. 0.74,
p
< 0.001), and a slightly lower specificity (0.96 vs. 0.98,
p
< 0.001). Higher sensitivity of the algorithm was also observed for each of the categories of severe or worse NPDR, PDR, and DME (
p
< 0.001 for all comparisons). The quadratic-weighted kappa for determination of DR severity levels by the algorithm and human graders was 0.85 and 0.78 respectively (
p
< 0.001 for the difference). Across different severity levels of DR for determining referable disease, deep learning significantly reduced the false negative rate (by 23%) at the cost of slightly higher false positive rates (2%). Deep learning algorithms may serve as a valuable tool for DR screening.
PurposeDrusenoid pigment epithelial detachments (PEDs) are a defined path to atrophy in age-related macular degeneration (AMD). We analyzed the relationships between retinal pigment epithelium (RPE) and drusen volume changes during the PED lifecycle, using spectral-domain optical coherence tomography (SD-OCT).MethodsTwenty-one cases of drusenoid PED tracked using SD-OCT through periods of growth and collapse were evaluated. Volumetric calculations and piece-wise linear regression analysis were used to determine the breakpoint between growth and collapse. Spectral-domain OCT scans were independently evaluated for the appearance of intraretinal hyperreflective foci, acquired vitelliform lesions (AVLs), and disruptions to the RPE+basal lamina band. Timing of these events with respect to the breakpoint was statistically evaluated. Morphometric characteristics of drusenoid PEDs were correlated with rate of PED collapse and final visual acuity.ResultsMean age of subjects was 75.3 years and mean period of follow up was 4.1 years (median 4.5 years; range, 0.6–6.6 years). The lifecycle of drusenoid PEDs was asymmetric, in that the rate of collapse (0.199 mm3/month) is significantly faster (P < 0.001) than the rate of growth (0.022 mm3/month). Appearance of intraretinal hyperreflective foci and AVLs preceded the breakpoint (both P < 0.001). The timing of disruptions to the RPE+basal lamina band did not differ from the breakpoint (P = 0.510). Maximal height, volume, and diameter of drusenoid PEDs were inversely correlated with final visual acuity (all P < 0.001) and positively correlated with the rate of PED collapse (all P < 0.001).ConclusionsSpectral-domain OCT signatures, plausibly attributable to anteriorly migrated RPE and disintegration of the RPE layer, precede or occur simultaneously with changes in volume of drusenoid PED during the lifecycle of this lesion.
En face OCTA and structural OCT showed better detection of type 1 NV than either FA alone or en face OCTA alone. Combining en face OCTA and structural OCT information may therefore be a useful way to noninvasively diagnose and monitor the treatment of type 1 NV.
Treatment-naive neovascular age-related macular degeneration eyes with Type 1 neovascularization at baseline were less likely to develop GA than eyes with other types. The correlation between apparent GA growth and subtype of neovascularization is stronger when lesions are classified with an anatomic grading that utilizes both FA and SD-OCT.
With the advent of anti-vascular endothelial growth factor (VEGF) therapy, clinicians are now focused on various treatment strategies to better control neovascular age-related macular degeneration (NVAMD), a leading cause of irreversible blindness. Herein, we retrospectively reviewed consecutive patients with treatment-naïve NVAMD initially classified based on fluorescein angiography (FA) alone or with an anatomic classification utilizing both FA and optical coherence tomography (OCT) and correlated long-term visual outcomes of these patients treated with an anti-VEGF Treat-and-Extend Regimen (TER) with baseline characteristics including neovascular phenotype. Overall, 185 patients (210 eyes) were followed over an average of 3.5 years (range 1–6.6) with a retention rate of 62.9%, and visual acuity significantly improved with a TER that required a mean number of 8.3 (±1.6) (± standard deviation) intravitreal anti-VEGF injections/year (range 4–13). The number of injections and the anatomic classification were independent predictors of visual acuity at 6 months, 1, 2, 3 and 4 years. Patients with Type 1 neovascularization had better visual outcomes and received more injections than the other neovascular subtypes. There were no serious adverse events. A TER provided sustained long-term visual gains. Eyes with Type 1 neovascularization had better visual outcomes than those with other neovascular subtypes.
A greater number of intravitreal anti-vascular endothelial growth factor injections is associated with an increased risk for sustained IOP elevation in eyes with neovascular age-related macular degeneration receiving intravitreal ranbizumab and/or bevacizumab.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.