AI summary · may contain errors
x0-diffusion is the most label-efficient vessel segmenter for catheter-based angiograms, reaching Dice 0.882 and IoU 0.795 at 16 labeled images.
- Open access
Moving in Investigative Ophthalmology & Visual Science, Ophthalmology, Experimental Eye Research, medRxiv, Progress in Retinal and Eye Research, bioRxiv, Diabetes and eLife.
Rebuilt
Nothing in this topic has moved yet this week.
Newest first · last 60 days
AI summary · may contain errors
x0-diffusion is the most label-efficient vessel segmenter for catheter-based angiograms, reaching Dice 0.882 and IoU 0.795 at 16 labeled images.
AI summary · may contain errors
UNet-based deep learning quantifies regressed retinal capillaries in streptozotocin-induced diabetic mice by detecting Collagen IV+ CD31− basement membrane sleeves, matching manual sleeve counts.
AI summary · may contain errors
Adaptive optics enables in vivo, single-cell–level imaging of multiple human retinal cell types, including photoreceptors, RPE, immune cells, and microvasculature, across FIAO, AO-OCT, and AOSLO platforms.
AI summary · may contain errors
A supervised Random Forest classifier distinguishes descending arteries and ascending veins in whole-mouse penetrating vessel trees, reaching 96% balanced accuracy on an independent two-photon dataset and 93% on light-sheet datasets.
AI summary · may contain errors
In mouse cortex after stroke, divergent-to-convergent capillary subtypes show the highest capillary stall risk, with stalls and flow changes occurring within the same vascular networks.
AI summary · may contain errors
A standardized ring-based OCT protocol for mouse ganglion cell complex thickness detects significant thinning over time, measuring 68.8 ± 5.2 μm at baseline versus 60.7 ± 8.0 μm at day 15.
AI summary · may contain errors
RetiME, a deep learning biomarker from low-cost retinal photographs, robustly predicts all-cause and cause-specific mortality and reconstructs personalized lifespan trajectories linked to observed outcomes.
AI summary · may contain errors
AO-OCT detected ganglion cell layer soma loss in glaucoma subjects over 3–12 month intervals, with annualized rates as high as -7.8%/yr.
AI summary · may contain errors
In the Rotterdam Study (19,088 eyes; 1.1% perimetric glaucoma), the AIROGS deep-learning algorithm achieved 70.6% sensitivity at 95% specificity with ROC-AUC 0.911.
AI summary · may contain errors
Urban practice locations performed 67%–80% of diabetic retinopathy therapeutic and 65%–90% of diagnostic procedures in the IRIS® Registry, while nonurban sites contributed only 1%–35%.
AI summary · may contain errors
Combining glaucoma diagnosis with treatment evidence in UK Biobank participants yields stronger associations with severity biomarkers than diagnosis-only definitions, e.g., IOPcc OR 23.6 vs 11.4.
AI summary · may contain errors
Heatmap analysis across 54 convolutional neural networks for referable diabetic retinopathy shows most attention to the central/temporal macular region, with exudates and cotton wool spots highlighted most often.
AI summary · may contain errors
In spontaneously hypertensive rats, endothelin 2 is the most significantly upregulated retinal endothelin component and its knockdown restores retinal function despite persistently low intraocular pressure.
AI summary · may contain errors
Visual acuity in healthy eyes declines nonlinearly as cone dropout increases, but eye motion preserves performance at the highest dropout level versus static stimulation.
AI summary · may contain errors
CausalFund improves smartphone-domain AUC for glaucoma and diabetic retinopathy screening across seven deep learning backbones while keeping hospital-domain performance comparable.
Journals this month
Topics moving, week to 19 Nov 2026
By field