The Powered Research team will present two abstracts at the ARVO 2026 conference in Denver, Colorado.
DENVER, April 9, 2026 /PRNewswire/ --
Poster Title: Artificial Intelligence to Quantify Leakage of Neovascular Tufts from Fundus Imaging
Poster Session Date & Time: Tuesday, May 5th, 2026 - 1:15 PM – 3:00 PM
Poster Number: 3176 - 0621
Session Number: 347
Session Title: AI in retina II
Authors: Zsolt Ablonczy, Daniel Costa, Stephanie M. Fogerson, Justin Prater
Purpose
Neovascular tufts (NVT) in the DL-2-aminoadipic acid (DL-AAA) model leads to profuse vascular leakage and edema-like features that mimic those observed in patients with wet aged related macular degeneration and diabetic retinopathy. Current readouts for the pathology include quantification of leakage in vivo and neovascularization in vitro. Here we present a machine learning-based image analysis tool that enables rapid and accurate assessment of total vascular area from fluorescein angiography (FA) data collected in preclinical settings and validate it against classic endpoints.
Methods
Neovascularization was induced via an IVT dose of DL-AAA in one eye of 4–6-month-old Dutch Belted rabbits. Control rabbits received a single 1 mg IVT dose of aflibercept 2.5 weeks post-induction. FA imaging via OCT was conducted 2-3 days prior to aflibercept administration and once weekly post-dose. In vivo, total vascular area was analyzed from images recorded through late-stage FA via a trained pixel classifier with QuPath. The results were compared to traditional corrected total region fluorescence (CTLF) assessment in ImageJ as well as neovascular tuft artificial intelligence (AI)-assisted analysis in retina flat mounts.
Results
At baseline, our AI-assisted analysis found the average total vascular area decreased to 40% (SD) in two weeks post aflibercept administration and 41% at four weeks post aflibercept treatment. Consistent with the AI-assisted FA data, the in vivo vascular leakage and in vitro neovascular tuft area were approximately 35 % and 25 % in aflibercept treated eyes compared to untreated eyes respectively. In addition, the number of vessels within the medullary ray was significantly reduced in aflibercept-treated eyes. The data remained consistent across all three quantification methods throughout the course of the study.
Conclusions
Our novel AI-assisted analysis accurately quantified total vascular area in-vivo over a four-week period in DL-AAA induced eyes with or without aflibercept treatment, thus providing a novel readout for another aspect of the disease pathology. This new machine learning tool provides a rapid, reliable, and unbiassed assessment tool for evaluating the impact of therapies in this model. This novel analysis allows high-throughput and noninvasive testing of pharmaceuticals designed to treat neovascular diseases in the eye to treat wet-AMD and diabetic retinopathy.
Poster Title: Correlation Between Nerve Density and Corneal in Two Rabbit Strains
Poster Session Date & Time: Tuesday, May 5th, 2026 - 3:30 PM – 5:15 PM
Poster Number: 3447 - 0030
Session Number: 366
Session Title: Corneal neuropathy and ocular pain
Authors: Flavia Leao Barbosa, Stephanie M. Fogerson, Dan Costa, Justin Prater, Brian Gilger, David Culp
Purpose
The New Zealand White (albino) and Dutch Belted (pigmented) rabbits are commonly used in ophthalmology research because their eye sizes are more comparable to human eyes, making study results more translatable, especially in models of corneal surface disease. Corneal nerves are essential for maintaining ocular surface health, and their trophic function may be important not only in normal conditions but also in disease states. Understanding differences between rabbit strains, nerve morphology, and corneal sensation is crucial for developing therapeutic interventions and evaluating treatment efficacy.
Methods
A total of fifty-seven female were included in this study, including twenty New Zealand white (NZW) and third-seven Dutch Belted (DB) rabbits. All animals were free of ocular disease as determined by a board-certified veterinary ophthalmologist before enrollment in the experiments. In this study, we assessed central corneal sensitivity using Cochet-Bonnet esthesiometry and measured corneal nerve density with widefield fluorescence microscopy.
Results
The mean central corneal sensitivity was significantly lower in DB (2.53± SD 0.32) than in NZW rabbits (5.19± SD 0.78) (p<0.0001). Superficial nerve projections appeared denser in the corneas of DB rabbits compared to NZW rabbits, and sub-basal nerves were denser in NZW rabbits, which may relate to increased sensation in NZW rabbits as measured by esthesiometry.
Conclusions
Corneal nerve density and sensitivity can be affected by factors such as corneal injury or disease. Corneal sensitivity offers an indirect measure of the function of corneal innervation, and changes in corneal sensation are often linked to alterations in nerve density. We observed that sensitivity varied significantly between rabbit strains, with pigmented rabbits being less sensitive than albino rabbits, a finding that corresponds to the decreased sub-basal corneal nerve density in DB rabbits. These findings emphasize the importance of selecting the appropriate strain for modeling corneal pain and innervation.
About Powered Research:
Powered Research specializes in preclinical research in ophthalmology and oncology. Our staff has extensive expertise in the design and development of non-GLP preclinical data packages, including tolerability and pharmacology models, drug distribution studies, and custom experimental designs. We strive to design the right studies for the product to support our clients' development strategies. Powered Research is well known for rapid study turnaround, and our scientific experts allow for timely study scheduling and execution. We have a collection of internal protocols for specific models and develop custom models based on client specifications and needs.
For more information, visit www.poweredresearch.com.
Media Contact
Calesha Lucas
Research Triangle Park
(984) 884-4936
[email protected]
www.poweredresearch.com
SOURCE Powered Research, LLC
Share this article