Official websites use .gov

A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS

A lock ( 🔒 ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Need Help

Utilizing intrinsic data structure with self-supervised computer vision methods to segment key pathologic features in clinical images from patients with age-related macular degeneration

Award Number: F30EY038544
ORGANIZATION: NATIONAL EYE INSTITUTE
OPDIV: NIH
AWARD CLASS: DISCRETIONARY
AWARD ACTIVITY TYPE: FELLOWSHIP/SCHOLARSHIP/STUDENT LOANS
PERIOD OF PERFORMANCE START DATE: 12/21/2026
PERIOD OF PERFORMANCE END DATE: 12/20/2027

Group Awards By:

View Award Description

Utilizing intrinsic data structure with self-supervised computer vision methods to segment key pathologic features in clinical images from patients with age-related macular degeneration - Project Summary/Abstract In age-related macular degeneration (AMD), the retina of the eye progressively degenerates leading to irreversible blindness. Over the course of the disease, multiple lesions appear on the retina. These lesions evolve with time to spread or prelude the development of different lesions at their locations. Many lesions are apparent on optical coherence tomography (OCT) and near-infrared reflectance (NIR) imaging which patients with AMD regularly obtain in clinic. AMD lesions on OCT and NIR imaging have been connected through histological analysis to biological mechanisms of AMD. However, in clinical management of AMD, there remains great unexplained uncertainty and complexity in which patients will progress and how lesions will develop. As such, there is a need for data-driven understanding of AMD. Identifying patterns in AMD across patients and with time will inform studies in the mechanism of disease, provide better, personalized care to patients, and allow the design of earlier clinical trial end points that would facilitate development of AMD treatments. To this end, it is a foremost priority to develop automated tools for accurate segmentation of the many lesions on AMD clinical imaging. Reticular pseudodrusen (RPD), drusen, and geographic atrophy (GA) are three especially important lesions to segment. RPD and drusen are pathologic deposits in the retina while GA is the clinical sign of end-stage retinal degeneration. RPD are especially difficult to segment since the lesions are small and may blend with background noise. Indeed, previous RPD segmentation models only partially identify large RPD. We developed a model for segmenting RPD from OCT and NIR images that identifies over thirty times more RPD than the leading existing model. Furthermore, unlike prior models, our model was developed without any manual data annotations. Aim 1 will extend the RPD segmentation model to be robust to varied OCT scan formats. Aim 2 will develop a unified model for segmentation of RPD, drusen and GA. The approaches developed in both aims will translate to other retinal segmentation tasks, benefiting analysis of additional findings in AMD and other retinal diseases. This work will be conducted at an institution with strong research support for both computer science engineering and ophthalmology. The fellowship training plan, focused on rigorous engineering, scientific, and clinical training, will support the PI in an independent physician-scientist career leading an interdisciplinary research group that develops imaging analysis methods and applies them to discover patterns of importance to clinical management of ocular disease.


Issue Date FY	Funding FY	Legal Entity Name	Legal Entity Address	Legal Entity City	Legal Entity State	Legal Entity Zip Code	Legal Entity COUNTY	Legal Entity COUNTRY	Assistance Listing	Award Code	Budget Year	Action Date	Action Type	Action Amount

Issue Date FY: 2026 ( Subtotal = $45,986 )
2026	2026	REGENTS OF THE UNIVERSITY OF MICHIGAN	1109 GEDDES AVE STE 3300	ANN ARBOR	MI	48109	WASHTENAW	USA	Vision Research	000	1	4/27/2026	NEW	$45,986
														Subtotal = $45,986

Grand Total All Awards = $45,986

Top

All Categories

About

Search

Reports

Data Submission

Award Information

Utilizing intrinsic data structure with self-supervised computer vision methods to segment key pathologic features in clinical images from patients with age-related macular degeneration

Award Number: F30EY038544

ORGANIZATION: NATIONAL EYE INSTITUTE

OPDIV: NIH

AWARD CLASS: DISCRETIONARY

AWARD ACTIVITY TYPE: FELLOWSHIP/SCHOLARSHIP/STUDENT LOANS

PERIOD OF PERFORMANCE START DATE: 12/21/2026

PERIOD OF PERFORMANCE END DATE: 12/20/2027

Federal Websites

Department of Health & Human Services

HHS Operating Divisions

HHS Staff Divisions

Download A Document Viewer