Primary care screening

Last updated March 2025

Tags:Real-world dataPrimary careRetinal photoNW400NW500SIGNALDiabetic retinopathy

This dataset comprises color fundus photography of hundreds of thousands of individuals with diabetes undergoing eye disease screening in primary care settings. Each eye is graded by an eye care specialist for image quality and eye disease. This single-visit dataset provides a broad snapshot of screening outcomes in primary care, which can be useful for development and validation of screening algorithms.

643

Locations

146K

Subjects

291K

Eyes

386K

Images

REQUEST DATASETS

Dataset overview

Dataset attributes	Details
Publishing Frequency	New major versions of the dataset will be released periodically with additional data collected from new sites and/or existing sites
Last Updated	March 2025
Geographic Coverage	USA
Locations	643
Devices	TRC-NW400 (Topcon Corp., Tokyo, Japan), NW-500 (Topcon Corp. Tokyo, Japan), Signal (Optomed Plc, Oulu, Finland), Horus Scope (Medimaging Integrated Solution Inc., Hsinchu, Taiwan)
Image Formats	JPEG

Metadata	Details
Demographic	Age, Gender, ZIP code (first 3 digits only)
Image Metadata	Camera model name

Labels	Details
Diagnosis	Image quality grade, diabetic retinopathy grade (ICDR), other retinal features
Model-generated	AutoMorph¹ segmentations, image quality prediction, vascular metrics Retinal pigmentation score²

References

Zhou Y, Wagner SK, Chia MA, Zhao A, Woodward-Court P, Xu M, et al. AutoMorph: Automated retinal vascular morphology quantification via a deep learning pipeline. Transl Vis Sci Technol. 2022 Jul 8;11(7):12.
Rajesh AE, Olvera-Barrios A, Warwick AN, Wu Y, Stuart KV, Biradar MI, et al. Machine learning derived retinal pigment score from ophthalmic imaging shows ethnicity is not biology. Nat Commun. 2025 Jan 2;16(1):1–14.

Dataset overview

Dataset attributes

Publishing Frequency

New major versions of the dataset will be released periodically with additional data collected from new sites and/or existing sites

Last Updated

March 2025

Geographic Coverage

USA

Locations

643

Devices

TRC-NW400 (Topcon Corp., Tokyo, Japan), NW-500 (Topcon Corp. Tokyo, Japan), Signal (Optomed Plc, Oulu, Finland), Horus Scope (Medimaging Integrated Solution Inc., Hsinchu, Taiwan)

Image Formats

JPEG

Metadata

Demographic

Age, Gender, ZIP code (first 3 digits only)

Image Metadata

Camera model name

Labels

Diagnosis

Image quality grade, diabetic retinopathy grade (ICDR), other retinal features

Model-generated

AutoMorph¹ segmentations, image quality prediction, vascular metrics

Retinal pigmentation score²

References

Zhou Y, Wagner SK, Chia MA, Zhao A, Woodward-Court P, Xu M, et al. AutoMorph: Automated retinal vascular morphology quantification via a deep learning pipeline. Transl Vis Sci Technol. 2022 Jul 8;11(7):12.
Rajesh AE, Olvera-Barrios A, Warwick AN, Wu Y, Stuart KV, Biradar MI, et al. Machine learning derived retinal pigment score from ophthalmic imaging shows ethnicity is not biology. Nat Commun. 2025 Jan 2;16(1):1–14.