NIH researchers build first large-scale database of cone photoreceptor cell size in the living human eye

Top shows color fundus photograph where "X" denotes the fovea and small boxes represent regions of interest (ROIs). Bottom photos (B) represent non-confocal split detection adaptive optics (AO) images from the ROIs shown in fundus photo. Bottom photos (C) show same ROIs from B, but with segmented cones overlaid.

National Eye Institute researchers have created the largest known database providing a reference for the size of individual cone photoreceptors, the cells in the retina responsible for color vision, across the lifespan. This resource will help clinicians distinguish early disease-related damage from typical aging, enabling earlier diagnosis and better tracking of whether new treatments are working at the cellular level. The work was published in Investigative Ophthalmology & Visual Science.

"Until now, there has been no large, publicly available dataset showing what healthy cone photoreceptors look like at different ages, sexes, and retinal locations," said senior investigator for the study, Johnny Tam, Ph.D., of NIH's National Eye Institute.

Diseases such as retinitis pigmentosa, Usher syndrome, and choroideremia are already known to alter the size and shape of cone photoreceptors. However, without a reference for "normal", it is difficult to gauge what changes are due to disease versus ordinary aging or natural variation between men and women.

Twenty-eight healthy volunteers, evenly split between men and women, and ranging in ages from 12 to 84, each had one eye imaged. To resolve individual cone cells, researchers used a technique called adaptive optics to correct for distortions to light as it passes through the structures of the eye. The team used this technology to image cells along a strip of retina extending from the center of the retina, outward.

Once the images were collected, the researchers used an artificial intelligence (AI) algorithm to trace the outline of each cone cell. Experts carefully reviewed every tracing, removing poorly defined cells, and adding any that the AI missed, refining the results until at least three independent reviewers agreed on the final tracings. This rigorous process produced a verified dataset of 9,350 measured inner segments (the metabolically active part) of cone cells.

Statisticians analyzed the measurements, accounting for each person's age, sex, and eye length. This approach allowed the researchers to determine how much variation in cone size was truly associated with age or sex, rather than to other differences between individuals.

The analyses showed a clear pattern: cones were smallest near the fovea, the central region of the retina required for sharp vision and gradually became larger with increasing distance from the fovea. Women's cones were on average about 5% larger than men's. With advancing age, cones gradually shrink over the decades, with the most noticeable shrinkage occurring at moderate distances from the fovea, suggesting these regions may be particularly susceptible to the effects of aging.

"This database represents the culmination of a decade of advancements in retinal imaging and artificial intelligence to make analyzing such a large number of cells possible," Tam said.

"As an open-source database, we hope it will enable the vision science community to carry out future studies of patients with retinal diseases by tracking subtle, cell-level changes over time as new therapies are developed and tested," said Nancy Aguilera, B.S., first author on the study and an engineer in Tam's lab.

The researchers plan to expand the database to include images beyond the areas of retina already included, and they plan to explore how cone size changes in a variety of blinding conditions, including retinitis pigmentosa, choroideremia, and Stargardt disease.

The work was supported by the NEI Intramural Research Program.

Reference: Aguilera N, Iyer S, Cruz MJ, Li J, Wall M, Kallem M, Volkov A, Brooks BP, Zein WM, Huryn LA, Liu T, Tam J. "In Vivo Human Cone Photoreceptor Inner Segment Diameter Database using Adaptive Optics Retinal Imaging" published August 19, 2026 in Investigative Ophthalmology & Visual Science. https://doi.org/10.1167/iovs.67.10.47