Using AI to map the retina

Researchers have created a reference map of the retinal pigment epithelium, outlining five distinct subpopulations of cells

SP eye map
Pixabay/Rudy and Peter Skitterians

New research published in Proceedings of the National Academy of Sciences has described five different subpopulations of cells within the retinal pigment epithelium (RPE).

Scientists from the National Eye Institute in the US applied artificial intelligence (AI) technology to RPE images in order to map distinct categories of cells.

The single-cell resolution map provides a framework for understanding the vulnerability of different areas of the RPE to disease.

Director of the National Eye Institute, Dr Michael Chiang, explained: "The findings will help us develop more precise cell and gene therapies for specific degenerative eye diseases."

AI was used to determine the external dimensions and shape of each cell. Compared to RPE cells in the periphery, foveal RPE tend to be perfectly hexagonal and more compactly situated.

The researchers found that the peripheral retina contains a ring of RPE cells with a similar area to those found around the macula.

A computer was trained using labelled RPE images to analyse the entire RPE monolayer from nine donor eyes with no history of significant eye disease.

Researchers also used AI to examine the RPE of donor eyes with AMD and choroideremia.

The scientists determined that different RPE subpopulations are vulnerable to different types of disease.

Dr Davide Ortolan, of the National Eye Institute, highlighted: "Overall, the results suggest that AI can detect changes of RPE cell morphometry prior to the development of visibly apparent degeneration.”