Skip to main content
. 2026 Jul 1;14:1873901. doi: 10.3389/fcell.2026.1873901

FIGURE 2.

Flowchart illustrating a retinal image processing pipeline, starting with an original eye image followed by preprocessing, channel separation into grayscale components, convolutional neural network (CNN) downsampling blocks, linear projection, multiple transformer layers, CNN upsampling, and output showing segmentation of blood vessels overlaid on the retinal image.

Flowchart of artificial intelligence-assisted retinal image processing and quantitative analysis.