Connectome of a human foveal retina

The foveal retina is a primate specialization which presents a feasible site for obtaining a complete connectome of a human CNS structure. In the fovea, cells and circuits are miniaturized and compressed to densely sample the visual image at high resolution, initiating form, color, and motion perception. Here we present a sample of human foveal retina analyzed by deep learning-based segmentation of all cells and synaptic connections. Cells in the volume are provisionally classified into 51 morphological types. We discovered synaptic pathways that are absent in non-human primates that may play a role in human color vision. We also demonstrate a biophysical model based on electrical synapses among short-, middle-, and long-wavelength sensitive cone photoreceptors. Segmentation of the ganglion cells, the output cells of the retina, suggests only 11 visual pathways, with 5 high-density pathways accounting for over 95% of the foveal output to the brain; this number is substantially lower than the 40+ types observed in mouse retina. The connectomic resource presented here reveals distinct features of a human neural system and highlights the potential for deep learning-based computational methods to advance understanding of the human brain.