From EyeWire
Jump to: navigation, search

The Retinal Connectome

Our challenge is to map the neural connections of the retina by analyzing images that were acquired using serial electron microscopy at the Max Planck Institute for Medical Research in Heidelberg, Germany. A retinal volume of size 350×300×60 μm3 was imaged, amounting to about one terabyte of data.

Error creating thumbnail: convert: no decode delegate for this image format `/tmp/magick-XXZOgyil' @ constitute.c/ReadImage/530.
convert: missing an image filename `/tmp/transform_ddefb7587f3e-1.jpg' @ convert.c/ConvertImageCommand/2838.
Image of the retina from serial block face scanning electron microscopy (SBEM). Scale bar = 100 microns. Voxel size = 16.5×16.5×23 nm3.

Game 1: Reconstructing Neurons

The first step of the challenge is to reconstruct the tree-like shapes of retinal neurons by tracing their branches through the images. You will accomplish this by playing a simple game: helping the computer color a neuron as if the images were a three-dimensional coloring book. The collective efforts of you and other players will result in three-dimensional reconstructions of neurons like this:

Error creating thumbnail: convert: no decode delegate for this image format `/tmp/magick-XXUiWyjm' @ constitute.c/ReadImage/530.
convert: missing an image filename `/tmp/transform_2d3ddf85dca1-1.jpg' @ convert.c/ConvertImageCommand/2838.

Playing the game does not require any specialized knowledge of neuroscience — just sharp eyes and practice. If you like, you can stop reading this page, and proceed to detailed instructions for the game here or simply start playing. On the other hand, if you’d like to know more about the scientific plan, read on.

Game 2: Identifying Synapses

Reconstructing neurons involves tracing their branches, which are like the “wires” of the retina. This by itself is not enough for finding connectomes; we also need to identify synapses. This kind of image analysis will be accomplished through another game that will be introduced on this website in the near future. The identification of synapses will involve subtleties, due to limitations of the dataset, as will be discussed in detail later on.

Rules of Connection

Playing either of the above games will produce information that will be valuable for understanding how the retina functions. How exactly will the information be used? To answer this question, we should confront the issue of variability. We expect that every retina will be wired somewhat differently. In that case, would mapping the connections in one retina tell us anything that is applicable to other retinas? We expect that retinal connectomes will obey invariant rules of connection, and it is these rules that really interest researchers. Many of the rules are expected to depend on neuronal cell types, i.e., of the form “Cell type A receives synapses from cell type B.” Some such rules are already known, but the vast majority remain undiscovered.

How can we extract such rules from a retinal connectome? A neuron’s cell type can be recognized from its distinctive shape, and hence from the 3D reconstructions that you will help create by playing the coloring game. The cataloguing of retinal cell types has not yet been completed, and your reconstructions will also contribute to this endeavor.

Relating Connections to Activity

Neurons of the retina respond to visual stimuli with electrical activity. Such neural signals eventually travel along the optic nerve from the eye to the brain. Therefore, if we want to understand the role played by the retina in vision, we must also measure the activity of retinal neurons. Furthermore, we should relate the connections of retinal neurons to their activity.

This can be done in two ways. First, before the researchers imaged retina neurons with serial electron microscopy, they used another method of imaging, two-photon microscopy, to measure the activity of the same neurons. Therefore it is possible to relate the activity of retinal neurons to their connectivity. Second, the cell type of a retinal neuron can be identified by its shape, and each cell type responds to visual stimulation in a distinctive manner. Therefore, we can relate the connections of retinal neurons in this dataset to the activity of neurons of the same cell types measured in other retinas.