Announcing EyeWire II

Clusters of neurons from the Ganglion Cell Layer that have recorded activity via Calcium imaging.

Thirteen years, 350,000 players, and over 6,000 neurons later…
What began in 2012 with EyeWire has evolved.  The sequel is here, fueled by a decade of AI improvements and the brilliance of a global community.

Introducing an entirely new brain mapping platform: EyeWire II. Get on the citizen science wait list. If you are a researcher, request access by emailing support@eyewire.org.

Discussion in the EyeWire Forum.

Unlike Eyewire, which is (by today’s standards) a small-ish volume measuring 300 x 350 microns, the new volume is 1,000 x 1,000 microns and includes over 100,000 neurons.  AI can now create 3D segments that make up the majority of each cell. But the AI still makes many mistakes, missing entire branches and merging multiple cells together. Most cells have only a few errors, so you can sometimes complete the cell in a few minutes. Though others can of course take longer, as was the case of the contenders for most complex cells ever mapped at Seung Lab, the CT1 fly eye-spanning neuron and the APL fly memory neurons, which took literally weeks of proofreading.

EyeWire won’t have cells that hard! In fact, you’ll recognize most of the cells because the new dataset is also a retina.

In addition to mapping the shapes and connections of retinal neurons, Eyewire II also lets us peek into what these neurons are doing — how they respond when the retina “sees” something – through the use of calcium imaging.

Scientists recorded activity from 378 cells in the ganglion layer, which contains the cells that send precomputed visual information to the brain — while showing the retina different visual scenes (visual stimuli). These included movies from the natural environment of the animals. By combining what neurons look like, how they’re wired, and what they do, Eyewire II is building the most complete maps of retinal function so far.

This kind of research helps us answer big questions, like:

  • How does the brain decode the visual world?
  • How do specific types of neurons collaborate to create sight?
  • What goes wrong in diseases like glaucoma, macular degeneration, or retinal dystrophies?
  • And can we build better brain-inspired AIs by understanding the retina’s neural code?

We’re beginning proofreading with Starburst Cells. You may remember them from EyeWire’s Starburst Challenge. You can see candidates for proofreading below:

Do you accept the mission to find and map Starburst Cells in the retina?

Click here to join the ranks of citizen scientists venturing into the future of brain mapping.

EyeWire II launches in alpha mode to Flyers (citizen scientists who contributed to flywire.ai) on Monday, March 31st, 2025. It will soon open to EyeWirers rank Scythe and higher.

By contributing your time and curiosity, you’re not just mapping neurons — you’re joining a movement to decode the brain, one connection at a time.

For Science!

–

Example proofreading challenge: the 2 yellow cells are fused together

Proofreading starburst amacrine cells in EyeWire II.

 

Diagram of retinal layers:

Layers of the retina. The Eyewire II dataset spans the solid box and sometimes includes what is within the dotted line.

 

You can see a demonstration of the new interface, called Neuroglancer, below, and explore it with the microns dataset to get a sense of navgation in EyeWire II at https://www.microns-explorer.org/gallery-mm3 :

Muller Cells have been recently identified in the EyeWire II volume. First appearance since Grimm’s Haunted Carnival 2018. Cell Guide coming soon!

Can’t wait? Browse Self Guided Training.

A deeper dive into Calcium Imaging

Calcium imaging is a technique that visualizes neural activity by detecting changes in calcium concentration within neurons, typically using fluorescent indicators. Scientists use it to understand how groups of neurons communicate and coordinate activity, revealing insights into brain function, behavior, and cognition.

Here’s how it works:

Neurons talk by firing signals, and every time one fires, calcium ions rush inside. Calcium imaging uses special dyes or proteins that light up when calcium is present. So, when a neuron is active, it literally glows — like watching fireworks in the brain.

Two-photon imaging is a high-resolution way of watching this glowing in live tissue, kind of like having night vision goggles that can see deep into the retina without damaging it.