How citizen science, gaming and AI are mapping the brain

Posted on: 06 April 2018

In a fascinating public lecture, international science communicator, Amy Sterling, recently explained how she uses gaming and AI approaches to engage a host of citizen scientists and transform the way we map the most complex thing in the universe — the human brain.

Amy Sterling is the Executive Director of EyeWire and Neo – two games that have been designed to help map the brain. EyeWire crowd-sources neuroscience from gamers, challenging hundreds of thousands of players around the world to solve 3D puzzles and map out neurons, which in turn allows neuroscientists to chart synaptic connections and model circuitry.

In her talk, which was part of the Neurohumanities Public Talks series, Amy described the history of the project, its recent successes, and her efforts in getting the games and gamers to work with AI in order to take the scale of the project to the next level.

Mapping the brain

Billions of neurons, connected through trillions of synapses, make up the most complex thing in the universe: the human brain. The field of connectomics aims to map these billions of neurons and trillions of synaptic connections, because understanding the architecture of our connectomes is crucial for understanding how thoughts, emotions, and consciousness originate in our brains.

To create neural wiring diagrams by hand, researchers need to start with cross-sectional images of brain slices. Each image features cross sections of individual axons and dendrites, which the scientist traces and tracks through the stack of images to painstakingly reconstruct circuits.

In 2012, neuroscientists working at MIT and Princeton launched Eyewire, a crowd-sourced game in which players trace neural wiring on a computer. An army of more than 100,000 ‘citizen neuroscientists’ have played the game, generating an unprecedented neural dataset. The output has already helped identify a new cell type in the retina of the eye and has shown how some retinal neurons respond to movement direction. The development of graphics processing units in parallel with more advanced artificial AI algorithms have further revolutionised connectomics, resulting in software that can reconstruct some brain areas with super human accuracy.

Despite these advances, the next phase of the project is to reconstruct a 1,000-fold larger volume of cortical brain tissue and will require a combination of AI and human supervision. To that end the next-gen crowdsourcing game, NEO, will find different ways for humans to interact with AI. In this new citizen science game people team up with other citizen neuroscientists to strategically decipher the spectacular spiny dendrites and vast axons of neurons found in neocortex of a mouse brain.