Scientist??s game helps map the brain
Source: Carolyn Y. Johnson
Sebastian Seung (left) in his MIT lab with Ashwin Vishwanathan. Seung’s EyeWire game enlists help with his research.
When Sebastian Seung read that each day people around the world spend 600 years collectively playing Angry Birds, he saw not a huge waste, but a big opportunity.
Seung, a professor of computational neuroscience at the Massachusetts Institute of Technology, thought that if he could divert even a fraction of that attention to a game with a loftier goal, he could help create the “connectome” ― a map of the vast number of connections in the brain that underlie vision, memory, and disease.
“We believe the connectome is such a difficult challenge that neuroscientists alone can’t do it,” said Seung, whose lab is developing the combination of artificial intelligence and citizen science that could rapidly map a cubic millimeter of brain.
But, he said, “if we were 1 percent as fun [as Angry Birds], we could do this in a year.”
This spring, the physicist-turned-neuroscientist and his team launched EyeWire (eyewire.org), an online game that aims to harness the energy of people working as volunteer “scientists” to build 3-D maps of the cell networks that are crucial for vision.
Players, who could range from anyone interested in science to parents looking for an educational game to play with their children, are presented with black-and-white electron microscope images of slices from the mouse retina, each image like a page from an online coloring book.
The computer highlights individual neurons in blue, but the computer program is set to stop the blue highlighting anywhere it could potentially cross a cell boundary, sometimes leaving bare patches within the cell. Participants are asked to correct the computer’s work and fill in the missing data, using their cursor like a paintbrush dipped in light blue paint.
When all those slices have been colored by people sitting at computers around the world, their efforts are combined to trace the structure of each neuron and the synapses ― the junctions between cells.
EyeWire is just the first step for Seung, who is scientific director of a nonprofit called Wired Differently that will raise funds and mobilize volunteers to aid other brain-mapping efforts.
Eventually, he said, he hopes to focus on diseases such as epilepsy or schizophrenia, in which the brain’s connections may be disturbed, or to examine the role the mind’s wiring plays in memory.
Seung’s brain-connectivity-mapping effort is one of a number of such projects nationwide, each of which is focusing on the brain at a different level of detail, using different technologies.
Seung’s approach is extremely zoomed in, using images taken with an electron microscope to assemble the jigsaw of connections between individual neurons. By Seung’s calculations, tracing all the neural connections in a cubic millimeter of brain would take one person working around the clock 100,000 years. Aided by the computer programs his lab has been building, that task would be slightly more doable, requiring 1,000 years of work.
But if large numbers of people could be persuaded to give a little of their time filling in the bare patches, an intricately tangled, colorful map of the brain’s wiring could emerge rapidly, turning a nearly impossible task into a valuable scientific resource.
At the other end of the scale is the Human Connectome Project, an effort to use sophisticated imaging technology to make regional maps of the connections between different brain areas.
Between those two extremes is an effort headed by Partha Mitra, a neuroscientist at Cold Spring Harbor Laboratory in New York, who is building maps of connections in mouse brains at an intermediate level of resolution, using images taken with a light microscope. He said he believes that reconstructing connectivity at this scale, which is focused on groups of cells and the connections between them, will be more technically feasible than the amount of data collection and analysis required for more finely detailed maps, but will provide necessary resolution missing from the regional maps.
All the techniques have challenges and potential advantages. It remains to be seen which approach will be most fruitful, or whether they all may be necessary to understand the complex wiring of the brain and how it functions.
“It’s different from the genome,” Mitra said, because the genome could be read out as a series of letters. “The brain is a whole different beast; it’s not written in a digital language that we can just read.”
All the approaches depend on the generation and analysis of massive amounts of data.
So far, only a handful of cells have been completely colored in on the black-and-white mouse retina images provided for EyeWire by Seung’s collaborators at the Max Planck Institute for Medical Research in Heidelberg, Germany. But Seung’s efforts are both an attempt to begin to make these complex maps and a way to track how effectively the volunteers can be used to generate rapid and accurate data.
The hope, he said, is that human players’ responses will help make the artificial intelligence accurate and more capable, ultimately allowing scientists to unravel and understand the role the brain’s connections play in perception, thought, and disease.
| }
|