Neuroscience: Big Brain

Neuroscience: Big Brain, Big Data

Neuroscience: Big Brain, Big Data

As big brain-mapping initiatives go, Taiwan's might seem small. Scientists there are studying the humble fruit fly, reverse-engineering its brain from images of single neurons. Their efforts have produced 3D maps of brain circuitry in stunning detail.

Researchers need only a computer mouse and web browser to home in on individual cells and zoom back out to intertwined networks of nerve bundles. The wiring diagrams look like colorful threads on a tapestry, and they're clear enough to show which cell clusters control specific behaviors. By stimulating a specific neural circuit, researchers can cue a fly to flap its left wing or swing its head from side to side — feats that roused a late-afternoon crowd in November at the annual meeting of the Society for Neuroscience in San Diego, California.

But even for such a small creature, it has taken the team a full decade to image 60,000 neurons, at a rate of 1 gigabyte per cell, says project leader Ann-Shyn Chiang, a neuroscientist at the National Tsing Hua University in Hsinchu City, Taiwan — and that's not even half of the nerve cells in the Drosophila brain. Using the same protocol to image the 86 billion neurons in the human brain would take an estimated 17 million years, Chiang reported at the meeting.

Read Also:
The Impact of Big Data and Analytics on Manufacturing Companies

Other technologies are more tractable. In July 2016, an international team published a map of the human brain's wrinkled outer layer, the cerebral cortex1. Many scientists consider the result to be the most detailed human brain-connectivity map so far. Yet, even at its highest spatial resolution (1 cubic millimeter), each voxel — the smallest distinguishable element of a 3D object — contains tens of thousands of neurons. That's a far cry from the neural connections that have been mapped at single-cell resolution in the fruit fly.

“In case you thought brain anatomy is a solved problem, take it from us — it isn't,” says Van Wedeen, a neuroscientist at Massachusetts General Hospital in Charlestown and a principal investigator for the Human Connectome Project (HCP), a US-government-funded global consortium that published the brain map.

So it goes in the world of neurobiology, where big data is truly, epically big. Despite advances in computing infrastructure and data transmission, neuroscientists continue to grapple with their version of the 'big data' revolution that swept the genomics field decades ago.

Read Also:
2016: The year AI got creative

But brain mapping and DNA sequencing are different beasts. A single neuroimaging data set can measure in the terabytes — two to three orders of magnitude larger than a complete mammalian genome. Whereas geneticists know when they've finished decoding a stretch of DNA, brain mappers lack clear stopping points and wrestle with much richer sets of imaging and electrophysiological data — all the while wrangling over the best ways to collect, share and interpret them. As scientists develop tools to share and analyze ever-expanding neuroscience data sets, however, they are coming to a shared realization: cracking the brain requires a concerted effort.

Scientists can chart the brain at multiple levels. The HCP seeks to map brain connectivity at a macroscopic scale, using magnetic resonance imaging (MRI). Some labs are mapping neural tracks at a microscopic level, whereas others, such as Chiang's, trace every synapse and neural branch with nanoscale precision. Still others are working to overlay gene-expression patterns, electrophysiological measurements or other functional data on those maps. The approaches use different methods — but all create big data (see 'Big data by the numbers').

Read Also:
This is how the 'Internet of Things' is changing the business model of the world's biggest technology companies

In part, this is because the brain, no matter the species, is so large and interconnected. But it also stems from the cells' unwieldy dimensions. A mammalian neuron's main extension — its axon — can be 200,000 times as long as its smallest branches, called dendrites, are wide.



Read Also:
Teradata (re)embarks on a solutions journey
Read Also:
3 Advantages of Using Neo4j Alongside Oracle RDBMS
Big Data Innovation Summit London
30 Mar
Big Data Innovation Summit London

$200 off with code DATA200

Read Also:
Why machine learning is the new BI
Read Also:
9 Hot Big Data And Analytics Startups To Watch

Leave a Reply

Your email address will not be published. Required fields are marked *