[tt] mapping the brain

[Eugen Leitl]

http://www.wired.com/science/discoveries/news/2008/01/connectomics?currentPage=all

Mapping the Most Complex Structure in the Universe: Your Brain

By Alexis Madrigal Email 01.24.08 | 3:00 PM

The ATLUM machine cuts ultrathin slices of mouse brain to prepare them for
mapping the connections that link millions of neurons.

Photo: Lichtman Lab

Harvard scientists have embarked upon an ambitious program to create a
circuit diagram of the human brain, with the help of new machines that
automatically turn brain tissue into high-resolution neural maps.

By mapping every synapse in the brain, researchers hope to create a
"connectome" -- a diagram that would elucidate the brain's activity at a
level of detail far outstripping today's most advanced brain-monitoring tools
like fMRI.

"You're going to see things you didn't expect," said Jeff Lichtman, a Harvard
professor of molecular and cellular biology. "It gives us an opportunity to
witness this vast complicated universe that has been largely inaccessible
until now."

The effort is part of a new field of scientific research called connectomics.
The field is so new that the first course ever taught on it recently ended at
MIT. It is to neuroscience what genomics is to genetics. Where genetics looks
at individual genes or groups of genes, genomics looks at the entire genetic
complement of an organism. Connectomics makes a similar jump in scale and
ambition, from studying individual cells to studying swaths of the brain
containing millions of cells. A full set of images of the human brain at
synapse-level resolution would contain hundreds of petabytes of information,
or about the total amount of storage in Google's data centers, Lichtman
estimates.  Machine Peels Brain, So Scientists Can Map Synapses

It slices, it dices and it heralds the arrival of a new era of neuroscience
that focuses on industrializing the process of mapping the brain.

It's a neuroscience gadget called the automatic tape-collecting lathe
ultramicrotome (ATLUM), and the name says it all. An ultramicrotome is a
piece of laboratory equipment that cuts samples of flesh into very thin
slices. The lathe allows the machine to cut continuously, which makes the
process faster. Already, the prototype has collected more than a hundred
half-centimeter-long sections of mouse brain.

Once the slices have been stuck onto a piece of transparent tape, the
scientists use a scanning electron microscope to actually image the cells.
Harvard molecular biology professor Jeff Lichtman's lab partnered with
optical equipment company JEOL to automate the process of imaging and
ordering those images.

"We will go to each section of tissue that the ATLUM has deposited and
identify the region of that section that contains the important information,
like the wiring of the neurons," said Charles Nielsen, a product manager and
vice president at JEOL. "Then we'll do a series of montage maps on each
section."

Continued on page 2

A map of the mind's circuitry would allow researchers to see the wiring
problems that might underpin disorders like autism and schizophrenia.

"The 'wiring diagram' of the brain could help us understand how the brain
computes, how it wires itself up during development and rewires itself in
adulthood," said Sebastian Seung, a computational-neuroscience professor at
MIT.

But with 100 billion neurons in the human brain, mapping them is an
impossibly complex task for humans alone. An early "by hand" connectomics
effort by Sydney Brenner of the Salk Institute studied the roundworm and its
meager 300 nervous-system cells: It took a decade to complete.

Michael Huerta, associate director of the National Institute of Mental Health
for scientific technology research, said that connectomics will fill a key
gap in our understanding of the brain.

"You could conceivably know every chemical and every molecule of every cell
in the brain, but unless you understand how those cells are connected to each
other, you have no idea how information is being processed," Huerta said.
"The connectome, in my opinion, is really what it's all about."

Lichtman's lab is creating what could be the equivalent of the genome
sequencing machine, which dramatically sped up the race to map the human
genome. It's an automated brain peeler and imager they call ATLUM (sidebar,
left).

ATLUM uses a lathe and specialized knife to create long, thin strips of brain
cells that can be imaged by an electron microscope. Software will eventually
montage the images, creating an ultrahigh-resolution 3-D reconstruction of
the mouse brain, allowing scientists to see features only 50 nanometers
across.

"It works like an apple peeler," Lichtman said. "Our machine takes a brain,
peels off a surface layer, and puts it all on tape. These technologies will
allow us to get to the finest resolution, where every single synapse is
accounted for."

Connectomics differs from other efforts to map the brain not just because of
its methods, but also the type of information it seeks. While the Brain
Atlas, funded by Paul Allen, maps the genes of a mouse brain, Lichtman's lab
is gathering anatomical detail. He's looking at the physical features of
cells, like the size of their synaptic vesicles, which store
neurotransmitters essential for cell communication.

"My background is in neuroanatomy, and to see (connectomics) data is
stunning," Huerta said. "Like the Human Genome Project, this work is giving
us a whole new level of information. The neuroscience community in general is
very excited about it." Machine Peels Brain, So Scientists Can Map Synapses

Continued from page 1

The technological hurdles of stitching together thousands of images (each
5,000 x 4,000 pixels) into a 3-D reconstruction of the brain is daunting. The
team wants to complete the mouse-brain reconstruction in four years, but to
hit that goal, Nielsen said the team would need up to 10 more electron
microscopes to speed image taking.

"In the old days, we'd make an injection and see a few cells light up, and
that was that," said Michael Huerta, associate director for scientific
technology research at the National Institute of Mental Health. "But as areas
in science mature, they get to the point where they are generating huge
amounts of data: in this case, data about connectivity in tissues."

Better image-recognition technology, which turns photographic images into
information that computers can use, could also increase the speed at which
pictures of the brain are transformed into wiring diagrams.

"If our computers could automatically identify the synapses in the images,
and trace axons and dendrites to their parent neurons, then they would be
able to generate brain-wiring diagrams," said Sebastian Seung, a
computational neuroscience professor at MIT. "Although we have made progress,
we are still far from making computers 'smart' enough to do this reliably.
This is a challenge at the frontier of computer science and artificial
intelligence."

Though he's working on a massive scale, Lichtman's inspiration comes from the
desire to understand individual neurons. Specifically, he wants to understand
how neurons go from having dozens of connections at birth to having just a
few. Each cell pares down many weak connections, keeping just a few strong
ones.

"Each baby nerve cell connects to 20 times the amount of nerve cells that it
will have as an adult," said Lichtman. "We try to understand what the rules
of pruning are. If a nerve cell has 100 connections and needs to prune that
down to five, the question is, which five?"

The neurons fight to stay connected, and each competition affects the outcome
for the rest of the cells, Lichtman said.

"So to understand the competition's impact on one cell, you have to
understand all of the competitions," he said.

The net effect of all that neural "hand-to-hand combat" is what we call brain
development, and it's what transforms a baby who can't walk, talk or operate
a Blackberry into a modern, adult human being.

While connectomics researchers are very excited, they're still just getting a
handle on mouse-sized brains. It could be a decade before data-crunching
technology will be available to map the complexity of the human brain.

"Some say that the brain is the most complex structure in the universe," said
Seung. "Right now it would be an incredible achievement just to find the
connectome for a tiny animal like a fly."

But the ATLUM could turn out to be as useful for connectomics researchers as
technologies like sequencers turned out to be for genomics researchers. Then
Lichtman and his colleagues would be able to answer some of the most
fundamental questions about what happens when you take unprogrammed human
beings and release them into the world.

It's the wiring, after all, that provides us with the flexibility that
Lichtman calls "the magic of being human."

"When a dragonfly is born, it has to know how to catch a mosquito," Lichtman
said. "But for us, none of this is built in. Our brains have to go through
this profound education period that lasts until our second decade. What is
changing in our brains?"