Skip to main content

Diffusion Imaging - Mapping the Connectome

From The Human Connectome Project Is a First-of-its-Kind Map of the Brain's Circuitry:
Working with $30 million and just half a decade, the Human Connectome Project aims to create a first-of-its-kind map of the brain’s complex circuitry, detailing every connection linking thousands of different regions of the brain. ...
The project aims to tap state-of-the-art brain scanning technologies, including diffusion imaging, various MRI methods, and magnetoencephalography to map not just how messages move through the brain, but how various regions work together via networks and networks of networks to achieve the complexity that is the human mind. With map resolutions down to the voxel – small swaths of grey matter containing about one million neurons each – researchers estimate the HCP will generate about one petabyte of data, which will require its own supercomputer to process.

All that scanning, data gathering, and analysis should pay off though, HCP researchers say. The end result will be an open platform that other neuroscientists can use to test their own theories, hypotheses, and findings against. Such a map should help scientists find their way to deeper understandings of how the brain works as well as cures for complicated neurological disorders.

Diffusion Tensor Magnetic Resonance Imaging:

Understanding Diffusion MR Imaging Techniques: From Scalar Diffusion-weighted Imaging to Diffusion Tensor Imaging and Beyond by Patric Hagmann et. al provides a nice overview of how Diffusion Tensor MRI works. Basically, MRI is used to detect the displacement distribution (a.k.a. diffusion) of water molecules along the 'pipes' formed by axons in the brain. Experimental evidence suggests that the tissue component predominantly responsible for the anisotropy of molecular diffusion observed in white matter is not myelin, as one might expect, but rather the cell membrane. The degree of myelination of the individual axons and the density of cellular packing seem merely to modulate anisotropy. Furthermore, axonal transport, microtubules, and neurofilaments appear to play only a minor role in anisotropy measured at MR imaging. In a conventional MRI, every 3D position is assigned a grey-level value, whereas Diffusion Tensor MRI assigns it a 3D image that encodes the molecular displacement distribution.

Hubs and Networks
Some interesting findings are already starting to be discovered using this technology. From Study: A rich club in the human brain:

"We've known for a while that the brain has some regions that are 'rich' in the sense of being highly connected to many other parts of the brain," said Olaf Sporns, professor in the Department of Psychological and Brain Sciences in IU's College of Arts and Sciences. "It now turns out that these regions are not only individually rich, they are forming a 'rich club.' They are strongly linked to each other, exchanging information and collaborating."

The study, "Rich-Club Organization of the Human Connectome," is published in the Nov. 2 issue of the Journal of Neuroscience. The research is part of an ongoing intensive effort to map the intricate networks of the human brain, casting the brain as an integrated dynamic system rather than a set of individual regions.

Using diffusion imaging, which is a form of MRI, Martijn van den Heuvel, a professor at the Rudolf Magnus Institute of Neuroscience at University Medical Center Utrecht, and Sporns examined the brains of 21 healthy men and women and mapped their large-scale network connectivity. They found a group of 12 strongly interconnected bihemispheric hub regions, comprising the precuneus, superior frontal and superior parietal cortex, as well as the subcortical hippocampus, putamen and thalamus. Together, these regions form the brain's "rich club."

Most of these areas are engaged in a wide range of complex behavioral and cognitive tasks, rather than more specialized processing such as vision and motor control. If the brain network involving the rich club is disrupted or damaged, said Sporns, the negative impact would likely be disproportionate because of its central position in the network and the number of connections it contains. By contrast, damage to regions outside of the rich club would likely cause specific impairments but would likely have little influence on the global flow of information throughout the brain.

Sporns said the cohesive nature of the rich club's interconnections was surprising and unexpected. It would not have been implausible to have highly connected nodes that did not interact or influence each other to the same degree.

"It's a group of highly influential regions that keep each other informed and likely collaborate on issues that concern whole brain functioning," he said.

Connectivity vs. Functionality
One of the things I find both annoying and almost funny is the marketing language that is being used for supercomputing simulations that try to equate the number of computations per second a supercomputer can make to an 'equivalent' level of neurobiology. IBM says that they can apparently do 'cat-scale' simulations. This, in spite of the fact that we don't fully understand how even the simplest neural networks work at a detailed level. Neurobiologist Henry Markram has gone as far as calling the IBM Cat Scale Brain Simulation a Hoax.

So it's important to take a step back and look at how this work fits into the larger scheme of things... From a comment made by Olaf Sporn in Brian Science Podcast interview with Olaf Sporn:
I think it would be simple-minded to reduce the brain to a wiring diagram. That’s certainly not my intention, and I think it would be simple-minded if one were to propose that. You mentioned the worm, C. elegans, earlier. It has about 300 neurons—something like that—fairly stereotypically connected to each other. And we’ve known that particular wiring diagram now for 25 years, as a result of the heroic efforts of researchers who reconstructed this meticulously in the early ‘80s. But we still don’t really understand how the nervous system of C. elegans works in its entirety.

So, it is something that we need to know—sort of like the genome. We really do want that information. But it doesn’t fully explain the functioning of the organism or of the nervous system; it only gives us a foundation. It’s necessary, but not sufficient.

In addition to mapping connectivity in the brain, Diffusion Tensor Imaging (DTI) is also providing insight into brain injuries such as concussions. From Dr. Randall Benson (quoted in the Brain Damage Blog (Jan 8, 2010): ):

Closed head injuries (non-penetrating) including concussion are caused by sudden acceleration or deceleration of the head which causes local deformations of the brain within the cranium. The anatomical and biomechanical properties of the brain are such that white matter fibers are stretched and damaged, resulting in diffuse axonal injury (DAI) which is the hallmark pathology and accounts for most of the neurological disability in TBI (Traumatic Brain Injury).

The typical cognitive deficits in TBI, i.e., slowed information processing, decreased attention and memory, and psychiatric symptoms are caused by damage to the “cables” which allow for efficient transmission of information between neurons. TBI reduces brain network efficiency resulting in decreased capacity and global functional impairment. Concussive injury such as occurs in football with high speed collisions also causes deformation of brain substance and is felt to account for many of the immediate and delayed symptoms including the post-concussive syndrome. ERP studies of sports related concussion suggest that symptomatic recovery may occur while neurologic and brain metabolic functioning continues to be impaired from weeks to months after injury.

Incurring a second concussion before neurologic recovery has been shown to worsen outcome and may begin a downward spiral culminating in chronic traumatic encephalopathy (CTE) but this is not known. Diffusion tensor imaging (DTI) is able to detect damaged white matter fibers (axons) which have altered flow of water molecules compared with healthy axons.


Check out the Brian Science Podcast interview with Olaf Sporn, which covers the work he has been doing on brain networks.

Networks of the Brain by Olaf Sporns.


Popular posts from this blog

Neurotransmitters - molecular messages

You often hear about neurotransmitters in the news and in science magazines in a kind of off-hand way that assumes everyone must surely know what these things are. But, um, what are they, exactly?

From Sandra Ackerman's book "Discovering the Brain": To be recognized as a neurotransmitter, a chemical compound must satisfy six conditions: It must be

synthesized in the neuron, stored there, released in sufficient quantity to bring about some physical effect; when administered experimentally, the compound must demonstrate the same effect that it brings about in living tissue; there must be receptor sites specific to this compound on the postsynaptic membrane, as well as a means for shutting off its effect, either by causing its swift decomposition or by reuptake, absorbing it back into the cell.

OK, well, what about hormones? They're chemical messengers too - how are hormones different from neurotransmitters? A hormone, by definition, is a compound produced by an endocrine…

It ain't what you don't know that gets you into trouble. It's what you know for sure that just ain't so.

-- Mark Twain

So far, most of the posts in this blog have been focused on building a 'bottom-up' understanding of how the brain works - from how DNA works up to how individual neurons work. Lots of good science to base all of this stuff on. It is difficult to go further 'up the stack' in this way, however.  How do neurons work together to do useful things? How are small-scale networks of neurons structured and how do the neurons interact in order for us to do simple things like rhythmically tap a finger?

Are we there yet?
Every decade or two the scientific community gets wildly optimistic that we will be able to fully understand how cognition works and be able to replicate the process in some non-biological system. It's been named many things over the years - cybernetics, artificial intelligence, computational intelligence, cognitive computing (see for a nice overview).  And yet, with all of the money tha…

Looking at Sound

Lately I've been listening a lot to Kate Bush's album Aerial - beautiful, wonderful stuff. The album cover is interesting too - the 'islands' that are reflected in the water are actually the amplitude envelope of a recording of some birds singing.

This idea of 'looking at sound' in different ways has been something I've really enjoyed exploring over the last several years. To help visualize the harmonics in a piece of music, I wrote a program a while back that analyses the frequency content of a sound waveform and creates a spectrogram (spectrum over time) of it, colour coding the intensity levels of each frequency.

I think I've found the bird song shown on the cover - it's 2:25 from the start of the song 'Aerial'. Here's what its spectrogram looks like:

The parallel contour lines that are stacked one on top of each other are the harmonics of the bird song. (A synthesizer's been added to the recording, which has changed the amplitue…