Hi-res brain topology map reveals network hub


In a very cool paper published yesterday in the open access journal PLoS Biology, an international team of researchers report that they have produced the most detailed and comprehensive map yet of the connections in the human cerebral cortex.

The cerebral cortex contains hundreds of billions of cells organized into thousands of discrete functional modules which act in parallel to generate all human behaviours and cognitive processes.

The new study uses neuroimaging to visualize more than 14,000 connections between nearly 1,000 of these modules, and reveals what the researchers call the brain's structural core, which contains numerous connector hubs that link it to other areas throughout the brain.

Recent advances in neuroimaging technology have led to several new methods collectively known as diffusion imaging, which measure a signal generated by the movements of water molecules to visualize the white matter tracts, or bundles of nerve fibres that connect ditant regions of the brain.

In the new study, led by Olaf Sporns of the Computational Cognitive Neuroscience Laboratory at Indiana University, one of these new methods, called diffusion spectrum imaging, was used to analyse the brain's so-called "connectome", a map of large-scale connectivity within and between the two hemispheres of the brain, in more detail and at a higher resolution than ever before.

The data reveal that the brain contains a central core consisting of 8 distinct subregions in the posterior medial area of the cortex. This core radiates a dense network of fibres to other parts of the cortex, and may act as an integrated system which co-ordinates the combined activity of the two hemispheres.

The researchers then used fMRI to visualize the participants' brains in the resting state. This showed that the central core region they had identified was more active than other regions, and so seems to be the brain's "default" network.

fMRI is conventionally used to measure the relative levels of activity in a small number of brain modules. Such data are of limited usefulness if regarded on their own and not in the context of the functioning of the brain as a whole. Because the new study shows how a large number of these modules are interconnected, it should help researchers to better interpret functional neuroimaging data.



More like this

The higher-order brain functions underlying complex patterns of human behaviour are poorly understood, not least because of the enormous number of neural computations involved. Complex behaviours require the parallel  and integrated activity of hundreds (or even thousands) of discrete brain modules…
Prosopagnosia is a neurological condition characterised by an inability to recognize faces. In the most extreme cases, the prosopagnosic patient cannot even recognize their own face in the mirror or a photograph, and in his 1985 book The Man Who Mistook His Wife For A Hat, the neurologist Oliver…
Healthy aging is characterized by a gradual decline in cognitive function. Mental processes such as attention, memory and the ability to process information are at their peak when people are in their 30s and 40s, but as we get older, we find it increasingly difficult to focus on relevant…
We are being constantly bombarded with news stories containing pretty pictures of the brain, with headings such as "Brain's adventure centre located". Journalists now seem to refer routinely to functional magnetic resonance imaging (fMRI) as "mind reading", and exaggerated claims about its powers…

Mo, just yesterday and again today you make mention of "thousands" of distinct modules.

Much has been made in the philosophical literature about what is and is not a module (from Fodor to Prinz -- I like Prinz's work a lot...
and there is of course neuroscientific work as well (Kanwisher, and later scientists using single and multiple-unit recording in face-selective areas of cortex). But you don't define the term here as you intend it. More importantly you don't define it in the way these paper authors do, where they identify approximately 6 modules, with criteria for selecting them as such.

I mean this criticism most respectfully, and am interested in any response you have.

friend - I'm not actually familiar with the literature on defining brain modules; perhaps I use the term too vaguely. And yes, it was sloppy of me to use a different deifintiion of the word from that of the authors.

Matt - I know exactly what I mean by "discrete", but perhaps my definition differs from yours.

I really do think that is a question of semantics. But thank you both for your comments - I'll take more care in choosing my words next time.

What surprises me is that the network hubs are so posterior? The authors didn't identify any network hubs in the prefrontal cortex. That is a very surprising result.

Jake - If you look at a measure like betweenness centrality, which captures the extent to which a node is involved in many short-cut pathways in the network, then you do indeed see prefrontal regions (e.g. in Figures 7A and B).

Your point is well taken, though, that most of the _mutually_ densely connected nodes were posterior.

This surprised me, too. However, I have not actually been able to find hard evidence for the conventional wisdom that frontal regions are most densely connected... (are you aware of evidence for this claim from e.g. tracer studies?) I would be interested to hear anybody's thoughts/opinions on this issue.

Part of the problem is that 'structural connectivity' can be measured at different scales and with different methods, and so a lot of the inferences that are made across scales and methods may not be valid.

In this case, the prefrontal regions may not have made it into the "core" because even though some of them are highly connected, they are not connected to many other hubs -- they are more isolated, and so they get pruned out of the "core", which is determined using a k-core decomposition.

[Disclosure: I was an author on this paper]

By Chris Honey (not verified) on 06 Jul 2008 #permalink

Nice feat, but visualizing small scale neurones, even in low res, is still far from visualizing large scale connectome in high res. Anyway, this achievement may pave the way for future researches to build upon it.