A new method for imaging whole-brain structural networks

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, each consisting of thousands of neurons.

For a real understanding of how the brain generates complex behaviours, we need detailed knowledge of the large-scale architecture of the neural networks involved. Visualizing this global architecture is possible with simple organisms.  For example, the nematode worm, Caenhorhabditis elegans, has a nervous system that consists of just 302 cells forming about 7,000 synapses, and the entire process of neural development has been mapped.

Current neuroimaging methods are inadequate for such a task in larger and more complex animals such as humans. But now, researchers from Lausanne in Switzerland, have developed a new technique with which they imaged, for the first time and at high resolution, whole-brain connectivity in healthy volunteers.

The new technique, described by Hagmann, et al in the open access online journal PLoS One, enables imaging of the structural networks of the entire brain in unprecedented detail, and at a higher resolution – approximately 2 millimeters – than earlier methods. It is also very efficient, because it is based on another magnetic resonance imaging (MRI) method; data from a conventional scanner, which takes about an hour to obtain, is processed by a computer to generate the images.

The primary set of data is obtained using diffusion spectrum MRI which visualizes the densities of nerve fibre tracts by detecting the movements of water molecules. Specialized computer programs are then used to perform two different operations on the data. One generates a white matter tractography, (similar to that obtained with diffusion tensor imaging), which shows the major fibres connecting different modules in the brain. The other generates an image that shows the interfaces between white and gray matter (that is, the junctions at which the fibres and the modules meet).

These data can then be combined to produce a graphical representation of millions of individual nerve tracts and thousands of module-tract interfaces. To demonstrate the power of their technique, the researchers used the data obtained from two volunteers to generate a detailed image of the well-characterized interconnectivity of five modules in the visual cortex called Areas V1- V5. (See animations of the whole brain and visual cortex imaging to YouTube.)

vis_sys.JPG

Above, the visual cortex can be seen from the left, top, front and back (a, b, c, d, respectively). The colours used for each of the fibre-module interfaces shows the various structures of the visual system (gray and blue for area V1 on the left and right respectively; orange for V2, cyan for V3; red for V5; magenta for the posterior thalamus) and their connections with each other (yellow for fibres connecting V1 to itself on the opposite side; red for fibres connecting V1 and V2, white for fibres connecting V2 and V3; green for fibres connecting V2 and V5; and blue for the connections between area v1 and the lateral geniculate nulceus in the thalamus).

This new large scale high-res technique will be useful for researchers investigating the evolution and development of the nervous system. It should also prove beneficial to clinicians investigating neurological diseases. Neuroimaging data are conventionally obtained piecmeal from a number of subjects, and then processed to generate images of an “average” brain. Because the new method can provide an image of whole-brain structural connectivity in individual subjects, it will be more useful than earlier methods for making comparisons between healthy and diseased brains.

One thought on “A new method for imaging whole-brain structural networks

  1. Diffusion MRI imaging has been around for almost a decade. Most of the techniques described in this paper are not new. The key unique aspect is the automatic parcellation of brain regions (which was described with too few details to remotely replicate). It is an interesting paper, but no where near as revolutionary or unique as to make it seem. Do some quick literature searches on DTI (or check the papers references) to see what’s already been done.
    Also, please correct your attibution of credit. This was a paper done by researchers in Swizerland with some analysis done by one person at Harvard. If it wasn’t clear from the author list, PLoS has a nice Author Contributions section at the end.

Comments are closed.