Imaging the spatial organization of brain connections

Understanding how the human brain is able to solve complex problems remains a major challenge in neuroscience research. Humans are capable, even at an early age, of performing seemingly effortlessly many tasks that are still incredibly difficult for the most powerful computers.
Scientists believe that part of the brain's extraordinary abilities comes from its complex wiring. The brain is thought to be organised into specialised regions. The dense network of connections between these regions enables fast, efficient and massively parallel processing of information.
However, the concept of a brain region that is specialised in a given computation is not satisfactory for brain scientists. Indeed, if that were the case, the human brain's abilities would be rather limited. Only a few hundred "regions" can fit in the 2.5 square feet or so of cortex that our brain contains; the size of a large pizza. If every region was only able to perform one task, we wouldn't be able to do complicated things like speaking, playing tennis or thinking about the meaning of life.
An intermediate view would suggest that the brain could indeed be subdivided into a reasonable number of regions. These regions may have a certain degree of specialisation, but may also perform a variety of different tasks in different contexts. I call these different patterns of brain activity "functional modes".
Now the key feature of these functional modes is that they can interact. A region that can do many things can also do combinations of these things. For example, a brain region can calculate our position in space, while another region has a map of our face. A third region that integrates these two pieces of information can do something even more complex: it can allow us to intercept an object that is going to hit us in the face, by calculating a transformation between the outside space and the face coordinates. The two maps can interact because they overlap.
The number and organisation of these functional modes is still largely unknown. My research will provide tools to measure these complex maps by measuring connections between brain regions. By looking at how a brain region connects to different other regions, we can determine its overlapping functions. The more complex the connectivity of a region is, the more modes it can potentially have, and the more complex functions it can potentially perform.
Nowadays, we can measure brain connections "non-invasively", meaning using techniques, such as MRI, that are completely harmless, and can be used on any normal healthy human. We can therefore get a better sense of the complexity of brain function by measuring connections in a novel way.
The research I am planning to conduct aims at providing brain scientists with the tools that will enable them to investigate this intricate relationship between the connections in the brain and the complexity of brain function.

This proposal aims to develop new approaches for defining and measuring functional organization in the human brain. The current paradigm in macro-connectomics is the idea that we can summarise brain connectivity using a network of "nodes" and "edges" - distinct brain units and the connections between them. However, the concept of "nodes" as segregated brain units remains elusive; and the simplistic view of the brain as a discrete set of units with edges representing their interactions overlooks complexities that have important functional consequences.

Brain connections often follow a spatial organisation. Tracer studies in non-human primates have shown that large-scale connections can be organised into local patterns such as topographies, gradients or sub-clusters. Crucially, different spatial patterns of connections can overlap within the same regions. These anatomical facts have important functional implications. The coexistence of spatial patterns of connections within a single cytoarchitectonic area suggests that neuronal activity may also follow spatial patterns dictated by superimposed connectivities. Overlapping connection patterns may interact in complex ways to enable more elaborate functions.

Spatial patterns of large-scale connectivity are currently ignored within the macro-connectomics research community. My proposal aims to fill a conceptual and methodological gap, and develop new approaches for defining and measuring the spatial organisation of human brain connections. These methodological developments will be carried out along two fronts: (i) development of novel methods for diffusion modelling and tractography; (ii) development of a mathematical framework for the characterisation of spatial connection patterns using in-vivo connectivity data. These novel methods will be useful to all researchers in the neuroimaging community, and will constitute a major advance in the way brain networks are currently modelled and interpreted.

Understanding the human brain is one of the greatest challenges of the 21st century. Several active fields of research are studying the brain at all available scales, from microbiology to systems and behavioural neuroscience. My research is concerned with studying the human brain at a macroscopic, systems level. The current most important question that the field needs to address is how to map the human brain connectome, that is, the set of anatomical connections that support brain function.

A mechanistic understanding of the human connectome is not only important for understanding normal brain function, but will also help tackle the even more complex pathological brain. A number of large projects from both sides of the Atlantic have recently started, with FMRIB Centre being a major contributor in many (e.g. The NIH-funded Human Connectome Project [HCP], The EU-funded "Connect" and "Developing Human Connectome Project" [dHCP] and the UK "Biobank Imaging"). This research area is clearly of great importance for the UK, and worldwide. In December 2012, Science Magazine identified the Human Connectome Project as an area to watch for the coming year.

These large multi-centre projects are centred on acquiring large state-of-the-art data sets. However, without state-of-the-art methods, and new ways of analysing network data, such as those I am proposing, researchers will fail to get the most out of this vast wealth of upcoming data. This project will open new ways of analysing Connectome data. The tools developed, as part of this project, will be shared with the scientific community through our freely available and widely used software FSL. This software is currently used in over 500 universities worldwide, and is contributing to the UK's well-recognised prestige in the field.

The primary beneficiaries of this research are scientists that are interested in systems level neuroscience, and how structure relates to function in the brain. In addition, the methods that will be developed for this project are novel, and will also open a new avenue in methods research and development.