This page is intended to be a broadly accessible discussion of some of the major themes raised by the research with which I have been involved. To date there have been three key themes, although many studies I have worked on stand alone, without falling directly into any category. One theme, which was the primary focus of my PhD thesis, is the mainly technical problem of repeatably finding specific white matter structures in the brain, which needs to be solved to facilitate other research. The second is a more practical question, of how the brain’s “wiring” changes as the body develops and gets older. Most recently I have been concerned with how to study the brain network as a whole, combining different sources of information to get a more complete picture than we could do using any one source alone.

These are discussed further below, with references to relevant published work from my group and me. Of course, these are active research areas under investigation by many groups, and fuller discussion of the relevant scientific literature can be found in each cited paper.

Finding specific brain structures

Diffusion magnetic resonance imaging (MRI) is a rich source of information about brain structure, particularly the white matter which forms the brain’s “wiring”, connecting areas of grey matter together. This medical imaging technology is available for the clinical MRI scanners installed in many hospitals, but its full clinical potential has probably not been realised yet. In particular, identifying the exact pathway of a specific white matter “tract” is hard to do consistently from one brain to the next, not least because these structures are three-dimensional and often twist and turn considerably. However, each brain generally contains the same set of major tracts, and each tract generally connects the same grey matter regions via essentially the same route.

Taking advantage of this similarity in tract shape across individuals, we developed a computational method which uses a “reference tract” to provide an archetypal trajectory for each structure. The tract of interest is then identified in each subject by using an algorithm to find the trajectory with the most similar shape to this reference. Proof of concept for this approach, called “neighbourhood tractography”, was demonstrated using a relatively simple version of the algorithm [Ref. 1]. A more sophisticated version of the algorithm was subsequently developed and refined, which overcame several limitations in the simpler initial version [Refs 2 and 3]. We later demonstrated how the same general approach could be used to “tidy up” the virtual representations of each tract after they have been identified [Ref. 4].

Tractography graphic

Unsupervised segmentation of the forceps major in four healthy adults across repeated scans [Ref. 3]

These methodological developments have been very useful in a number of research studies in clinical and nonclinical neuroscience, including those discussed in the next section. Implementations of the various algorithms are freely available in the TractoR software package [Ref. 5].


  1. J.D. Clayden, M.E. Bastin & A.J. Storkey (2006). Improved segmentation reproducibility in group tractography using a quantitative tract similarity measure. NeuroImage 33(2):482–492. [pdf (1061 KiB); PubMed; DOI direct link]
  2. J.D. Clayden, A.J. Storkey & M.E. Bastin (2007). A probabilistic model-based approach to consistent white matter tract segmentation. IEEE Transactions on Medical Imaging 26(11):1555–1561. [pdf (1040 KiB); PubMed; IEEE Xplore]
  3. J.D. Clayden, A.J. Storkey, S. Muñoz Maniega & M.E. Bastin (2009). Reproducibility of tract segmentation between sessions using an unsupervised modelling-based approach. NeuroImage 45(2):377–385. [pdf (1724 KiB); PubMed; DOI direct link]
  4. J.D. Clayden, M.D. King & C.A. Clark (2009). Shape modelling for tract selection. In G.-Z. Yang, D.J. Hawkes, D. Rueckert, A. Noble & C. Taylor (eds), Proceedings of the 12th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI). Lecture Notes in Computer Science, vol. 5762, pp. 150-157. Springer-Verlag. [pdf (1056 KiB); DOI direct link]
  5. J.D. Clayden, S. Muñoz Maniega, A.J. Storkey, M.D. King, M.E. Bastin & C.A. Clark (2011). TractoR: Magnetic resonance imaging and tractography with R. Journal of Statistical Software 44(8):1–18. [pdf (1876 KiB); Journal of Statistical Software; 2016 updated version]

Development, ageing and the connected brain

Brain development is a very complex and long drawn-out process, which continues well beyond birth. In particular, the formation of myelin around bundles of neuronal axons is reported to be a process which continues into adolescence, and possibly right up to the onset of adulthood. (This is an important aspect of brain development because the myelin sheath enables signals to be sent rapidly and reliably through the longer connective pathways.) It is by now well established that changes in the characteristics of white matter which are visible to diffusion MRI accompany normal postnatal development, indicating that white matter tracts gradually gain structural “coherence” over the years. Roughly opposite changes occur during one’s old age.

A few research groups have reported that many tracts tend to develop in tandem with one another. This may not seem surprising, since the brain grows in size during childhood and development of different parts of it may be expected to proceed in parallel, but the ability to use medical imaging to “observe” changes to white matter microstructure is very valuable. In our work, we have been able to separate out groups of tracts which “vary together”, and look at their relationships with age, gender and intelligence [Ref. 1]. We found that, over the age range of 8 to 16 years, age is the primary driver of change to white matter “integrity”, as expected, but the trajectory of change is substantially different between the two sexes. Even more interestingly, we observed a link between more subtle components of the diffusion MRI–based tract characteristics that we measured, and intelligence. These latter components were secondary to the age effect, and independent of age and gender. We have also shown a similar relationship between integrity in a group of tracts and cognitive processing speed in old age [Ref. 2], as well as demonstrating that childhood intelligence has a substantial bearing on intelligence and the state of white matter in old age [Ref. 3]. These consistent links between particular white matter tracts and intelligence in early and late life paint a fascinating picture of the role of neural communication in facilitating cognition.

Tract development graphic

Cross-sectional age relationships of mean diffusivity in late childhood [Ref. 1]

An additional avenue of enquiry has been to use the methodological framework described above to look at morphological brain changes in later life. As we age, the brain tends to atrophy somewhat, changing the shape and arrangement of brain structures. The tract modelling framework that we developed provides one way of quantifying changes in tract morphology over time, and we have shown the effects to be particularly obvious in the frontal part of the corpus callosum, which connects the two brain hemispheres together [Refs 4 and 5].


  1. J.D. Clayden, S. Jentschke, M. Muñoz, J.M. Cooper, M.J. Chadwick, T. Banks, C.A. Clark & F. Vargha-Khadem (2012). Normative development of white matter tracts: Similarities and differences in relation to age, gender and intelligence. Cerebral Cortex 22(8):1738–1747. [pdf (1632 KiB); PubMed; DOI direct link]
  2. L. Penke, S. Muñoz Maniega, C. Murray, A.J. Gow, M.C. Valdés Hernández, J.D. Clayden, J.M. Starr, J.M. Wardlaw, M.E. Bastin & I.J. Deary (2010). A general factor of brain white matter integrity predicts information processing speed in healthy older people. The Journal of Neuroscience 30(22):7569–7574. [PubMed; DOI direct link]
  3. I.J. Deary, M.E. Bastin, A. Pattie, J.D. Clayden, L.J. Whalley, J.M. Starr & J.M. Wardlaw (2006). White matter integrity and cognition in childhood and old age. Neurology 66(4):505–512. [PubMed]
  4. M.E. Bastin, J.P. Piątkowski, A.J. Storkey, L.J. Brown, A.M. MacLullich & J.D. Clayden (2008). Tract shape modelling provides evidence of topological change in corpus callosum genu during normal ageing. NeuroImage 43(1):20–28. [PubMed; DOI direct link]
  5. M.E. Bastin, S. Muñoz Maniega, K.J. Ferguson, L.J. Brown, J.M. Wardlaw, A.M. MacLullich & J.D. Clayden (2010). Quantifying the effects of normal ageing on white matter structure using unsupervised tract shape modelling. NeuroImage 51(1):1–10. [PubMed; DOI direct link]

Multimodal brain networks

The interconnections between regions of the brain are vital to its function, and several neurological diseases are thought to be at least partly attributable to breakdowns in these communication channels. Medical imaging provides multiple windows into this network of connections, often called the “connectome”, and it has recently become common to represent these in terms of an abstract “graph”. I have reviewed the processes and pitfalls involved in using so-called structural data to reconstruct these connectomes [Ref. 1], but networks can also be built up using functional images, by looking for pairs of regions which are consistently active at similar times.

Graph reconstruction stages

A cortical parcellation, combined with diffusion tractography, can be used to build a structural “connectome”, an abstract representation of interconnections in the brain [Ref. 1]

The structural and functional connectomes should be complementary, in the sense that they are two (admittedly limited) views of the same system. There is therefore a strong drive towards understanding how to combine the information most effectively. For our part, we have investigated the nature of the relationship between functional connectomes derived from different modalities [Ref. 2], and between structural and functional connectomes [Ref. 3], by attempting to predict one from the other consistently.

Although graphs in general are well understood—social networks, such as Facebook connections, are described using the same structures—there are reasons for treating brain networks specially. One such reason is that cognitive tasks do not necessarily use the whole brain, relying instead on functionally specialised subnetworks. One of our more technical developments was therefore to explore a novel way to subdivide brain networks, producing so-called “principal networks” [Ref. 4]. This technique is now available in TractoR, and was quickly put to use in a study of children [Ref. 5].


  1. J.D. Clayden (2013). Imaging connectivity: MRI and the structural networks of the brain. Functional Neurology 28(3):197–203. [pdf (9880 KiB); PubMed; Functional Neurology]
  2. F. Deligianni, M. Centeno, D.W. Carmichael & J.D. Clayden (2014). Relating resting-state fMRI and EEG whole-brain connectomes across frequency bands. Frontiers in Neuroscience 8:258. [DOI direct link]
  3. F. Deligianni, D.W. Carmichael, G.H. Zhang, C.A. Clark & J.D. Clayden (2016). NODDI and tensor-based microstructural indices as predictors of functional connectivity. PLoS ONE 11(4):e0153404. [PubMed; DOI direct link]
  4. J.D. Clayden, M. Dayan & C.A. Clark (2013). Principal networks. PLoS ONE 8(4):e60997. [pdf (1520 KiB); PubMed; DOI direct link]
  5. J. Bathelt, H. O'Reilly, J.D. Clayden, J.H. Cross & M. de Haan (2013). Functional brain network organisation of children between 2 and 5 years derived from reconstructed activity of cortical sources of high-density EEG recordings. NeuroImage 82:595–604. [PubMed; DOI direct link]