Thomas Bolton

Thomas (Figure 1) was born in Lausanne, Switzerland, and spent his youth in the Morges area, where he perfected his knowledge of fine Swiss chocolate and developed a strong interest towards racket sports. After completing his high-school track at Gymnase de Marcelin, it was time to decide what university studies to pursue; as a very rational individual holding full certitude about his future, Thomas resorted to the only possible decision for this purpose: a heads and tails assessment (with unbiased coin), which kicked off a Bachelor cycle in Life Sciences and Technology at Ecole Polytechnique Fédérale de Lausanne.

Following many headaches and numerous peaks of stress uniformly distributed across exam sessions, Thomas completed his university studies, going for a Master in Neuroscience; this time, he did not require a coin toss to decide what to do, as neuroscience clearly stood out as the most attractive option.

Having greatly enjoyed his Master project, done in the field of electrophysiology (Seeking for a role of Synaptotagmin 9 in exocytosis), Thomas started a PhD in the Laboratory of Sensory Processing, where he was to carry out in-vivo electrophysiology experiments to probe the dendritic activity of pyramidal cells from the mouse barrel cortex. Perhaps frightened by the very complicated running title of his thesis, he decided to redirect himself after 10 months, and to move towards more practically applicable work that would not involve animal experimentation, but would instead put his engineering skills to the test.

Despite stupidly miswriting the time of his first meeting with his future professor, Dimitri van De Ville, and thus arriving two hours late for the appointment in front of a closed door, Thomas had the chance to be hired in the Medical Image Processing Laboratory, first for a three-month internship in early 2015, and then for a PhD which commenced in early May 2015.

His general research topic is the development and application of dynamic functional connectivity (dFC) tools to better understand neurodegenerative and neurodevelopmental brain disorders. To word this into more understandable terms that a novice reader can easily grasp (yes, Thomas also does his best to be a nice guy, and so, he really cares about people understanding what he writes in his profile page, aside from trying to make it entertaining), the main idea behind dFC is the fact that the brain is always highly dynamic: at a certain time point, specific sets of regions will be synchronised in activity, and those interactions will dynamically evolve over time. One wishes to capture this evolution, which may possibly differ in a diseased brain.

More specifically, the main axes of Thomas' work so far include:

  • Synthesis of the dFC literature existing to date, isolating key axes of methodological development and suggesting novel avenues for future developments in the field

Figure 2: An overview of most dFC analytical efforts to date.

  • Application of inter-subject functional connectivity (ISFC), a dFC approach in which connectivity between brain regions is tracked over time through cross-subject measurements (which attenuates sources of connectivity changes that are uncorrelated across subjects, i.e. noise and spontaneous changes), to a dataset of individuals diagnosed with autism spectrum disorders who all watched an audio-visual scientific documentary. The goal here is to extract movie-driven changes in brain activity that may distinguish autistic subjects from healthy controls

Figure 3: ISFC analysis.

  • Implementation of a first toolbox version for the application of co-activation pattern analysis, a frame-wise dFC tool with which the different networks co-activating with a chosen seed region can be disentangled and analysed in terms of spatial and temporal features
  • Application of co-activation pattern analysis to subjects at risk of developing schizophrenia, and to bipolar subjects lying in different mood states, to try and extract dynamic neuronal correlates of those conditions (done in collaboration with Diana Wotruba, ETHZ, and Gwladys Rey, UNIGE, respectively). Those works are done on resting-state data, for which scanned subjects were explicitly asked not to do anything particular (i.e., they probably underwent mind wandering, for instance by wondering when that tedious scanning session would finally end; believe it or not, a lot of relevant information can be extracted from such data)

Figure 4: co-activation pattern analysis.

  • Application of the total activation (TA)/innovation-driven co-activation patterns (iCAPs) pipeline to the same dataset as in point 2. TA is a dFC approach performing deconvolution of the signals acquired with fMRI, so that the signals are undone from hemodynamic effects; through sparsity-pursuing regularisation, cleaned signals with a sparse set of changes (reflecting the activation or deactivation of brain networks) are recovered. The frames containing those changes are then clustered into a set of iCAPs, which stand for the brain networks of interest

Figure 5: total activation/innovation-driven co-activation patterns.

  • Development of a novel dFC methodology enabling to assess cross-couplings between different brain networks, that is, to understand whether some brain systems, when turning active, will modulate the propensity of others to become active or deactive. Briefly, the approach involves the modelling of network activity time courses through hidden Markov models, which are also allowed to influence each other through sets of modulatory coefficients describing those cross-network couplings. To keep the dimensionality of the problem affordable, and fit with our physiological expectation that only subsets of networks will influence each other, sparsity in modulatory influences is imposed through an L1 regularisation constraint. Upon preliminary analyses on resting-state data, such cross-network couplings do exist, and may be promising targets as biomarkers of neurodegenerative disorders.

Figure 6: sparse coupled hidden Markov models.

Aside from those projects, Thomas is also attempting to follow up on his hobbies outside of the laboratory, which involve a great interest for the Japanese language and culture, as well as an extended knowledge of most tennis-related facts from the past 20 years (in particular when they involve Roger Federer, indisputably the greatest player of all time; see Figure 7).

Figure 7: Roger Federer waving his willsome wand at (from left to right): Australian Open 2017 (W,A), London Olympics 2012 (F), World Tour Finals 2015 (F,A), Cincinnati 2015 (W), World Tour Finals 2010 (W), Indian Wells 2015 (F). W, winner; F, finalist; A, attended part of the tournament myself.


Figure 1: Thomas Bolton. Glasses chosen with the expert advice from Selin Anil and Lorena Freitas; hair style and knitted scarf courtesy of Lorena Freitas (yes, we also like helping each other and giving each other nice presents in the laboratory).