How individual connectomes got me through my dissertation defence


A few days ago, I successfully defended my PhD thesis (with no corrections!). The sense of relief and high that comes with the culmination of 5 years is quite hard to describe. I found myself surprised to be *almost* enjoying the defense. What began as one of the more stressful days I have encountered ended with a tremendous feeling of accomplishment, and of course, Cevapi and Vodka shots ;)

The central tenet of my work is to understand individual variability in human brain connectomes (connectome = the brain simplified as a matrix, a network of nodes – individual regions, and edges – the connections between regions). I study how individual structural connectivity (the physical white matter pathways) and functional connectivity (information exchange that happens along these pathways) are linked. I find that, despite the common assumption that individual structural connectivity is the best predictor of individual functional connectivity, this is actually not necessarily the case. See recent blog post by Randy and myself and here is the paper. Knowing this, I set out to understand if we can still use the individual variability that exists in connectomes to explain individual differences in cognition. And we found that, indeed we can! What’s more, the brain patterns that express cognitive variability in structural and functional connectivity don’t overlap. This means that structure and function each provide unique and distinct information for cognition. Basically, we need both! (Check out the paper; Zimmermann et al., 2018a). But we didn’t stop there. Acknowledging the limitations of individual connectomes, yet having shown that there is meaningful individual variability, I wanted to devise a way to study structure and function together within individual subjects. And what better way than TheVirtualBrain. By integrating the two modalities together at multiple scales within individuals, we showed that biophysical features (such as interactions between long-distance regions, and between individual neuronal populations) can actually predict individual cognition even better than the individual structural and functional connectomes themselves! (see Figure 1 below and the paper; Zimmermann et al., 2018b). This was very exciting, because it shows the potential for a biophysical connectome-based computational model like ours to answer challenging questions about individual differences.


Figure 1. Integration of individual SC/FC within TVB workflow (left), model biophysical parameters covariation with cognition exceeds individual SC/FC

A few more words before I go… Reflecting back on my past 5 years @ the McIntosh lab, I realize how fortunate I am to have had the experiences I did – an incredible mentor that is supportive yet at the same time gave me the freedom to explore my own path, great peers, and two research fellowships (@UCI with Ana Solodkin and @Charité with Petra Ritter, both wonderful mentors and friends that made me feel very much at home). So I wanted to end off by thanking everyone who made me experience what it was! I’m looking forward to continued work with all of you.

McLab Icon_C8A698 mini.png