Ilaria Carannante: Beyond pairwise connectivity: topological data analysis from striatal microcircuits to whole-brain connectomes

Abstract.

The brain can naturally be modelled as a graph: neurons (or brain regions) are the vertices, and the connections between them are the edges. However, conventional connectivity analysis often focuses only on pairwise interactions, overlooking higher-order structures that may be important for understanding network function and disease stages.

In this seminar, I first present an in silico model of the striatal microcircuit¹, in which the connectivity is generated from reconstructed single-neuron morphologies, and higher order structure is captured by directed cliques. We analysed both this healthy network and its counterpart in Parkinson’s disease.
I then describe ongoing work that pushes this analysis further, examining for example how neurons change their role within cliques and quantifying their organization using measures such as entropy.

Finally, I focus on my main postdoctoral project, concerning human whole-brain structural connectomes reconstructed from diffusion MRI tractography. Here we aim to determine whether topological representations of white-matter networks can reveal distributed, disease-specific alterations that conventional connectivity measures may miss.

¹ The striatum is a deep-brain nucleus central to movement, and one of the regions most affected by Parkinson’s disease.