Research

A cross-disciplinary approach to understand the brain

Understanding the brain is hard. I see it as a complex machine where we can’t see exactly what is happening inside, but we can measure it from different sides, and when we do, we get information at many levels: electrical, hemodynamic, behavioral, optical, etc. My research develops statistical and computational methods that try to tie these measurements together, and I have found that connectivity analysis is a versatile tool to explain brain processes and their dynamics.

A multidisciplinary approach

Connectivity analysis, i.e., the analysis of brain networks, requires tools from different fields. As a consequence, my work inherently combines statistics, signal processing, AI, and scientific computing. In my lab, we release all of our toolboxes as open-source software, as I see it as part of working in an interdisciplinary field.

Long-term goals

Coupling between electrical and hemodynamic activity

How do the frequency bands of EEG relate to the slower hemodynamic response measured with fNIRS? What is the direction of that information flow?

Statistical inference for brain connectivity

There is no single connectivity, but a range of approaches to construct networks that inform about different aspects of the brain. Which ones are the most appropriate? How can we quantify their uncertainty? Which networks inferred by deep learning methods are explanatory for neuroscience?

Real-life applications of connectivity

Real-time connectivity estimates can be used in real-life settings, for example, to monitor seizure networks during long-term iEEG recordings. To get there, we need to understand how to estimate networks on commodity computers, without requiring large workstations.

Current work

We pursued through the projects of my lab, and the results are in my publications.