I will give an overview of the current approaches to estimate connectivity from neuroelectrical imaging. I will go through the promises and pitfalls of each approach, in particularly spanning the axis that goes from model-free approaches (transfer entropy, mutual information), to model-based, data-driven approaches (correlations, Granger causality), to biologically inspired models (Dynamical causal models).

The learning objectives will include: (i) a comprehensive overview of freely available toolboxes; (ii) an idea of convenient datasets to benchmark and compare them; (iii) a critical perspective on the advantages and caveats of each approach.

 

 

 

 

I will describe the issues relates to multiplets of variables sharing the same amount of information about the future state of a system, and present some approaches to address them. I will then present applications to neuroimaging and physiological data.

 

%%%%%%%%%%%%%%%%%%%%%%

 

Daniele Marinazzo is Associate Professor at the Department of Data Analysis of the University of Ghent. He obtained his Master Degree in Physics in 2001 and his PhD in Physics in 2007 from the University of Bari, Italy. His main area of expertise concerns methodological and computational aspects of neuroscience research. His research has always been interdisciplinary, being a physicist who has always worked side by side with clinical and experimental neuroscientists, even performing electrophysiology experiments himself, on rats and cats during his postdoc at CNRS in Paris (2008-2011). In 2011 he was appointed as Tenure-Track Professor in Ghent, and has been promoted to the rank of Associate Professor in 2016. His research group studies new techniques for the analysis of physiological data rooted in theoretical physics and machine learning, with a particular focus on retrieving brain connectivity from the analysis of neuroimaging data. He is involved in several fruitful collaborative projects involving advanced data analysis with the departments of neurology, psychiatry and clinical psychology at the University of Ghent. Prof. Marinazzo is also involved in active collaborations with theoretical and experimental groups in Belgium, Spain, Italy, France and China. He teaches techniques of neuroimaging data analysis. He is member of an international consortium aimed to promote exchanges and support to Chinese students (Plan 111). He organizes regularly workshops, conferences and journal special issues around connectivity methods and their applications to neuroscience. He is member of the scientific committee of the Ghent Institute for Neuroscience, serving as an umbrella for interdisciplinary and translational research. Daniele Marinazzo is editor of PLOS One, Network Neuroscience and PLOS Computational Biology, and referee for many journals in the field of neuroscience, applied physics and mathematics.

He strongly advocates open science and science communications.