Fusing connectivity information

The main goals of this project are to provide tools:

  1.  To acquire more efficiently diffusion MR images: obtaining the diffusion MR images required for this project requires long acquisition times, and high diffusion gradient strength to fulfill the Nyquist conditions, which is feasible but not clinically practicable, particularly so as the subject may move during the dMRI session (in which case the data is compromised). Using compressed sensing to reduce dMRI acquisition times is a promising technique which we seek to develop in this project. The sampling strategy has been extensively studied for Q-Ball Imaging, but so far, very few studies have addressed the question of optimal experiment design for multiple Q-shell acquisition. We have just started to explore some related ideas and would like to pursue exploring this work with our partners within this project.
  2.  To use the structural information provided by dMRI to define better models and regularization schemes for spatio-temporal MEEG source reconstruction: integrating the dMRI information in the resolution of the ill-posed inverse problem in MEEG. In general, regularization is based on prior information mostly based on parcellation of the cortex. This spatial prior can be enlarged in order to take into account the anatomical connectivity between these parcels using information provided by dMRI.
  3. To use MEEG data to better understand the task-dependent spatio-temporal structure of connectivity patterns: the way the “brain network” is used is highly dependent on the task at hand. From the structural connectivity prior in the inference model for localizing the functional sources and networks, we estimate the causal interactions between the sources from EEG, MEG and combined MEG+EEG, dMRI being used as a spatial prior for the inverse problem.