Introductory Workshop on MRI data analysis using Independent Component Analysis, at the Rabinovitch House

The most common method of analysis of fMRI data consists in comparing the BOLD signal in the brain with time series representing the various conditions of the experiment (e.g. task vs rest). Similarly, resting-state data is often analyzed by selecting a region of interest, computing
the average BOLD signal in that region, and comparing it with signal elsewhere in the brain.
Both these methods produce statistical maps (for activation or functional connectivity) which
then give insight into the underlying brain processes.
Attractive by their simplicity, these types of analysis have the drawback of being heavily
assumption-driven and usually do not extract all the interesting information contained in the
data. To complement these, so-called data-driven methods such as Independent Component
Analysis (ICA) have been developed that give a much richer view of the information content of
the data. However, these methods can be a little intimidating as they rely on complex numerical
analysis algorithms, and require special knowledge from the user to be used correctly.
The goal of this workshop is to allow participants to confidently use ICA (as implemented in
MELODIC in the FSL package) to extract and interpret results from their fMRI and resting-state
data. We will first introduce the basic concepts behind the ICA method and how they are
implemented in MELODIC to produce statistical maps. We will then apply ICA to test fMRI and
resting-state datasets and go over many important steps encountered in using ICA in real life
situation: classification of components into signal of interest and artifacts, results interpretation,
data cleanup, network reconstruction and evaluation of functional connectivity changes. We will
also cover group ICA analysis using temporal concatenation ICA as well as tensor ICA and dual
regression.
This will be a theoretical workshop (no practical sessions are organized) and it requires a
minimal background in data analysis to benefit from attending. Participants with nonmathematical
background are encouraged to read through the text (PDF) to introduce
themselves to the necessary notions of matrix algebra.

Lecturer : Thomas Gisiger, Research Associate at the CRBLM

When: Monday, December 5th, 2016, at 10:30-12:30

Where: Rabinovitch House, 3640 Rue de la Montagne, room #101

Note: Please note that this same workshop will also be presented at BRAMS on December 7th. We look forward to seeing you at either event.

 

The workshop is free.
To register, please, fill in the form below. The workshop is limited to 20 participants.