He is currently assistant professor at Telecom ParisTech and scientific consultant for the CEA Neurospin brain imaging center. My work is on statistical machine learning, signal and image processing, optimization, scientific computing and software engineering with primary applications in brain functional imaging (MEG, EEG, fMRI). Before joining Telecom ParisTech, I worked at the Martinos Center for Biomedical Imaging at Harvard in Boston, and for five years at INRIA in the Parietal Project Team and the Athena Project Team. I am also an active member of the Center for Data Science at Université Paris-Saclay.
Alexandre Gramfort
PRNI 2016 - TUTORIAL 5
MNE Python: sensor- and source-space analysis of M-EEG data
 |  |  |  |
|---|---|---|---|
 |  |  |  |
 |  |  |  |
MNE (http://martinos.org/mne) is a software package for processing magnetoencephalography (MEG) and electroencephalogra-phy (EEG) data. It provides a comprehensive workflow for data preprocessing, forward modeling (with FreeSurfer) using boundary element models (BEM), source imaging using distributed source models, time–frequency analysis, non-parametric and multivariate statistics, decoding and connectivity measures. All analyses can be conducted either in the sensor or in the source space. The aim of this tutorial is to present what MNE can do starting from raw data using the Python programming language. The tutorial will be hands on based on IPython notebooks so the attendees can run the full analysis on their machines and experiment with the tool, eventually on their own MEG or EEG data.
MONDAY, JUNE 20th
08:30 :: Registration
09:00 :: Introduction to MEG, EEG and MNE
Content :
- Lecture on MEG/EEG basics (physics, acquisition, signals vs. noise)
Objectives :
- Have the background on MEG/EEG signals and electrophysiology
- Discover what is MNE (origin, project, people)
- Setup MNE-C & Looking at raw data with with mne_browse_raw (use interactive SSPs, set filtering parameters)
- Identifying what is signal and what is artifacts
11:00 :: Sensor space analysis with MNE
10:30 :: Coffee Break
Content :
- Hands-on with MNE-Python and jupyter notebook
Objectives :
-- From Raw to Evoked
- Scripting preprocessing (filtering, artifact rejection, maxwell filter)
- Visualization of Raw, Epochs and Evoked data
12:30 :: Lunch Break
13:30 :: Source localization basics: MNE/dSPM
Content :
- Quick lecture on MEG/EEG source localization methods
(theory behind noise covariance and minimum norm, source spaces)
- Hands-on with notebook
Objectives :
-- Do a coregistration with mne_analyze
- Freesurfer cortical source spaces visualization with freeview
- BEM / Forward modeling
- Do noise covariance estimation & inspection
- Inverse operators to STC
- Viewing STCs with mne_analyze and scripting in python to make nice figures
15:00 :: Coffee Break
15:30 :: Source localization basics (continued)
Content :
- Hands-on with notebook
Objectives :
- Beyond cortical source spaces (volume, discrete)
- Beamformers on volume grid
- Dipole fitting
TUESDAY, JUNE 21st
09:00 :: Decoding / Machine Learning with MEG/EEG
Content :
- Lecture on introduction to machine learning and decoding on MEG/EEG
- Hands-on using MNE and Scikit-Learn
Objectives :
- Understand what is decoding. Pro and cons.
- Know how to use MNE-Python decoding module and scikit-learn to do time by time decoding and time generalization
11:00 :: Time-frequency Analysis
10:30 :: Coffee Break
Content :
- Lecture on time-frequency analysis (power spectrum density estimation, induced activity, ERS, ERD)
- Hands-on with MNE-Python
Objectives :
- Know the different strategies to highlight time-frequency content in MEG/EEG data
(Welch method, Morlet wavelets, Multi-taper)
- Know how to baseline such data
12:30 :: Lunch Break
13:30 :: Independent Component Analysis (ICA)
Content :
- Short lecture on ICA
- Hands-on with MNE-Python
Objectives :
- Understand the basic theory behind ICA
- Know how to correct MEG/EEG biological artifacts with ICA
15:00 :: Coffee Break
15:30 :: Reporting and presenting your neuroscience results
Content :
- Presentation of the mne-report tool
- Presentation of matplotlib to make publication ready figures
- Questions time
Setting up your system
Download and install MNE following the instructions at:
http://martinos.org/mne/stable/getting_started.html
You’ll need a Linux or Mac to have the MNE-C working but you’ll be able to do most of the course with a windows machine.
Download and install Anaconda with Python 2.7:
https://www.continuum.io/downloads
Install mayavi:
conda install mayavi
Install the latest mne-python release and pysurfer by running in a terminal:
pip install --user mne
pip install --user pysurfer
pip install –user nibabel
Download MNE sample data (auditory), SPM faces data, and MNE somato data by running the following 3 examples:
http://martinos.org/mne/stable/auto_examples/plot_from_raw_to_epochs_t o_evoked.html
http://martinos.org/mne/stable/auto_examples/time_frequency/plot_time_fr equency_sensors.html
http://martinos.org/mne/stable/auto_examples/datasets/plot_spm_faces_d ataset.html
Download MNE command line tools for Linux or for Mac (only sorry)
For additional information you may contact tutorial5@prni.org