A guide for running standard analysis steps at the center.

Starting guide

Data at the CENIR is stored on the ICM network. Below are guides to find and work on your data as needed.

• Were are my data: ICM data storage and environment
• Organizational recommendations: subjects, scripts, where and how.
• My very first steps on Linux at CENIR
• Software: where, what.
• Reporting guidelines: the COBIDAS recommendations for MEEG experiments

Each of the three main external toolboxes used at the MEEG center have their own environment and way of thinking about the data. We strongly advice you to choose wisely once and for good for a given project.

See this page for guidance: Choose your toolbox

The document below provides a complete processing pipeline with in depth description (using the MNE toolbox).

Data acquisition and preprocessing

Specifics of data acquisition, modalities and hardware at the MEEG center require your attention.

Data acquisition and fusion

At the MEEG center, you can acquire data in various modalities (MEG, EEG, eye tracking, fNIRS, MRI...). Each of these has its own specifics (file format, sampling frequency, number of channels, coordinate system...). We have in-house tools to merge data from different modalities within a consistent framework accepted by all external toolboxes.

• Anatomical data should be acquired at the center following the specifications on this sheet Fiche_protocole_generique_Anat_MEG_Prisma_Verio_v0.pdf
• If you use MEG, all data is converted from vendor-specific file formats to the widely used .fif file format, originally implemented by Elekta, our MEG vendor.

• If you use EEG (and no MEG), you may use a different file format. Laurent HUGUEVILLE
• Eyetracking data is merged with MEG with perfect synchrony right at acquisition. You do not need to do anything. In your MEG data, there are three additional channels of type 'MISC', with numbers 007 008 and 009.
• Eyelink basic preprocessing: detecting saccades and creating associated markers with datahandler
• Conversion Laurent HUGUEVILLE
• polhemus Laurent HUGUEVILLE
• synchro (markers, conditions...), Laurent HUGUEVILLE (see this page)
• fusion Laurent HUGUEVILLE

Preparation

• Maxfilter, (a.k.a. tsss): Elekta's noise cancellation and head realignment
  

• T1 MRI segmentation
• headmodel and source space (grid for beamformer) creation in FieldTrip

Cleaning

• Detect and mark bad data segments
• Use ICA to correct blink and cardiac artifacts

• Rereference

• A warning about filters!

Time-locked data analysis

Epoched data uses segments of data recorded during repeated occurrences of an event.

• ERPs (with FieldTrip)
• Time frequency analysis: See the first part of this tutorial.
• Sources:
  
  • How do dipolar sources project to sensors?
  • Evoked potentials: MEG wMNE analysis / EEG wMNE analysis
  • Single trials: MEG D.I.C.S (one subject), EEG D.I.C.S (one subject), MEG D.I.C.S (group analysis), MEG LCMV (virtual channel)
• Connectivity: D.I.C.S (cortico-muscular coherence)

Continuous data analysis

Continuous data uses unsegmented streaks of data recorded during (usually) long periods without specific events.

• Resting state
• Epilepsy

Experimental techniques

Tests and evaluation of new experimental approaches.

• Combination of magnetometers and gradiometers in the inverse problem using D.I.C.S in FieldTrip
• Creating surrogate data: swapping cardiac beats across trials mariana.baborebelo (Unlicensed)

Statistical analysis

Statistical tests are closely related to the experimental design

• Single vs multiple factors Lydia SAFIEDDINE (Unlicensed)
• Group vs Subject analysis Lydia SAFIEDDINE (Unlicensed)
• Sensor space Lydia SAFIEDDINE (Unlicensed)
• Source space Lydia SAFIEDDINE (Unlicensed)
• Time Frequency domain Lydia SAFIEDDINE (Unlicensed)