This is a collaborative space. In order to contribute, send an email to maximilien.chaumon@icm-institute.org
On any page, type the letter L on your keyboard to add a "Label" to the page, which will make search easier.

Philosophie generale

Preprocessing

1. Artefact detection and/or correction

SSS and tSSS guidelines

SSS and tSSS are used to pre process our Elekta MEG Data. Using SSS or tSSS (through the maxfilter command) is a required step to get rid off most the environmental noise of our system. Both methods modelize the MEG signal by projection on mathematical functions (spherical harmonics) able to separate internal (brain) and external (environmental noise) sources of magnetic field. tSSS, by correlating internal and external decompositions of the signal, is able to get rid of strong external noise sources (as the elevator behind the MEG room) that are not sufficiently attenuated by regular SSS. It is so an alternative to the PCA decomposition to solve this noise problem.

You will find below few points to insure that maxfilter is correctly used and to insure a good processing.

  1. SSS and tSSS processed signals are not comparable (from an amplitude and noise level points of view). If you plan to pool data from several subjects you have to use the same method (either SSS or tSSS) on all your subjects. 
  2. tSSS specific configuration: To avoid any discontinuity in the tSSS output, you have to set the analysis buffer length to the duration of your run by adding the duration in second after -st option (for example -st 600 for a 10 minutes run). When tSSS is used the automated bad channel detection is switched off. Saturated channels and static bad channels are still automatically excluded. See 2) point below, use -bad option if necessary.
  3. Always check the output signals: Automatic bad channels detection does not always work and it may miss channels misbehaving (channels with few steps for example). These remaining artifacts will be spread on all you channels by sss or tSSS. The easiest way to check for that, is to take a look at your signals after SSS or tSSS processing. If one or several bad channels were not detected you must rerun the maxfilter command with an explicit declaration of the misbehaving  channels using the -bad option. 
  4. Use the correct calibration files: SSS or tSSS use specific calibration files to fine tune the processing. These files may change during a preventive maintenance of the system (usually performed once a year in september). The default command line "maxfilter" takes by default the current calibration files. It is perfectly ok if you are processing newly acquired data. If you are (re)processing old raw data, chance is high that the calibration files are not the good ones! First point, the change in the SSS or tSSS output between two calibration is very small - do not panic. If your files are older than April 6th, 2013 ask to the MEG team. Otherwise for a given subject and a given date (for example in my_exp_my_subject_04/20130610) you should find a directory called sss_config together with your raw data. This directory contains the necessary sss calibration files. You can use the maxfilter_cenir alias to apply sss. This alias will automatically find the calibration file in sss_config and will use them to apply SSS or tSSS.
  5. Do not hesitate to contact the MEG/EEG team if you have any question!
Blinks, saccades, ECG, muscle , alpha

Here

2. Managing Markers

Decoding the subject's responses
Combining markers
Analyzing behavior

3. Analyzing data at the sensor level

Computing ERFs
 

Important note concerning Magnetometers and Gradiometers (how to average signals)

Computing Time-frequency plots
Visualizing data
Statistical analysis
  • Defining spatial regions and temporal windows of interest
  • Cluster-based approach
 

4. Source localization

  • Minimum Norm or Beamformer?
  • Toolboxes