MNE

Hi,
I am planning to use the MNE software for the EEG data analysis. Data
(both EEG and MEG) was collected with VectorView system at the
University of Technology, Espoo, Finland. We have averaged fif files
preprocessed using Neuromag software (Plotting). We also have MRI data
for each subject (not raw data, but already prepared sets and slices
for the MRILab).
I never used the MNE before, and my the first question is just "How to
start?". It is not clear how to setup the environment variable
SUBJECTS_DIR. Then, is it possible to perform the MNE for the group
average files (EEG data only) to see roughly the differences between
the two groups?
Thank you very much for your kind cooperation,
Irina Anurova,
University of Helsinki (currently working at Georgetown University,
Washington DC).

Dear Irina,

I only use MNE for MEG analysis, I do not use MRILab, so I cannot help with specific details.

However, I strongly recommend that if you are a new user of MNE that you go through the sample tutorial as explained in Chapter 12 of the MNE manual document because then many things will become clear. The sample dataset contains both MEG and EEG data so it should be relevant.

To set up the environmental variable SUBJECTS_DIR for the directory that contains the MRI on a linux-style terminal, you need to type in the terminal:

setenv SUBJECTS_DIR ~/<path-to-your-data-directory>/subjects

SUBJECTS_DIR is the directory in which the anatomical MRI data is stored, and it contains the individual directories for the anatomical data, named after each subject in the experiment. This is explained in section 3.2 of the manual. For example, if my experiment is called Expt1, my SUBJECTS_DIR is .../data/Expt1/subjects. If I have two subjects called subject1 and subject2, their anatomical MRI folders will be located inside the SUBJECTS_DIR: .../data/Expt1/subjects/subject1 and .../data/Expt1/subjects/subject2. Etc.

As I understand it, the noise covariance matrix (NCM, filename *cov.fif), which is necessary for MNE source reconstruction, can only be made from raw files, using the MNE command 'mne_process_raw'. Therefore you might need your raw files as well as your preprocessed average files in order to do your source reconstruction. However, I am not sure so maybe someone else who does EEG will know more about this than I.

Best,
Nela