[Mne_analysis] averaging in MNE

[Matti Hamalainen]

HI Gustavo,

On Apr 13, 2009, at 10:39 AM, Gustavo Sudre wrote:

> Hello,
>
> I have a couple questions regarding averaging MEG data with MNE and  
> I was wondering if you could help me.
>
> 1) Is it possible to average a certain event only if something else  
> happened after it? For example, let's say I have conditions 1 and 2.  
> I want to average each condition, but only when the trigger happened  
> and I received a response from the subject, which is coded as a 3 in  
> the trigger line. In other words, average all the occurrences of 1  
> or 2 only when there's a 3 after them. I wrote a script to do that  
> in Matlab using the MNE toolbox, but since then lots of people have  
> joined the lab and they are not very familiar with matlab, so I was  
> wondering if there was a way to do such average within MNE, either  
> by using the GUI or a combination of .ave and .eve files.

Unfortunately, there is no direct way to do conditional averaging in  
mne_process_raw/mne_browse_raw (yet). However, the easiest way around  
this is to output a text format event file a and manipulate this file  
so that new event numbers are assigned to the events meeting the  
requested combinations. I guess such a script could be even written in  
Excel.

> 2) Something else my script does is to average conditions across  
> different fif files. In the example above, I could have several fif  
> files with such events, and the final .fif averaged file would  
> represent the averages across all files. I imagine this is possible  
> to do with the MNE software by using the global average parameter.  
> Is that correct? Is there a different way to do it besides  
> specifying a different .ave file for each of my raw .fif files? It's  
> usually not a problem to do that for only a couple .fif files, but  
> when there're about 10 different .fif files to be averaged, it  
> starts to get a little cumbersome to keep track of all the .ave  
> files, which look very similar except for the output parameter.

You can, indeed, make a grand average file using mne_process_raw. You  
can use the --saveavetag option to create the output file names  
automatically. Also, if mne_process_raw runs out of ave file name  
names when processing the raw files, the last one available will be  
used. If you use custom event files, you can specify them on the  
command line (in version 2.6).

Let's assume

my.ave			is your averaging file
my1_raw.fif and my2_raw.fif		are your raw data files
my1.eve and my2.eve 			are your event files

You can do

mne_process_raw <other options>  --raw my1_raw.fif --raw my2_raw.fif -- 
eve my1.eve --eve my2.eve --ave my.ave --saveavetag -ave --gave my- 
gave.fif

This will produce the averages in my1-ave.fif and my2-ave.fif and log  
files my1-ave.log and my2-ave.log independent of what is specifie in  
my.ave. Also, the my1.eve and my2.eve override the event file  
specification (if any) in my.ave.

Similarly, the --savecovtag option can be used to generate the noise- 
covariance calculation output file names.

I hope this helps.

- Matti






-------------

Matti Hamalainen, Ph.D.
Athinoula A. Martinos Center for Biomedical Imaging
Massachusetts General Hospital
Building 149, 13th Street, Mailcode 149-2301
Charlestown, MA 02129-2060
USA

e-mail          msh at nmr.mgh.harvard.edu




-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20090414/291ac729/attachment.html