Re: [Marsbar-users] Getting different results when using "Extract ROI full options"

[Matthew B. <mat...@gm...>]

Hi,

On Thu, Sep 6, 2012 at 11:30 PM, Angela Mayes <ang...@mc...> wrote:

> -          From a localiser task I have extracted out some ROIs based on a one sample t test
>
> -          Used the transform option to combined these ROIs with the aal ROIs
>
> -          Then I want to extract the data for each participant associated with these ROIs for the active or task run (separate task than the localiser where I have modelled the data already and have beta images, contrast images etc, however I plan on using the Beta images to extract out the data)
>
> -          I go to data…extract ROI data full options
>
> -          I select the ROIs I have created (the combined ones)
>
> -          Use SPM design – no
>
> -          Modality of images to scale – other
>
> -          Number of subjects – so I did this as 1 to make sure I got the same values as when I did this as 26 (the actual total number of participants)
>
> -          Select the beta images for either the 1 subject or the 26 (saying done between each participant)
>
> -          Scaling from – raw data
>
>
>
> The whole purpose of me doing this is to extract out the values for each of my ROIs for each participant to put into SPSS so I can complete a MIXED ANOVA taking into consideration laterality and other factors.
>
>
>
> However before even getting it into SPSS when I just look at the values in matlab that marsbar spits out for the two different ways: 1) 1 subject, 2)26 subjects (that includes the subject from 1) the values for the SAME participant are different depending if I extract the ROI values from the Beta images as either a single subject or as the whole 26.
>
>
>
> Does that make more sense?

Yes, that makes a lot of sense.  I think what you are seeing is that
the voxels included in the ROI are those which are not NaN for any
image.  So, if there is any difference in the positions of NaNs across
subjects, then some voxels will be shaved off the pool of voxels over
which MarsBaR takes the mean at each time point.  Do you see what I
mean?

>
> Also am I meant to be getting these responses in email? As  I only knew you had responded because I get the newsletter, and should I be replying on the sourceforge website or something?

How you get the responses depends on how you subscribed.  I always
subscribe asking for emails as they arrive, but another common option
is only getting a digest of emails say once a day or once a week -
maybe you did that subscription option?  I think you can change that
at your subscriber page;  I think you can reach that with:

https://lists.sourceforge.net/lists/listinfo/marsbar-users

>
> Sorry for being a little clueless.
>

Don't worry - thanks for hanging in there even though I'm being slow to respond.

Best,

Matthew