Dear Tom and Bettyann thanks a lot for your valuable comments!
I would like to confirm that I understand what you wrote and commented previously. So my understanding is that I would need to run a second level analysis ON EACH subject SEPARATELY using a paired t-test,
i.e, ONE second level FEAT for subjects 1:
input pre_subj_1 and post_subject 1; design paired-t-test; output: COPE1 pre> post for subject_1, COPE 2: Post > pre for subject 1, and doing that for each subjects. In that design I would like to model the mean positive correlation and negative correlation So, COPE 3 mean pre_condition, COPE 4: mean _post_condition
Let's say I have four subjects, I would then have to run two randomize separately as follow:
1) randomise -i COPE1 pre> post for subject_1 COPE1 pre> post for subject_2 COPE1 pre> post for subject_3 COPE1 pre> post for subject_4 -o pre>post.nii -d single_group_average.mat -t single_group_average.con -T -m <mask>
2) randomise -i COPE2 Post > pre_for_subject_1 COPE2_Post > pre_for_subject_2 COPE2_Post > pre_for_subject_3 COPE2 Post >pre_for subject_4 -o Post >pre.nii -d single_group_average.mat -t single_group_average.con -T -m <mask>
Then if I want to mask the positive correlation and negative correlation I would need to mask the output from randomize as follow:
pre>post.nii masked with COPE 3 mean pre_condition (mean form second level analysis)
Post > pre masked with COPE 4: mean _post_condition ((mean form second level analysis)
Questions:
A) Am I on the right track?
B) Is it the same to input in the second level analysis in FEAT all of the fist level analysis together as explained at:
http://www.fmrib.ox.ac.uk/fslcourse/lectures/practicals/feat2/index.htm instead of doing the second level analysis for each subjects separately (as I described above)?
C) Following the web page http://www.fmrib.ox.ac.uk/fslcourse/lectures/practicals/feat2/index.htm, and the example described above, Is it possible to finish the analysis in the second level and to try to publish those results? If not, Why is it necessary to perform a third level analysis in this case?
I would really appreciate your comments and advice on this matter
Thank you very much
Lorena Jiménez Castro, MD
________________________________________
Re: [FSL] Round 2: Change in functional connectivity between pre- and post-conditions using randomise
FROM: bettyann
TO: [log in to unmask]
Correction on my previous post ...
Since my experiment consists of two conditions -- Pre and Post -- and I want to model the _difference_ between these conditions, I need to follow the 'Paired t test' example shown at:
http://www.fmrib.ox.ac.uk/fslcourse/lectures/practicals/feat2/index.htm
This practical states after the bullet description of 'Level 2':
Note that as well as the COPEs, FEAT passes the variance of these COPEs ("varcopes"), and even the uncertainty in the variance of these COPEs ("tdofs" - degrees-of-freedom), between the different levels.
This is what I want. Once I complete this paired t-test 2nd level analysis, I can then do the 3rd level, between-subject analysis described under "We are now ready to setup the third level, between-subject, analysis."
> Not surprising, the FSL group has provided a discussion and example of how to implement this 2nd level fixed effects analysis. I'll follow this:
>
> http://www.fmrib.ox.ac.uk/fslcourse/lectures/practicals/feat2/index.htm
> see:
> Group difference with multiple sessions for each subject
>
>
>
>> Ah, yes! This is good. I do want to use the information provided by the varcopes.
>>
>> Would I feed the resulting cope1 volumes into randomise? Or would I use the zstat1 volumes as input to randomise?
>>
>> This gets back to my original question about using beta weights (cope1) as a measure of functional connectivity (no noise weighting factor) vs. using z-scores (zstat1) as a measure (includes noise weighting; is a transformed expression of correlation).
>>
>> Thanks,
>> * ba
>>
>>
>>
>>> In actual fact, the best thing to do would be to perform a fixed
>>> effects 2nd level analysis on each subject's connectivity data to
>>> calculate the difference between sessions for each subject. The cope
>>> images from such an analysis would represent the change in
>>> connectivity from session 1 to session 2, and could then be entered
>>> into a randomise group analysis. This would have the advantage of
>>> using the varcopes from the 1st level connectivity analysis.
>>>
>>> -Tom
|