Hi,
It's not very clear what model you're using - I guess that at second
level you subtracted one session from the other for each subject
separately? And then at 3rd level just took the group mean across
subjects?
Wrt your histogram-based comparisons - I'm afraid I really can't tell
what you were doing here. Were you feeding in _difference_ images? I
would not compare histograms from separate runs, but would create a
difference image and look at the histogram on that.
Cheers.
On 12 Sep 2007, at 21:13, Vishwadeep Ahluwalia wrote:
> Hi,
> my dataset contains 15 subjects, each with 2 sessions and with 2
> runs in
> each session. My aim is to find if the paradigm i used gives
> reproducible
> results or not(i.e 2nd session is the replica of the 1st session).
> so i ran
> 1st level GLM analysis for each run. at the 2nd level i did a Fixed
> effects
> analysis combining the 2 runs for each session and each subject(i.e
> i have
> 21 session1 cope images and 21 session2 cope images). At the 3rd
> level, i
> did a paired comparison between the 2 sessions using FLAME-1. i did
> not get
> any significant differences. then i found the group mean of each
> session
> using flame and i ran featquery with an ROI and found that the 90th
> percentile COPE and VARCOPE values of the two sessions were quite
> different.
> 90-percentile Session1 Session2
> COPE 38 25
> VARCOPE 19 20
> Even though the variances are similar the COPE is changing
> significantly.
> therefore should'nt i get significant differences between sessions
> at the
> 3rd level paired comparison analysis? any suggestions as to why
> this might
> not be the case. also, would this 3 level analysis be an
> appropriate method
> of assessing the reproducibility of activation in a particular ROI?
> Thanks!
> -vish
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|