Hi Livio,
On Thu, 3 Mar 2005, Livio Finos wrote:
> hi all!
> I'm performing a Hierarchical model analysis (at least I'm tring). I'm
> using a summary statistics approach since the design of experiment is
> equal for every subject (Am I right?)
That's right - use first-level analysis in FEAT for each session analysis,
and then carry out a second-level (cross-subject) analysis with a
"higher-level" analysis also in FEAT.
For more information see:
the FEAT manual http://www.fmrib.ox.ac.uk/fsl/feat5/
the FSL lectures http://www.fmrib.ox.ac.uk/fslcourse/
and the FEAT FAQ http://www.fmrib.ox.ac.uk/fslfaq/#title_feat
> the experiment consider 10 subjects. Each subject-session have 5 scans,
> 3secs/scans (15 secs alltogether) under rest and 5 (15 secs) under
> stimulus. 80 scans (i.e. 240 secs) alltogether.
> If I well understud the summary statistics approach, I just need to
> perform 10 separated analyses, one for each subjects (t 2 independent
> samples, 80 scans/observations), hence a second level 1 sample t test for
> 10 subjects/observations. is it right?
yes that's right
> I set the first level analysis in this way:
> data: 10 analyses, total volumes=80, TR=3 high pass filter=45 (shold be
> 15*1.5=22.5?).
45 is fine - 30 would be ok too.
> Pre-stats: slice timing correction: none, motion correction: none, BET
> brain extraction: ON, spatial smoothing FWHM =0 (my images are already
> smoothed by SPM software), temporal filtering: Highpass=ON, lowpass=OFF,
You _should_ do motion correction and it would be better NOT to pre-do the
smoothing before this. It would be best to take the raw data, turn on
motion correction and do spatial smoothing in FEAT.
> Melodic ICA=OFF.GLM:
> Evs: Numb of original EVs=1, skip=0, off=15, on=15, phase=15. I also ask
> for a GAMMA convolution: phase=0, stddev=6, mean lag =3 (as default).Contrast & F-test: setup contrast & F-test for REAL EVs, contrasts=1,
> F-tests=1, C1: 1, -1, F1=ON.
> Am I doing OK for the first level analysis?
This seems ok - check that the model looks right according to your design.
> second step should be 1 sample t test. is it right?
Yes - the easiest setup at higher-level - see
http://www.fmrib.ox.ac.uk/fsl/feat5/detail.html#SingleGroupAverage
> I also suppose the whole analysis could be (better) done in an automatical
> way in FSL. could someone address me? some tutorial for this design of
> experiment or something like that?
Sure - this is pretty automated - in the sense that you setup these
settings in FEAT and then it does everything for you.
Good luck - cheers, Steve.
--
Stephen M. Smith DPhil
Associate Director, FMRIB and Analysis Research Coordinator
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
|