Am Montag, den 10.11.2008, 11:17 +0100 schrieb Mathieu d'Acremont:
> Dear Volkmar,
>
> I removed the first and last 4 volumes of each session and made all the
> analysis as if it was a single session. It's somehow a simplification,
> but it allows me to compute T-test on each of my parametric modulator
> and to avoid F-test.
If you can model a t test for one session, it should be straightforward
to create an t contrast that averages over all your sessions.
> I also entered the realignment parameters so i think it somehow models
> differences between sessions (movements were more pronounced between
> sessions).
No, each session may have a different mean signal intensity, and the
assumptions made for temporal autocorrelation and high-pass-filtering
are violated if you concatenate the sessions.
> Thanks for you help.
>
> Volkmar Glauche wrote:
> > Dear Mathieu,
> >
> > a very brief answer: You should remove the first few scans from each
> > session. Also, you should treat the sessions separately in realignment,
> > slice timing (if applicable) and statistics. If you would treat your
> > data as coming from one single session in the model, SPM would not be
> > able to model e.g. different mean intensities and you would introduce
> > errors in estimation of temporal autocorrelation and high-pass
> > filtering.
> >
> > Hope this helps,
> >
> > Volkmar
> >
> > Am Mittwoch, den 05.11.2008, 19:39 +0100 schrieb Mathieu d'Acremont:
> >> Dear SPM users,
> >>
> >> I'm analyzing fMRI data with SPM8b. Data were recorded during 4
> >> functional sessions and 1 structural session:
> >>
> >> - session 1, functional, 408 vol.
> >> - session 2, functional, 408 vol.
> >> - session 3, structural, 192 vol.
> >> - session 4, functional, 408 vol.
> >> - session 5, functional, 408 vol.
> >>
> >> In each functional session, 4 volumes were recorded before the beginning
> >> of the task (vol 1-4) and 4 after the end of the task (vol 405-408). So
> >> only 400 volumes in each functional session are interesting for model
> >> estimation (vol 5-404).
> >>
> >> For model estimation, i want to treat all volumes as if they were coming
> >> from a unique session (because it is basically the same task in the 4
> >> functional sessions and i need to simplify the model). I will also use
> >> the realignment corrections computed during the preprocessing as
> >> regressors in the model.
> >>
> >> I have 2 questions concerning the preprocessing:
> >>
> >> - should i define 4 sessions for the realignment or make as if data were
> >> coming from a unique session?
> >>
> >> - which volumes should i delete during the preprocessing (e.g. vol 1-4
> >> of the first session or of all 4 sessions, what about vol 405-408)?
> >>
> >>
> >> Cheers,
> >>
>
>
--
Volkmar Glauche
-
Department of Neurology [log in to unmask]
Universitaetsklinikum Freiburg Phone 49(0)761-270-5331
Breisacher Str. 64 Fax 49(0)761-270-5416
79106 Freiburg http://fbi.uniklinik-freiburg.de/
|