Bernhard -
> We are analysing a multi-subject epoch related fMRI study comparing two
> different groups (12 subjects each). For each subject there are two runs
> (120 scans, TR 3 seconds, whole brain). Group comparison should be done
> using a second level analysis.
> When looking through the archives I was wondering wich way is the best
> and most valid for first level single subject analysis creating the
> contrast images for the second level (concerning temporal filtering and
> global normalisation).
>
> 1) Should Global normalisation (scaling) applied on the first level.
> (There were also comments that this can negatively affect the data).
If you want to do global normalisation, it should be done
at the first level. A more difficult question to answer is whether
one should ever globally normalise fMRI data - the problem arising
when the estimate of global is correlated with your effects of
interest (eg as can occur if you effects of interest cause large +
extensive changes that contaminate the global estimate) - see the
SPM archives for more discussion.
> 2) Should a low pass filter be applied (in case, which one is
> appropriate hrf/Gaussian)
If you are only ever looking at second-level analyses,
there is no need to lowpass filter - such filtering is only
used to correct for autocorrelation in first-level models -
an issue not relevant to second-level models. Indeed,
lowpass filtering typically reduces the efficiency with which
parameters are estimated (you are essentially losing high
frequency information; see Friston et al, 2000, "To Smooth
or Not to Smooth"), so you should not lowpass filter.
Rik
--
---------------------------8-{)}-------------------------
DR R HENSON
Wellcome Department of Cognitive Neurology
& Institute of Cognitive Neuroscience
University College London
17 Queen Square
London, WC1N 3AR
England
EMAIL: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~rhenson
TEL1 +44 (0)20 7679 1131
TEL2 +44 (0)20 7833 7472
MOB +44 (0)794 1377 345
FAX +44 (0)20 7813 1420
---------------------------------------------------------
--
|