Dear Dan,
The 4D intensity normalisation (grand-mean scaling) does *not* correct
for within-timeseries drift. It is the 3D version that corrects for drift (although
we do not recommend using this - temporal filtering is better). The 4D
version is just used to correct for changes in overall mean value (across
time and space) between different subjects.
I hope that helps.
All the best,
Mark
On 28 Nov 2011, at 10:48, Dan Shaw wrote:
> Thanks for your prompt response - greatly appreciated.
>
> I guess the question is how multiplying all 180 volumes comprising a time-series - the gand mean of each being affected independently by slow-frequency drifts - by the same single value can normalise the entire time-series. Surely this would simply re-scale the time-series (and artefacts)?
>
> Am I still overlooking something painfully simple?
>
> Dan.
>
|