Hi
yes, it's possible - we do recommend, however, do carry out the
time-series GLM analysis in native space and then register the
*results* into a common space in order to avoid interactions between
the GLM estimation and interpolation. The registration into high-res
space can be carried out easily from within the FEAT GUI - if you only
want to compare results between sessions (and not across subjects) you
might want to change the registration target to the subjects high-res
image rather than the standard brain.
cheers
christian
On Thursday, Jul 31, 2003, at 18:40 Europe/London, Keith Schneider
wrote:
> I'm not going to concatenate the time series from different sessions, I
> just want the different sessions to be in register so that I can see
> whether the same small structures identified in the two separate scans
> are in exactly the same place or not. Is this possible to do?
>
> keith
>
> On Thu, 31 Jul 2003, Stephen Smith wrote:
>
>> Ah - in this case, you probably _don't_ want to do this at all then.
>> We do
>> not recommend concatenating time series from different sessions (see
>> the
>> FSL email archive for details) but recommend processing each session
>> separately and then combining across sessions and/or subjects with
>> higher-level analyses. Also, see the Feat manual for more detail:
>>
>> http://www.fmrib.ox.ac.uk/fsl/feat5
>>
>> Thanks, Steve.
|