Hi Ted,
see Feat's bubble help on, for example, highpass filtering under the data
as weel as prestats tab: the same filtering that is applied to the data is
applied to the model internally.
In my experience, it does not matter too much for motion correction
parameters. But you are certainly right: to be 100% consistent, you would
want to do it.
But you can easily do it via the Glm(_)gui - apply temporal filtering to
the design, and test different cutoffs. You can right away tell the
difference.
Hth,
Andreas
PS: You would then extract the filtered EV from the *.mat file.
Am 03.11.11 19:49 schrieb "Ted Satterthwaite" unter <[log in to unmask]>:
>Dear FMRIB,
>
>In working with resting-state data, as is common I have been bandpass
>filtering the timeseries data using fslmaths -bptf outside the prestats
>GUI in FEAT. When adding motion parameters to the model, it occurred to
>me that this data has not been filtered in the same way. This prompted
>two questions:
>
>1) Normally, in the FEAT GUI, there is a button to "apply temporal
>filtering" to a given EV. Is temporal filtering also normally applied to
>motion parameters included in the FEAT model?
>
>2) If one is doing temporal filtering outside of the GUI, can fslmaths
>-bptf be used on text files such as the mc/.par file? (I have not had
>luck with this). If fslmaths will not work, is there a tool available
>within FSL to do so? I imagine there must be given the temporal filtering
>of other EVs in the FEAT GUI. Of course we could do this in matlab, but
>I know that -bptf in fslmaths is not the standard butterworth filter,
>and I would like to use the same parameters as used on the timeseries
>data.
>
>Thanks for your advice.
>
>Cheers,
>Ted
>
>__________
>Ted Satterthwaite, M.D., M.A.
>University of Pennsylvania
>Department of Psychiatry
>Brain Behavior Laboratory
|