Dear colleagues,
I am writing to both SPM and FLS mailing list. Sorry for those who have
received this message twice.
I have some questions about registration of volumes in fmri. As I understand
both programs are doing this registration appliying 6- parametre
transformations, that is, considering the brain as a rigid solid. However,
taking in account the ratio TR/delay the most probable is that movements are
produced during volume acquisition instead between two acquisitions. In this
case, different slices should be not parallel so the whole brain is deformed
in a way that it can't be recovered by any affine transformation (neither a
12-parameter one). I agree that if pacient's movements are very slow this
deformation may be irrelevant but short movements are usual (as it can be
seen in registration graphs).
My questions are: Should registration be done slice by slice in orther get
more accuracy? In that case, is there any tool to reslice a volume from non
parallel slices? And finally, is there any way to measure if brain
deformation produced by pacient's movement during acquisition is relevant or
not (quality assurance of data)?
Thanks for your attention
Yours,
Carles Falcon
Hospital Clinic Barcelona
University of Barcelona (Spain)
[log in to unmask]
|