| few questions about normalisation.
| I coregistered mprage.img (as T1, object) and meanbold.img (as T2, target),
| choosing the option 'Coregister only'.
| Then, I normalised the registered bold*.img, estimating parameters from
| meanbold.img and selecting the EPI.img template.
| Finally I normalised the mprage.img, estimating parameters from the
| mprage.img and selecting the T1.img template.
|
| Is this procedure correct?
| Can I overlay the activations on the normalised mprage?
A preferable way, assuming that the coregistration step has worked well,
would be to estimate the normalization parameters once, and apply the
same parameters to both the functional and structural images. The ".mat"
files generated by the coregistration step are incorporated into the
spatial transformations from the spatial normalization, so you don't really
need to reslice any of the images before applying the spatial normalization
parameters.
The coregistration is never exact unless the fmri images have been distortion
corrected (because a rigid body registration can not match warped functional
images to an unwarped structural). At the FIL we tend to estimate our spatial
transformations from the functional images (using the EPI template), and apply
the same spatial normalization parameters to the structurals. We find that
the functional images of different subjects are slightly better registered
when we do it this way.
I hope this helps,
-John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|