The reason for the problem is that the T1 was moved by the coregistration
step. The original transform would warp the T1 from its original space, to
the MNI space. However, after coregistering the T1 (source) to the EPI
(reference/target), it had been moved, so the spatial normalisation
parameters were no longer appropriate.
The geneal recommendations are:
1) Coregister all your images before doing spatial noemalisation.
2) Coregister the an EPI image (selecting all the other EPI images as "other")
to the T1, before writing out the spatially normalised images.
Best regards,
-John
> Thanks for your response. I now have a question about applying the
> normalization parameters from the normalized T1 (normalized with a
> lesion mask) to the EPI time series data.
>
> The T1 normalized with a very good outcome. I then went back and
> coregistered (mutual information coregistration) the unnormalized T1 to
> the mean realigned EPI and resliced the entire time-series.
>
> The T1 and EPI are now in qualitatively anyway, near perfect register.
> However, when I apply the normalization parameters to the EPI
> time-series data and use Check Reg to inspect the spatially normalized
> T1, a normalized EPI scan and the T1 template, the EPI scan its way out
> of register with the others. It seems its origin is lower by several
> slices.
>
> Can you identify any misteps on my part? Should I be using the lesion
> mask with the EPI time series data as well?
|