Hi -
This is quite a common thing after using Brain extraction.
The reason is that brain extraction includes the CSF on the outside
of the brain in the T2 (where it is bright), but not in the T1 (where
it is dark).
Depending on the contrast in your images, you may well find that this
brain-extracted boundary provides the information that drives the
registration.
It is well-worth checking T1-to-T2 registrations for this.
I agree that inverting the T2 image intensities is a good solution,
but you may want to do it before brain extraction.
T
On 4 Dec 2006, at 09:04, Tuomas Neuvonen wrote:
> Dear FSL experts,
>
> I did a registration of a structural image (t1_brain) and bedpost
> data (nodif_brain) using the registration step in FDT (cost functions:
> mutual information and correlation ratio; DOF: 12).
>
> When looking carefully at the registration results in fslview I found
> that gyrii in T1 image (t1_brain) are aligned with sulci in the T2
> image
> (nodif_brain) for both cost functions. I obtained the best
> alignment of
> sulci only after inverting the T2 image intensities (making the
> contrasts somewhat
> similar to a T1). Any comments on using this kind of approach?
>
> thanks,
>
> Tuomas
|