[Insight-users] resampling problems

[Lydia Ng]

> Okee, that helps me to understand...
> So, I should use something like:=20
> resample->SetTransform(transform->GetInverse()); ?

Both the ITK registration component and the resampling
works with "fixed domain to moving domain" transform
paradigm.

So, if you are using the results directly from
the registration you shouldn't need to inverse.

Have you tried running your data using the=20
MultiResMIRegistration app? If so, are you getting the
same problem?

-Lydia

> -----Original Message-----
> From: J. Van Dalen [mailto:j.vandalen@rad.umcn.nl]
> Sent: Monday, November 18, 2002 9:33 AM
> To: Lydia Ng
> Subject: RE: [Insight-users] resampling problems
>=20
>=20
>=20
> Hi Lydia,
>=20
> > It is the numerical parameters that looks inverse to you
> > or the final resampled image?
> It's the final resampled image that looks inverse.
>=20
> > As you might already be aware, that in registration
> > we are typically trying to find the transform which maps
> > a point from the fixed image domain to the moving image
> > domain - which is the reverse of how people usually think
> > in terms of transforming images.
> Okee, that helps me to understand...
> So, I should use something like:=20
> resample->SetTransform(transform->GetInverse()); ?
>=20
> > --------------------
> > Have you tried this?
> > resample->SetInterpolator(interp.GetPointer());
> >=20
> > -----------------------
> That's the answer!=20
>=20
> Thanks,
> Jorn.
>=20
>=20
>=20
>=20
>=20
>=20