Thanks John,
I normally coregister it first (T1-FA) and then use a non linear
normalization ( FA of the subject - FA template (in house built)) and then
rewrite all the required images using the transformation matrix generated
from the normalization. This seems to pull back the genu of T1 image a
little bit to the top. Not a 100% match but just looking for another way
with your help.
This procedure doesnt have any drawbacks I believe , isn't it !
Thanks
Regards
Tilak
On Tue, Feb 24, 2009 at 1:31 PM, John Ashburner <[log in to unmask]>wrote:
> If there is a lot of distortion in the phase-encode direction of your
> images,
> then this could be a bit problematic. These distortions are caused by
> changes in the magnetic field that arise because of different magnetic
> susceptibilities between tissue and air. This subject seems to have quite
> large frontal sinuses, so there may be a bit more spatial distortion than
> usual.
>
> Coregistration in SPM only uses a rigid-body transformation, and does not
> attempt to correct for these spatial distortions. This means that it is
> not
> always possible to achieve good alignment between distorted EPI data, and
> relatively undistorted T1w.
>
> If you collect field maps when you scan, then it may be possible to correct
> the artifact. If not, then it's a bit tricky as there's nothing yet
> included
> in SPM that could model this distortion. Others have published papers on
> correcting it using nonlinear image registration, but I'm not aware of any
> freely available software that would do the job.
>
> Best regards,
> -John
>
> On Tuesday 24 February 2009 16:49, ramtilak gattu wrote:
> > Thanks a lot,
> >
> > that clearly works . I just segmented the T1 image and used the WM as
> the
> > target image .
> >
> > But I have a small minor problem, if you look at the attached image where
> > the cross hairs are pinpointed on the sagittal view.
> >
> > The top end of the genu of the corpus collosum on the FA image is matched
> > to the center of the genu of the corpus collosum on T1 image. it
> literally
> > looks like the genu in FA is pulled down. Is there a way to correct it (
> > for a better view look at the axial slice).
> >
> > Thanks again for the valuable suggestion.
> >
> > Regards
> > Tilak
> >
> > On Tue, Feb 24, 2009 at 5:56 AM, John Ashburner
> <[log in to unmask]>wrote:
> > > Maybe try coregistering with a skull-stripped version of the T1w. You
> > > can do
> > > this by segmenting, and masking the T1w (or a bias corrected version)
> > > with the sum of the GM, WM and CSF. The masking can be done using
> > > ImCalc.
> > >
> > > Best regards,
> > > -John
> > >
> > > On Tuesday 24 February 2009 02:48, ramtilak gattu wrote:
> > > > Hello SPM users,
> > > >
> > > > I apologize if it is a re-posting of my message
> > > >
> > > > I am trying to co-register Fractional Anisotropy (DTI_FA -
> > > > resolution-1.8*1.8*4-33 slices) of a subject to his T1 image (
> > > > resolution-0.98*0.98*3-60 slices) using spm 8b. I even tried spm5.
> Both
> > >
> > > the
> > >
> > > > images are in Analyze format
> > > >
> > > > I played with different objective functions but of no use.
> > > >
> > > > Please see the attached image before registration (top row) and after
> > > > registration (bottom row).
> > > >
> > > > Could anyone guide me whats going wrong with coregistration .
> > > >
> > > >
> > > >
> > > > Thanks for the help in advance
> > > >
> > > > Regards
> > > > Tilak
>
>
--
Ramtilak Gattu
21530 Green Hill Rd
#127
Farmington Hills
Michigan-48335
ph: 313-887-1766 (home)
313-820-8862 (cell)
313-993-8896 (MRRF-office)
313-745-2487 (MS-office)
