I've seen NaNs even when using the template bounding box, which is more
than enough coverage for my region of interest.
What worked for me recently was feeding SPM Minc files with origin
specified in the minc start variable and making sure input coordinates
were relative to that center. See this post:
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind04&L=spm&P=R374133&I=-1
Making sure SPM knew the volume center going into spatial normalization
seemed to be the key.
Donna
On 01/20/2005 06:35 AM, John Ashburner wrote:
>>1. For point normalization, I first get the inverse deformation followed
>>by the code at Gem 5, Normalizing [x y z] at "John's SPM2 Gems".
>>However, the corresponding position on template of lots of points away
>>from the volume center, including landmark as LPA, Nasion, and RPA, cannot
>>be found by this method. The inverse deformation seem not having all the
>>mapping points. I'd like to use these landmark to verify that the
>>deformation is correct. How to get these point?
>>
>>
>
>This procedure is the one that involves writing out a deformation field and
>inverting it. In order to avoid too many unknown mappings (NaNs) in the
>inverse deformation field, the forward deformation field needs to have good
>coverage. Try increasing the bounding box when this is created. This may
>involve changing the spatial normalisation defaults.
>
>
>
>>2. I would also like to use deformation to get corresponding position on
>>individual's MRI from template. Can I use the deformation, y_xxxx.img, in
>>the very similar way as Gem 5, Normalizing [x y z] at "John's SPM2 Gems".
>>If it is implausible, how to do it?
>>
>>
>
>The same procedure.
>
>c = [x y z]'; % co-ordinates in mm (as shown by Display)
>P = spm_get(1,'y*.img','Select forward warp');
>P = [repmat(P,3,1) [',1';',2';',3']];
>V = spm_vol(P);
>
>vx = inv(V(1).mat)*[c ; 1]; % The voxel in the deformation to sample
>
>w_coord = [...
> spm_sample_vol(V(1),vx(1),vx(2),vx(3),1)
> spm_sample_vol(V(2),vx(1),vx(2),vx(3),1)
> spm_sample_vol(V(3),vx(1),vx(2),vx(3),1)]
>
>
>
>>3. Are both "deformation, y_xxxx.img" and "inverse deformation,
>>iy_xxxx.img" already containing the information of affine matrix?
>>Therefore I can apply them without additionally adding affine matrix. Is
>>that correct?
>>
>>
>
>The affine transform as well as the nonlinear deformations are included in
>these deformation fields.
>
>
>
>>4. At last, I'd like to learn more about the deformation field in SPM,
>>where can I get the manual or any reference recommended for me to read?
>>
>>
>
>There's no manual I'm afraid. The best source of information (in the form of
>examples) is currently the archives of the Mailing list. Try a search of
>http://www.jiscmail.ac.uk/cgi-bin/webadmin?S1=spm for "Deformations toolbox"
>
>Best regards,
>-John
>
>
>
|