[MINC-users] minctracc error after initialization fromregister-based xfm: one last question

[Audette, Michel]

Hi again, 

and sorry to be a pest. This conversation came up a little while back, on another project of mine, and is relevant to my discussion of today. It has to do with which parameters to tweak on minctracc, if one is not working with 1mm resolution MR, but something significantly more high-res. In my case 0.08mm and 0.2-0.3mm for the two volumes, with one volume only covering a small part of the second. 

Also, starting from a pretty good estimation: a rigidly registered resampled source volume, based on a transformation obtained via register. 

Is it -step value that gets tweaked, or -w_translations, for smaller volumes? 

Also, since the orientation alignment is pretty good, does this have implications for -w_rotations? 

Thanks for your consideration. 

Cheers, 

Michel


-----Original Message-----
From: minc-users-bounces at bic.mni.mcgill.ca on behalf of Andrew Janke
Sent: Fri 8/17/2007 1:03 PM
To: MINC users mailing list
Subject: Re: [MINC-users] minctracc error after initialization fromregister-based xfm
 
On 8/17/07, Audette, Michel <Michel.Audette at medizin.uni-leipzig.de> wrote:
> maudette at icaw164201:~/work/earData> minctracc -transformation HumanS16885_etzoldCT.xfm HumanS16885_Stack_nuc_xyz.mnc etzold_emma_12792115_003002_mri_xyz.mnc HumanS16885_Stack_nuc_xyz_nonlin.mnc
> WARNING: (rotmat_to_ang.c:187) step one: rz not in the range -pi/2..pi/2
> Cannot convert R to radians!  -0.717591    0.045836   -0.694954
>    0.408952   -0.779968   -0.473716
>   -0.563756   -0.624138    0.540954
> Error in minctracc in file minctracc.c, line 536
> Could not initialize transformation parameters

Hi Michel,

What does:

   xfm2param HumanS16885_etzoldCT.xfm

return?  I suspect a rotation larger than pi/2.

There is no way around this in minctracc given how it optimises
rotations (Euler Angles). But there is a way!

First resample your source file using the xfm in question:

   mincresample ....  -transformation HumanS16885_etzoldCT.xfm \
      HumanS16885_Stack_nuc_xyz.mnc grimpflegroob.mnc

Then do your nonlinear part of the registration starting with an identity xfm

   minctracc -identity HumanS16885_Stack_nuc_xyz.mnc \
      grimpflegroob.mnc HumanS16885_Stack_nuc_xyz_nonlin.mnc \
      nonlinear.xfm

Once you then have your xfm, concatenate the two and resample. Of
course things are not best they can be given that you are working from
resampled data, but at least you only resample once for your final
image.


a
_______________________________________________
MINC-users at bic.mni.mcgill.ca
http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users