[MINC-users] minctracc on 2D volumes?

[Paul GRAVEL]

Dear all,

Thank you very much for your replies!

Mallar, Soren, and Andrew, I apologies but I will use Claude's approach, 
hope you don't mind ;-)! Claude has solved the problem (many thanks 
Claude!).

Have a good day!

Paul

On Thu, 14 Jul 2011, Andrew Janke wrote:

>> Soren said:
>>  Adding to that, you may want to try padding with two all-zero slices in
>> each z-extreme (on top of what Mallar suggested).
>> I had to do that to force and in-plane registration only.
>
>> Mallar said:
>>> It’s tricky to get minctracc to work on 2D slices in linear mode.  I would
>>> suggest using the -w_translations and -w_rotations options that you
>>> basically “turn off” the out of plane components by using an absurdly small
>>> value.
>
> I'll bow to Mallars experience with this here...  But to add a bit of
> "practicality" here's how I would approach using this advice.
>
> First, remember that minctracc expects a 3D volume, so let's give it one.
>
> Make your z slice big and thick so that rotations out of plane
> are penalised by minctracc:
> (beware! minc_modify_header modifies your file in place!)
>
>   $ minc_modify_header -dinsert zspace:step 20 xtrue_rot45deg.mnc
>
> Then add a "blank" slice either side to further penalise minctracc for
> doing out of plane rotations:
>
>   $ mincreshape -dimrange zspace=-1,3 xtrue_rot45deg.mnc expanded.mnc
>
> And not run minctracc using the arguments that Mallar suggests:
>
>   $ minctracc -lsq6 -w_translations 1 1 0.0001 \
>         -w_rotations 0.00001 0.00001 0.017453 \
>         expanded.mnc expanded2.mnc out.xfm
>
> (you will have to apply the same treatment to your target file to make
> expanded2.mnc)
>
> Let us know how it goes, this seems to be a recurring question so it'd
> be nice to put this up on wikibooks somewhere if this recipie works
> well.
>
> Thanks
>
>
> a
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>