[Insight-users] inverting deformation fields

[Matt McCormick]

Hi Anja,

Yes, sampling and the grid play an important role.

The sampling locations of the output must be chosen, and they are not
necessarily the sampling locations of the input.

This is what the ResampleImageFilter and its Transform may operator in
the "opposite" direction what is expected -- what is always computed
is the output pixel location.

HTH,
Matt

On Wed, Apr 10, 2013 at 7:17 PM, Anja Ende <anja.ende at googlemail.com> wrote:
> Oh yes, of course! That is a subtle point.
>
> May I ask a follow up question then? In inverting a deformation field then,
> how do we decide the sampling of this grid of the inverse field?
>
> Say our original deformation field has a displacement at every voxel and
> these displacement vectors can point to any real location. So, is the
> inverse deformation field going to be calculated at these real locations?
> However, we need to have a uniform grid and this is not guaranteed by this
> scheme now. Can someone explain this to me on how this is done?
>
> Sorry for asking these naive questions. I could use the ITK filters out of
> the box but I would like to know some of these basics.
>
> Thanks,
> Anja
>
>
> On 10 April 2013 20:06, Matt McCormick <matt.mccormick at kitware.com> wrote:
>>
>> Hi Anja,
>>
>> If displacement vector field gives the displacement at location x, but
>> the negative of the displacement at location x is not the inverse;
>> what is required is negative of the displacement vector at x + \delta
>> x.
>>
>> HTH,
>> Matt
>>
>> On Wed, Apr 10, 2013 at 5:47 PM, Anja Ende <anja.ende at googlemail.com>
>> wrote:
>> > Hello all,
>> >
>> > This is a very naive question and I would be very grateful if someone
>> > can
>> > clarify this doubt for me.
>> >
>> > I see that there are quite a few filters and some very sophisticated
>> > methods
>> > for inverting deformation fields.
>> >
>> > If I have a dense deformation field, why is it not enough to simply
>> > negate
>> > the displacement vector to compute the inverse? I am sure there is a
>> > simple,
>> > logical explanation for this but I am having trouble figuring it out.
>> >
>> > Thanks a lot for your help!
>> >
>> > Anja
>> >
>> > _____________________________________
>> > Powered by www.kitware.com
>> >
>> > Visit other Kitware open-source projects at
>> > http://www.kitware.com/opensource/opensource.html
>> >
>> > Kitware offers ITK Training Courses, for more information visit:
>> > http://www.kitware.com/products/protraining.php
>> >
>> > Please keep messages on-topic and check the ITK FAQ at:
>> > http://www.itk.org/Wiki/ITK_FAQ
>> >
>> > Follow this link to subscribe/unsubscribe:
>> > http://www.itk.org/mailman/listinfo/insight-users
>> >
>
>
>