[Insight-users] deformation field interpretation, motion extraction

[Luis Ibanez]

Hi Christian,

   The units of deformation fields generated by ITK defomable
   registration algorithms are the same of the image spacing.

   For example if you are using CT images with spacing in
   millimeters, then the units of the deformation field will
   be in millimeters.

   If you are using microscopy images with spacing in microns,
   then the units of the deformation field will be in microns.


BTW: You may find interesting to use the 4D Slicer application
from the following paper:

     http://www.midasjournal.org/browse/publication/617
     "VV: a viewer for the evaluation of 4D image registration"



    Regards,


       Luis


---------------
c23562 wrote:
> Dear All, 
> 
> I’m working on a project dealing with motion extraction from 4D- CT data of
> human heart. 
> After registration of two consecutive volumes I try to get the motion of
> interesting points from deformation field. Therefore I fetch the deformation
> vector out of the first field on the interesting position. With the result i
> make a trilinear interpolation with the neigbouring deformation vectors of
> the next field. And so on. The motion vector I write to a new field. But at
> the end the optical flow doesn’t look like a possible motion of my
> seedpoint.  What is going wrong? 
> May I take the deformation vector [e.g. 0.3432 0.9387 0.6202] and interpret
> it as a motion in x,y,z-direction? What is the right unit, pixel or
> millimeters? At this time I assume that the unit is pixel and my seedpoint
> moved 0.3432 to the next x-index, 0.9387 to the next y-index and 0.6202 to
> the next 0.6202 z-index. Adding the the start position to the received
> displacement I get the new position for the next deformation field pointing
> to a non-grid position. Therefore I use trilinear interpolation between
> displacement vectors of the neighbouring grid-notes to get the next motion
> vector. 
> Is it right to use the deformation field in this way or is there any
> conversion of deformation field entries needed (negation, normalizing,
> scaling, …) ? 
> (For registration and creating the deformation field I used
> DeformableRegistration5, DeformableRegistration15 and
> DeformableRegistration16 .)
> 
> Best regards, 
> Christian
>