Bounding box correction

Support
, ,
1

Hi, I would like to ask you regarding the size of the transformation result.

The brief summary: I use the Elastix modul for non-linear normalization to get the transformation matrix (on a mean volume), which is then applied to a series of individual runs (having their position similar to the transformed volume) in the Transforms modul with the above created bspline matrix.

However, the end result dimensions are sometimes different (e.g. the bounding box sizes is altered, once it is [102 100 34], the next time is [100 101 32]). For the subsequent MATLAB analysis the software requires similar sizes. This is the error message which is currently presented in SPM:

  • "** The images do not all have the same dimensions. **
  • The function assumes that a voxel in one image corresponds with the same voxel in another. This is not a safe assumption if the image dimensions differ. Please ensure that you have processed all the image data in the same way (eg. check spatial normalisation bounding-boxes, voxel-sizes etc)."

Is there a way how I can ensure that the outcome of a transformation always have the same dimensions? (either with the Elastix, or with another module which adds a certain number of blank (black) voxels to the volumes’ margin to create the same-dimensioned bounding-box?)

Thank your for your suggestions!

2

I was thinking about another method: by cropping all the volumes with the same ROI box.
In this way the bounding box would be identical throughout the image series.

One question regarding to that: is there a way to apply the cropping to multiple volumes at the same time?