Determinant output obtained from deformation field not as expected


#1

Hello,

I have asked the ANTs team regarding this matter but they suggested I ask the SCT team instead. I previously registered a 14 slice mFFE scan to the template using SCT and am currently calculating the determinant of the outputted deformation field from sct_register_to_template. I calculated the Jacobian by only considering the coordinates reflective of the templated image as the remaining values in the deformation field are very large and negatively impacted determinant calculation. Nevertheless, the calculated Jacobian files still have several planar, light striations. The respective deformation field does not possess such characteristics. I have attached a sample of the Jacobian determination, deformation field, and target file here. Any help for why this is happening is appreciated.

Regards,
Ashwin


#2

Hi Ashwin,
This is likely related to the stretch (z-scaling), which aims at aligning respective vertebral levels between the source and the destination image, and which occurs as an intermediate step before doing the slice-wise registration.
Cheers,
Julien


#3

Hi Julien,

Thank for the prompt reply. I have a follow-up question. The volumes that were registered using such rotation algorithms as centermassrot and columnwise appear to have a Jacobian where cord matter is not present. I’ve uploaded the determinant, deformation field, and target image here. The deformation field is representative of the determinant. Any reasoning or help for why this is happening is much appreciated.

Regards,
Ashwin


#4

Hi,
I’m not sure what you mean by “cord matter is not present”, but the determinant you uploaded does have non-zero value within the spinal cord. Those deformations affect both the regions inside and outside the cord.
Best,
Julien