Inverse B0 for DKI processing

Hi
I would like to know if, in processing DKI data from neonatal spinal cord, inverse b0 image is useful and has to be used into SCT pipeline. Up-to-now I used this script (kurtosis_def.sh (3.2 KB) ) and didn’t use it , but maybe it can help processing (see figure for inverse b0 example).


best,
Rosella Trò

Hi Rosella,

by “inverse b0 image”, did you mean: b=0 image acquired with reversed phase encoding direction?

If that’s what you meant, then the usefulness/relevance of using this image depends on:

  • how much distortion there is in the EPI;
  • are the inverse/forward b=0 images acquired back-to-back, with minimum motion between them (otherwise the correction will introduce spurious deformation, instead of correcting them!)
  • is there sufficient SNR (for robust correction)
  • depending on the algorithm used for correcting the deformation, do both inverse/forward images contain the same information? e.g., if a bright structure (e.g. fat) only appears in one of the image, the distortion correction algorithm might create spurious deformation

this list is not exhaustive, but shows that there are multiple things to consider. Looking at this problem requires time and expertise. If you don’t feel comfortable making these kind of decisions and don’t have access to experts in your collaboration circle, i would recommend to skip that step as to avoid introducing spurious deformations.

yes I exactely mean a b=0 image acquired with reverse phase encoding. thank you for your suggestion. In case I decided to use this additional image, the proper SCT command is

sct_dmri_concat_b0_and_dwi
right?
thanks
Rosella

In case I decided to use this additional image, the proper SCT command is sct_dmri_concat_b0_and_dwi
right?

no, you should absolutely not merge in time those two very different datasets, as your DKI metrics will suffer from very large mis-alignment.

as i said above, unless you know exactly what you are doing with these inverse b=0 images, i suggest you do not use them

1 Like