Resampling glitch when the NiFTI header is oblique

Forums Forums BrainSuite Diffusion Pipeline (BDP) Resampling glitch when the NiFTI header is oblique

Viewing 3 reply threads
  • Author
    Posts
    • #1263
      robreid
      Participant

      Hi,

      I noticed a resampling glitch when the dMRI is warped to a T1 with an oblique sform:
      Resampling glitch

      The glitch appears as one or more diagonal fault planes that are more or less conspicuous depending on the local contrast – the screenshot shown is from the worst case on the inferior edge of the CC.

      The T1’s sform is
      srow_x 280 4 1.195487 -0.00832 -0.0876 -85.916519
      srow_y 296 4 0.029844 0.995965 0.197129 -165.741318
      srow_z 312 4 0.099596 -0.198567 0.992426 -124.586205

      There is no problem when the same T1 is used with sform
      srow_x 280 4 1.2 0.0 0.0 -99.663498
      srow_y 296 4 0.0 1.0156 0.0 -97.173988
      srow_z 312 4 0.0 0.0 1.0156 -128.981201

      Has anyone else seen similar problems with obliqued T1s, and is there a possibility of fixing the resampler?

      Thanks,

      Rob

    • #1264
      robreid
      Participant

      It does not look like the img button worked; try https://photos.app.goo.gl/PJyZ5xz0iQlfiGJi2 or email me.

    • #1299
      Anand Joshi
      Moderator

      Hi Rob,
      Thanks for reporting the issue. Can you please let me know which executable were you using? Was it BDP.sh?
      You can try the following workaround for now. You can simply load and save the T1 from brainsuite. It should save the header in brainsuite friendly format. Then proceed with the analysis using this saved T1.
      Anand

    • #1331
      robreid
      Participant

      Hi Anand,

      Sorry for not replying earlier. I was using BDP.sh, for scriptability, and have never actually gotten the whole brainsuite package running. If I understand correctly though, and I think I do, I could easily manipulate the T1 to have a diagonal (non-oblique) affine matrix outside brainsuite, and carry on. I do not mind resampling the T1 for this purpose since its native resolution is much better than the dMRI’s.

      Thanks!

      Rob

Viewing 3 reply threads
  • You must be logged in to reply to this topic.