FZJ-INM1-BDA / voluba

Browser based tool for interactive alignment of high-resolution volumes of interest to 3D reference brain models
https://voluba.readthedocs.io
Apache License 2.0
4 stars 0 forks source link

Resolution/Scale reset after selection of transform dialog #166

Closed JThoennissen closed 1 year ago

JThoennissen commented 1 year ago

The issue causes two problems:

  1. When selecting Hippocampus public volume it is displayed too big. By clicking on the brain icon to transform it, the volume is rescaled correctly but due to this moved outside of the reference space.
  2. When a user sets the resolution/scale, closes the transform dialog and opens it again, the resolution/scale is reset to the initial scaling that is retrieved from the precomputed info file.
JThoennissen commented 1 year ago

Find a screencast showing 1. below:

https://user-images.githubusercontent.com/32842931/231501995-f66aa0ec-f7d3-43d8-b0cc-d8c7dccae8fd.mp4

xgui3783 commented 1 year ago

merged and deployed to staging , can you verify if it's fixed?

https://voluba-staging.apps.hbp.eu/#/

JThoennissen commented 1 year ago

The second issue is fixed 👍 The first issue remains, but is not urgent. It seems the scale is initially set to 20,53 for Hippocampus volume ...

xgui3783 commented 1 year ago

Here, I am respecting the transform returned by the image service.

I am unsure why the image service returned to contain the specific transform. better asking @pacher

Why the Hippocampus public volume has a transform like that

[
  [
    20.531328320802004,
    0,
    0,
    -24353.02734375
  ],
  [
    0,
    20.531328320802004,
    0,
    -24353.02734375
  ],
  [
    0,
    0,
    16.66734486266531,
    -29998.779296875
  ],
  [
    0,
    0,
    0,
    1
  ]
]

A command line to retrieve the transform would be

curl 'https://zam10143.zam.kfa-juelich.de/chumni/list' | jq '.[1].extra.neuroglancer'
pacher commented 1 year ago

It seems that affine matrix stored in Hippocampus nifti is wrong which is not surprising given the origin of this nifti. I would suggest to close the issue and re-open if this problem is observed with any other user uploaded nifti generated by any standard software.

Upd.: I mean there is no software issue to fix, but we probably need to fix the Hippocampus nifti to improve user experience.