TIFF & STL do not seem to align

[jcfr]

After loading the following data:

• WN00105/Models/mc3/mc3_dcm_cropped.tif
• AUT_mc3.stl

It looks like the STL and associated TIFF stack do not align:

Caption:

• Top we have volume rendering of
  • TIFF loaded as LPS + converted to RAS + scaled
  • TIFF loaded as LPS + scaled
• Bottom, we have rendering of
  • STL loaded as RAS
  • STL loaded as LPS

TIFF

LPS Oriented

[NicerNewerCar]

After a discussion with @amymmorton, all of the models (on a per-study basis) were adjusted with several transforms. We are proposing the inclusion in each dataset of a modelSpaceToCTSpace transform. This transformation would allow anyone to convert to/from CT space.

Since both Autoscoper and Slicer transformations deal with world space we would expect that loading the 4x4 transformations from Autoscoper into Slicer would create the same orientation of the models in question. However, as shown below, this is not the case.

Autoscoper:

Slicer:

[amymmorton]

@BardiyaAk I think I have a handle on this - but need your input - emailing- you..

[amymmorton]

Note 1)

When I examine the Transforms module, it appears that the 3D Slicer environment is very interconnected with anatomical specific directions. (LR, PA. IS).

From my exposure to data processed in autoscoper, there does not exist the same emphasis. Manipulations of data are processed in XYZ 3space coordinates. In autoscoper, specifically the XYZ origin of the 3d world and its orientation is set during calibration by the placement of the calibration cube. The cube could be placed , relative to the anatomy in a different location each patient/specimen let alone between projects/ groups.

Navigating between the anatomical and 3d space coordinates may become tricky.

Communities like ISB have tried to standardize anatomical coordinate systems for joints/ bones to specific the xyz and anatomical orientations. If the researcher has adopted a isb standard, then the association would be joint specific.

[amymmorton]

TO DO:
Documentation path forward:

Many datasets uploaded and transferred via gdrive- but issues in documentaiont have been detected: #129

Sample data included in AUtoscoper needs to be fully documented

• _ upload legacy matlab code - which was used to produce datasets from 2018.

• -readme in the google drive - links to those code functions as a documented how-it-was done.

• Full dataset upload - WN00113 - all tfm, all wrl, all dicoms/tiffs. For rad, mc3

[amymmorton]

legacy processing code

WN00113 dataset