tra seem to be wrt dicom origin rather than 'autoscoper' origin

BrownBiomechanics / SlicerAutoscoperM

This 3D Slicer extension enables users to perform image registration.

https://autoscoperm.slicer.org

MIT License

0 stars 3 forks source link

tra seem to be wrt dicom origin rather than 'autoscoper' origin #102

Open amymmorton opened 1 month ago

amymmorton commented 1 month ago

When generating VRG from a static volume or from a seqence, the Origin2DICOM transform represents relationship between dicom space and aut space.

Dicom space	Origin2DICOM used to express volume in AUT space

:warning: AUT space (center of volume and VRGs) is the space from which .tra transforms should be expressed. Taking .tra : the positions are not in appropriate anatomical location. See below.

AUT space (center of volume and VRGs) is the space from which .tra transforms should be expressed. Taking .tra : the positions are not in appropriate anatomical location	@NicerNewerCar pointed- they appear translationaly to be in DICOM basis:

amymmorton commented 3 weeks ago

Attempting to branch at commit 9950058, mar 28th just before commit 24c43b6

amymmorton commented 3 weeks ago

unfortunately in that bracnh build the origin2dicom tfm was not being generated for either a static 3d ct or the seq volume

jcfr commented 2 weeks ago

Based on https://github.com/BrownBiomechanics/SlicerAutoscoperM/issues/102#issuecomment-2165950975 from @amymmorton , regression possibly introduced through pull request:

https://github.com/BrownBiomechanics/SlicerAutoscoperM/pull/68

originally intended to address:

https://github.com/BrownBiomechanics/SlicerAutoscoperM/issues/56

jcfr commented 2 weeks ago

@amymmorton To allow us to more easily reproduce the problem, would it be possible to provide the steps & input files[^1] allowing to reproduce the problem in SlicerAutoscoperM ?

[^1]: For example, you could provide a link referencing the dataset to download

amymmorton commented 2 weeks ago

Originally shared in email from @NicerNewerCar

Autoscoper Slicer Scene

amymmorton commented 2 weeks ago

The data originally pprovided 4/29: BN00107

amymmorton commented 2 weeks ago

Something odd about Origin2DICOM

when applied to the 'target' pose CT dicoms- the origin should be geometric center in the volume- but it's not:

yellow is neutral - bone color rendering is flexion with Origin2DICOM applied

NicerNewerCar commented 2 weeks ago

Something odd about Origin2DICOM

when applied to the 'target' pose CT dicoms- the origin should be geometric center in the volume- but it's not:

This is happening because the Origin2Dicom Transform is created from the 1: neutral volume. The 4: flexion volume has a smaller Z dimension and because of that ends up slightly off center from the origin in the z direction.

It Looks like the Origin2Dicom is not getting applied correctly or is getting removed before/during the VRG generation.

Slicer No O2D tfm	Slicer O2D	Autoscoper

Working on a fix for this now