Generalized procrustes alignment missing

[styner]

The current SlicerSALT shape analysis module does not provide an option to perform also Generalized rigid-body procrustes alignment as in the original SPHARM-PDM pipeline. Such a alignment is the norm for most shape analysis pipeline and should be added to the module. This is only necessary if there is (a) no prior registration template and (b) if so requested by the user (I think it should be "on" by default).
This would involve:

1. computing a mean shape from all the ellipse aligned data in the current output folder
2. rerunning ParatoSPHARM with (1) as the rigid registration template
3. recomputing the mean shape using the procrustes aligned data
4. rerunning ParatoSPHARM with (3) as the rigid registration template, and return to step (3) as long as the mean shape changes in a major way (e.g. a Hausdorff surface change of more than 1 voxel size)

[bpaniagua]

Hi Martin,

I remember we had this as an argument -mTemplate and we removed it... I cant recall why, do you remember?

I was thinking about enabling Procrustes as part of the surface toolbox revamp. The idea of this work is to port a bunch of the MeshMath functionality into Slicer in a way that is easy and that is format agnostic for using it both in commandline and through the GUI.

I have been working on a lab's page here.

Let me know what you think.
Bea

[styner]

Not sure why we removed the option, probably to simplify things (only one ParaToSPHARM computation). Maybe we wanted the user to compute the mean shape him/herself through such a Surface/Mesh/Model toolbox and then rerun the last step of the shape analysis themselves. It though seems a bit non-intuitive that a user would do this themselves rather than having an option directly in the analysis module.

Btw, adding the MeshMath functionality into a Slicer compatible module would make total sense, I often need to point folks to MeshMath and it would be better to have that in Slicer or SlicerSALT. Isn't there already a model/mesh toolbox that does triangle simplification etc. Maybe adding to that tool box would be good?

Martin

[bpaniagua]

Yes, if you read the labs page here you will see we plan to create small CLIs that can be called independently or through the Surface Toolbox.