help > error with SPHARM-PDM output?
Showing 1-5 of 5 posts
Results per page:
Jun 5, 2016  03:06 AM | Yann Chye
error with SPHARM-PDM output?

I'm having some problems running a couple of my images on SPHARM's shape analysis module. Specifically, some of the output files look weird, and the image can't be visualised on shapepopulationviewer.

For example,
  • the values for the procalign.vtk file are extremely large (i.e. 1.5e+24, -5.3e+28)
  • There was no .._pp_surf.vtk and .._pp_para.vtk file output
I'm not sure what's going on and am at my wits end. Any help would be much appreciated, please!!
Jun 13, 2016  05:06 PM | Martin Styner
RE: error with SPHARM-PDM output?
Hi Yann

Likely the surfaces were not simply connected and thus the spherical mapping went off.

You mentioned there is no pp_surf.vtk and no pp_para.vtk surfaces, really? I am surprised that you got any procalign surfaces, as you are missing the starting point (the surf and para vtk's). Without those surfaces, you would not get anything. Did you try loading the pp-segmentations into a viewer such as 3D Slicer or itksnap? How do they look?

Jun 17, 2016  03:06 AM | Yann Chye
RE: error with SPHARM-PDM output?
Hi Martin,

yes there was no individual pp_surf.vtk and pp_para.vtk surfaces, although there is a pp_surf_para.vtk surface. When I tried to view them (pp_surfSPHARM) in 3d slicer, the viewing window looks funny (images covers the entire viewing window like a giant curtain), despite the fact that the original gipl volume looks normal.

Jun 17, 2016  01:06 PM | Martin Styner
RE: error with SPHARM-PDM output?
those surfaces are nonsense, so do not read too much into them, as the original surfaces failed. What is the error output? My guess is that the surfaces must not have spherical topology (i.e. no hole or handles)?

Jun 17, 2016  01:06 PM | Beatriz Paniagua
RE: error with SPHARM-PDM output?

just adding a bit of information. Martin is correct saying that this is a problem in topology, but it can be caused not only because of holes or handles but also by "almost holes" or c-shaped topologies that can't be easily resolved by the method. I hope that helps,