Groups tutorial - sample data / guidance?

(Craig Stark) #1

I’m looking to take FreeSurfer segmentations of the hippocampus and identify any pre/post intervention changes in shape. I’ve gone through the very nice tutorial on getting SPHARM run on all my segmentations and I’ve gotten the SPHARM Shaple Analysis Module to run through cleanly.

But, I’m having trouble moving past this for group-level analysis. I’ve found the GROUPS tutorial PDF (I should note that the main salt.slicer.com page has an incorrect link) and it’s helping, but I’m not there yet. In particular, I’ve gone through and put down several fiducials on all my hippocampi, but I’m a bit lost on where my “Common unit sphere” is supposed to have come from. I believe the “Input Models” is the “Step3_ParaToSPHARMMesh” folder that’s created and my “Input Fiducial Files” is, of course, where I have those .fscv files I made with the Markups tool. But, I’ve clearly missed a step or more between the Shape Analysis Module and running a group analysis as even that first step of rigid alignment is hanging me up.

This might all be a good bit easier to work through if I could find the tutorial data that the PDF refers to (Summer2018/GroupsTutorial). Is that available somewhere? That and/or any other guidance on moving forward would be greatly appreciated.

Thanks,

Craig

0 Likes

(Beatriz Paniagua) #2

Hi Craig,

I have corrected the tutorial links in the salt.slicer.org page. Thank you for pointing that out.

GROUPS right now is just a way to adjust and optimize the SPHARM-PDM parameterizations to account for population information. The regular SPHARM-PDM computation only takes into account the geometry of each single case and that at times is insufficient.

If you refer to the option to do group statistical comparison and hypothesis testing based on shape, that module will be integrated into SALT as soon as Slicer finalizes the transition between Python 2 and Python 3.

We will make an announcement when it is ready, hopefully very soon.
Thank you!

Bea

0 Likes

(Craig Stark) #3

Thanks - I’ll be on the lookout!

Craig

0 Likes