Hi Guys,

I recently ran Shape Analysis Module (Slicer 4.3) on my amygdala dataset with 100 subjects (extracted from Freesurfer ASEG.MGZ file) everything ran good and I was able to do the quality checking on the ShapePopulationViewer as you guys recommended. As Next step, I want to try out the statistical analysis on my MacOS 10.10.4.

I understood from the forum's past discussions that running the MANCOVA is not possible on MacOS and By converting VTK to meta format I can run the StatNonParamTestPDM (downloaded for Yosemite) without covariates. I tried running the StatNonParamTestPDM and it was showing me errors about my META files are not readable and when I tried opening them in ITK-SNAP it was showing file not recognizable. I am not sure if the files created by the VTK2meta function is correct. I am attaching one of the meta files created and I will really appreciate if you guys can provide me some input about the quality and validity of my meta files.

Thanks,
Vinod

Hi Vinod,



You will not be able to load your *meta surface meshes in ITK-snap since they are surfaces and no longer image volumes (data that ITKsnap reads).
Meta is the Spatial Object format that was supported in old versions of ITK and that is no longer supported in ITK, since all surface functions are being implemented in VTK. We used to provide a mesh reader with SPHARM that could read that meta format meta, KWMeshVisu (distributed as part of the v.1.10.10 version, for linux only) but that is also no longer supported.

I looked at your file using our own compiled version of KWMeshVisu and everything seems fine.

Your problem with StatNonParamTestPDM has to have another source and not the data. Could you please tell us the commandline you use to run StatNonParamTestPDM and the exact error message?



Thank you!

Beatriz

Hi Vinod

One more thing, the amygdala is a structure that's fairly blobby/ellipsoidal and SPHARM usually has problems with such structures (in general structures that possess any rotational symmetry). Can you display your structures with the phi-coloring in ShapePopulationViewer (I think the procrustes aligned vtk surfaces have both phi and theta coloring attributed to the meshes) and check the correspondence? Maybe make a screenshot and post it here.

Btw, for that purpose simply load the output csv in the folder OutputGroupFile into ShapePopulationViewer and select the column with the procrustes aligned surfaces

Martin

Dear Martin and Beatriz,

Thanks for your insightful input and comments, I really appreciate it. I found problem with few of my VTK files (able to catch them in ShapePopulationViewer) and they were not converted into meta correctly and once I removed those subjects from analysis file, everything was fine. 

I was able to run the following command
"StatNonParamTestPDM NewAmygShapeDiff.txt -surfList -numPerms 20000 -signLevel 0.05 -signSteps 1000"
and got the output. But I am not sure if the output is correct. I am attaching the zipped outputs and I will appreciate if you can provide me some input on the validity of the output I got.   

attached are following files:

NewAmygShapeDiff.txt
NewAmygShapeDiff.txt_avgPval.txt
NewAmygShapeDiff.txt_corrPval.txt
NewAmygShapeDiff.txt_FDRPval.txt
NewAmygShapeDiff.txt_MeanDiff.txt
NewAmygShapeDiff.txt_meanDiffVec.txt
NewAmygShapeDiff.txt_rawPval.txt
NewAmygShapeDiff.txt_realStat.txt
NewAmygShapeDiff.txt_Tstat.txt

Thanks,
Vinod

Hi Vinod,



Great that you were able to generate results.
I think you would benefit now from the suggestions Martin is making to another Forum user, in order to be able to visualize her results.
Follow the whole conversation and feel free to comment here: http://www.nitrc.org/forum/forum.php?offset=25&max_rows=25&style=ultimate&forum_id=1281&thread_id=5724


Thanks,
Beatriz