I used SlicerSalt software to carry out a shape analysis but whichever database I used and for different substructures such as caudate, hippocampus and amygdala the pvalue was always non-significant. In particular, I performed the following steps:
- shapeAnalysisModule from which I have as input the mask of interest of the first subject. Then I set the coef of the first subject just obtained as a template and obtained the mesh of all the subjects.
- Procrustes Registration between all the elligned.vtk meshes taking the first one as reference and scaling.
- Population Analysis between the two sets of subjects obtaining the average surfaces
- Covariate Significance testing by putting different covariates such as age, years of illness.
I did not carry out the Correspondence Improvment as I could not find a way to obtain landmarks automatically for so many subjects. Is there something wrong with the procedure?
I would also have liked to perform shapeAnalysisMANCOVA with these models but as it is an old tool it does not read the new vtk. Is there a way to convert them to the old format? otherwise how can I see the pRAW?
thank you very much,