Hi Georg

StatNonParamTest should still be available though as part of one of the older distributions. Alternatively you can compile the tool yourself and enable the compilation of StatNonParamTest.

But, neither StatNonParamTest (which does not allow you to use a generalized linear model, and thus cannot correct for covariates such as gender or age), nor shapeAnalysisMANCOVA allows you to do a nice longitudinal analysis, e.g via linear longitudinal mixed models. What StatNonParamTest will allow you to do is to analyze any scalar/univariate feature/measurement at every surface location (shapeAnalysisMANCOVA won't allow you to do that). Thus, you could use MeshMath to compute longitudinal, pairwise differences between aligned meshes and then analyze those. This would allow one type of longitudinal analysis, though restricted to perfectly balanced data with exactly 2 time points.

We are working on getting better stats tool into the toolbox, but those are not yet ready.

An alternative would be to use http://www.math.mcgill.ca/keith/surfstat/ (you would need to take the vtk surfaces and convert them to OBJ or the free surfer format first), or possibly even better the longitudinal tool in Freesurfer: https://surfer.nmr.mgh.harvard.edu/fswiki/LinearMixedEffectsModels

The issue with the two tools above is that they can only handle univariate data. You would have 3 options:
a)  computing pairwise differences to an earlier time point (e.g. choosing a baseline and computing differences to that baseline image via MeshMath),
b) compute differences to a template (and introduce a template bias).
c) use the new thickness features, which are univariate. The newest versions of the tool will also compute a medial axis and the distance at the boundary to that axis. The corresponding mesh is called *MedialMesh.vtk and the corresponding local thickness and area are stored both in that vtk as well as a separate text file.

My suggestion would be to use option (c) above with the FreeSurfer tool. While I suggest that, I have never done it myself and you may run into a number of issues involving file conversions/data preparation for using the linear mixed model tool in FreeSurfer.

Best regards
Martin