Hi Georg

1) you could do that, i.e. that the textiles created with -MagNormDir and put those into a single datafile and work with StatNonParamTestPDM. We did not do that. In that particular case, we had data from multiple studies and multiple raters etc and wanted to covary for that. We did a bit more hands-on work by using R (Eric Maltbie did all the stats on that one). Like it says in the paper: " Repeated measures ANOVA (R statistical package) was employed to calculate the statistical significance of the local signed difference at each location independently resulting in raw significance maps (Fig. 4). All surfaces, i.e. from all subjects and all raters were included. RaterID, subjectID, and hemispheric side were modeled as covariates." I don't have the R scripts anymore that we used, but we simply loaded those text files into R, and ran locally a repeated ANOVA and wrote p-values out as a text file that was manually made to conform to the text format used by MeshMath. The p-values were added via MeshMath -KWtoPolyData . 

2) that depends on how the segmentation data is aligned. Are the segmentations (manual vs automatic) in exactly the same space, i.e. you could load them into a viewer such as Slicer and the segmentations are very close on top of each other? If so, then best would be to use surfSPHARM. If the segmentation images are not aligned, then _procalign would be needed to align the data (but in your case you don't want to do that if the segmentations are already aligned).

Martin