Hi Martin,

at the moment I just compare manual and automated segmentation, I asked for longitudinal approach to make sure I have within-subjects comparisons. From my understanding, shapeAnalysisMANCOVA is primarily used for between-subjects designs (comapring groups)?

Just to make sure I did not miss anything,  I recalculated the Meshes in Slicer 4.3 using the latest spharm-pdm extension; my system is Fedora 17, TCL/TK 8.5; ITK 3.4; VTK 5.8, QT 4.3.
PWD is  ../Mesh/SPHARM; The command is: MeshMath T0256_left_man_pp_surfSPHARM.vtk difference.txt -subtract T0256_left_automated_pp_surfSPHARM.vtk
It returns: 0
Incomplete file record definition
NDims required and not defined.
MetaObject: Read: MET_Read Failed
MetaMesh: M_Read: Error parsing file
Incomplete file record definition
NDims required and not defined.
MetaObject: Read: MET_Read Failed
MetaMesh: M_Read: Error parsing file

I also tried with eallingened meshes, same results. I visualized beforehand with ShapePopulationViewer and found that _pp_surfSPHARM were not alligned at all (both surfaces visible only, when Allign button pushed), Basically both Meshes look good and differences in segmentation are visible. In terms of volumes, the second (baseline) volume is larger. Switching the order of both volumes results in the same error.

Good news on StatNonParamTest: I have successfully tested the binaries of the new version (i.e. -- help works). Thanks so much!

Any ideas on MeshMath? Would you need Meshes?

Best regards,
Georg