Hi Ziad,
Thanks for your suggestion.
I tried it and loaded the output file onto a inflated std.141 surface. It looked like a mesh covering the brain surface, but the locations and size of BA areas looked fine.
I then tried to use MapIcosahedron directly by editing the .spec file: replaced the LabelDset name (lh.aparc.a2009s.annot.niml.dset) with lh.BA.thred.annot.niml.dset (the file generated with FSread_annot) and only kept smoothwm and sphere.reg in "States" and "NewSurface" to save time. After running MapIcosahedron, I loaded the output file (std.141.lh.BA.thred.annot.niml.dset) onto a inflated std.141 surface and the BA areas looked the same as shown with tksurfer on the original surface (no "mesh" looking).
So, I guess for converting a surface into standard mesh, MapIcosahedron may do a bettter job than SurfToSurf with a M2M file. I am not sure if this is a disply / colormap issue. Please correct me if I am wrong here.
BTW, can MapIcosahedron only convert the data assigned with the option "-NN-dset_map" or "-dset_map" without converting all the surface data included in the .spec file?
Thanks,
Xiangchuan