Dear Elveda,
there are two thresholds, 1) the height or voxel threshold and 2) the spatial extent or cluster threshold. Only those voxels whose actual timecourse is close enough to the expected timecourse will survive the height threshold. The extent threshold is used to define clusters of activation = a certain number of neighbouring voxels that initially surpassed the height threshold. If there are no activated voxels, then you can't define clusters. This can well happen on single-subject level depending on the chosen threshold / the paradigm.
Concerning the second issue, the main reason for problems during normalisation or segmentation are large displacements of the anatomical volume compared to the MNI space / SPM templates. There might be some instances in which one of the two approches 1) standard normalisation ("Normalise: Estimate & Write") or 2) "Segmentation" followed by "Normalise: Write", fails whereas the other still works well, but in general this points to displacements. So make sure the data is more or less oriented like the MNI space. See older posts for details, for example passage 2) in https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=spm;5e4642f3.1306
Hope this helps,
Helmut
|