Menu
▾
▴
#33 Relax the requirement for masking in GroupMap
In the groupmap and subsequent phase, a voxel is only considered in the statistical analysis step if and only if all subjects contributed values to the voxels. While this conservative move is good from a statistical point of view and make analysis easier to understand, at times, slight misalignment in individual subjects can reduce the voxels considered in the statisical analysis significiantly.
The proposal here is to include voxels where it is only missing one or two subjects. (The actual implementation should permit exclusion of arbitrary number of subjects.) This should only be an option and exercised with care. Nonetheless, it is a useful option to have.
Logged In: YES
user_id=783811
Originator: YES
Note that if this proposal is adopted, there must be a change in tsafx_extractstats.sh as the guarantee that "all statistical images used in the extraction of statistical has voxels in all clusters in the cluster map" is no longer true.
Logged In: YES
user_id=1377989
Originator: NO
I don't think this is a route we should go into. I don't know whose idea it was, but I think it is bad. One is reminded of the p-value slider for SPM (http://people.hnl.bcm.tmc.edu/cuixu/xjView)...