The goal of this MR is to change the scores the voxel-wise WB are resampling. Before, The WB was resampling the raw T- or F-scores. After this MR, it is resampling the equivalent Z- or X-scores. This is to match the behaviour of the SPM toolbox, which does the resampling this way.
The results are matching those of the SPM toolbox except for some WB p-values. This discrepancy comes from the fact the p-values are computed using
(bootstrappedScore >= originalScore) in the fsl version while they are computed using
(bootstrappedScore > originalScore - tolerance) with
tolerance = 10^-4 in the SPM version. The latter is due to the fact that the equality between floats may not work as expected.
Because we are quite in a hurry for the next release of FSL, I will already merge this request. I believe the discrepancy above can be dealt later in another MR and I will add this fact in a new issue.