Hi Cesar,
That is a mistake by FSL. Since the dataset is not a
time series, it should indeed have dim4=1 (dim4 is
reserved for a time axis, according to the standard).
It is dim5 that should be 11. To get more detail
about the header information, try nifti_tool:
nifti_tool -disp_hdr -infile melodic_IC_good.nii.gz
If you want to make an incorrectly formatted dataset to
appease FSL, create it with 3dTcat instead of 3dbucket.
- rick