[pymvpa] Antwort: Re: Antwort: Re: Loading normalized data with 'get_model_bold_dataset' fails

[Yaroslav Halchenko]

On Fri, 02 Jun 2017, lydia.hellrung at econ.uzh.ch wrote:

>    No, they all have 195 volumes, but the geometry information is apparently
>    saved differently.
>    The output says:

>    ./sub002/BOLD/task001_run001/bold_irbadvols.nii.gz    float32 [112, 112,
>     27, 195] 1.96x1.96x4.06x2.00  #exts: 1 sform
>    ./sub002/BOLD/task001_run001/bold_iwrbadvols.nii.gz   float32 [105, 127,
>    105, 195] 1.50x1.50x1.50x0.00  #exts: 1

>    while the first file could be loaded, the second one not. Could that be an
>    issue? And in case, what would be needed?

well -- you are saying   flavor='iwrbadvols'  so only that one should be
loaded, and I guess the 'brain.nii.gz'  mask is of the 'compatible'
dimensionality I guess.  otherwise I would have expected it to fail
earlier.

But the above shows the "culprit" --   1.50x1.50x1.50x0.00   so your TR
there is set to 0 sec...  so time indices are degenerate and I guess
that played out somewhere (may be compressed everything into a single
sample or smth like that)

You could fix it  via a quick python code using nibabel or using
nifti_tool  ... --help gives

       e.g. modifying nifti_1_header fields:
            -mod_field descrip 'toga, toga, toga'
            -mod_field qoffset_x 19.4 -mod_field qoffset_z -11
            -mod_field pixdim '1 0.9375 0.9375 1.2 1 1 1 1'

-- 
Yaroslav O. Halchenko
Center for Open Neuroscience     http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik