[pymvpa] Surface-based Searchlight?

[Nick Oosterhof]

Hi Matthias,

On Nov 29, 2013, at 6:17 AM, Matthias Ekman wrote:

>> Also please note that the most recent code is available here: https://github.com/nno/PyMVPA
> 
> Thanks for the link to the most recent code! 

You're welcome :-)

>>> Q1: Do I understand correctly that the EPI image must be coregistered to the anatomical before running ``prep_afni_surf`` (i.e. afni is not doing that 'magically' otherwise)?
>>> 
>> You can coregister surfaces to either an anatomical or EPI image ('-a' or '-e' options). If you have both, you should coregister them yourself, for example using AFNI's align_epi_anat.py.
>> 
> Just to make sure I really understand the difference between the -e/-a options. In case I want to perform the SL analysis in the native EPI space I'd follow these steps:
> 
> 1. Freesurfer/recon-all based on the T1 anat
> 2. coregister T1 to the EPI using e.g ``align_epi_anat.py``
> 3. use the following command (using only "-e")
> 
>     pymvpa2-prep-afni-surf -s SUBJ_ID -d fs/surf/ -e func/func.nii.gz -r OUTPUT_DIR -l 4
> 
> whereas if I coregister my EPI to the T1, I would use this command (using "-e" and "-a")
> 
>     pymvpa2-prep-afni-surf -s SUBJ_ID -d fs/surf/ -a anat/anat.nii.gz -e func/func.nii.gz -r OUTPUT_DIR -l 4
>     
> Is that correct? 

Almost:

- the options '-a' and '-e' are mutually exclusive. In the latter case (after aligning EPI to T1) you can use either '-e' or '-a', but not both.
- note that you can also coregister the T1 to the EPI (so that the T1 is transformed, not the EPI). Also then you can use '-e' or '-a'.
- assuming you have a good EPI image you can actually skip the T1-EPI coregistration step (step 2 you mentioned above) and run it with just '-e'. It well then take the anatomical from FS's recon-all and the output dir will have an anatomical (with suffix al2exp - "aligned to experimental volume") that is aligned to the EPI. This may be the preferred usage.

Also: 
- I recently added experimental support for images in template space (MNI/Talairach) - but this has only been tested properly with anatomical images ('-a'), not EPIs ('-e'). So in that case EPI-T1 alignment should be done manually.
- Always make sure to check the alignment visually. Usually the '_e3' files from AFNI's @AddEdge (generated by pymvpa2-prep-afni-surf) are most suitable for this. The latest version of the script should also generate a few "qa*.png" bitmap images that should help in inspecting the alignment.

Please let me know if you have further questions.

best,
Nick