Hello,

Is there a recommended pipeline I can use for volume based analysis in subject space? Or is volume based subject space analysis not possible at all?

Thanks.

PS: Sorry for double posting, I was not aware that there was a special forum for CONN.

Hi Tamer,

It is perfectly fine/possible, although of course you will not be able to run volume-based group-level analyses since those analyses require some form of inter-subject coregistration at the level of voxels.

For example I would probably start with the pipeline named 'preprocessing pipeline for surface-based analyses (in subject-space) when FieldMaps are available' and then simply remove the last three functional steps ("functional Resampling of functional data within the cortical surface", "functional Label current functional files as "surface-space functional data"" and "functional Smoothing of surface-level functional data") and replace those with a simple "functional Smoothing (spatial convolution with Gaussian kernel)" step instead. That will fully preprocess your data while keeping it in subject-space (coregistered with each individual's anatomical file). 

Hope this helps

Alfonso

Originally posted by Tamer Gezici:

Hello,

Is there a recommended pipeline I can use for volume based analysis in subject space? Or is volume based subject space analysis not possible at all?

Thanks.

 

PS: Sorry for double posting, I was not aware that there was a special forum for CONN.

Dear Alfonso,

Thank you so much for your response.

These are the steps I used in my pipeline. I don't have a fieldmap, so I used this pipeline:

'functional_label_as_original'    'functional_realign&unwarp'    'functional_label_as_realigned'    'functional_slicetime'    'functional_art'    'functional_coregister_affine_noreslice'    'functional_label_as_subjectspace'    'functional_smooth'    'functional_label_as_smoothed'    'structural_segment'

However, the noise mask and the structural image don't seem to be registered properly.

Dear Tamer,

Right, sorry I forgot to mention this. After preprocessing, if your analyses are going to be in subject-space make sure to change the default "analysis mask" (found in the Setup.Options tab, which by default is defined in MNI-space) to either 'implicit mask' (automatically derived from the functional data) or to another explicit mask of your choice -which should be also in subject-space-.

Hope this helps

Alfonso

Originally posted by Tamer Gezici:

Dear Alfonso,

 

Thank you so much for your response.

 

These are the steps I used in my pipeline. I don't have a fieldmap, so I used this pipeline:

'functional_label_as_original'    'functional_realign&unwarp'    'functional_label_as_realigned'    'functional_slicetime'    'functional_art'    'functional_coregister_affine_noreslice'    'functional_label_as_subjectspace'    'functional_smooth'    'functional_label_as_smoothed'    'structural_segment'

 

However, the noise mask and the structural image don't seem to be registered properly.

 

 

 

Dear Alfonso,

Thank you so much for your help so far. The analysis worked fine, but I noticed that the CONN's network and atlas ROIs do not match the subject space.

Is there a way to denormalize these ROIs using CONN, or should I denormalize these myself and set them as a subject specific ROI manually?

Dear Tamer,

Right, those atlases are both defined in MNI-space, so naturally they will not be aligned to subject-space data. I personally find it simpler to run analyes in MNI-space and then bring back any results of interest to subject-space when necessary, compared to keeping the analyses in subject-space and bringing any necessary MNI-space templates or masks to each individual's space. See for example this thread for an example of how to bring back results from MNI-space back to subject-space. 

Hope this helps

Alfonso

Originally posted by Tamer Gezici:

Dear Alfonso,

 

Thank you so much for your help so far. The analysis worked fine, but I noticed that the CONN's network and atlas ROIs do not match the subject space.

Is there a way to denormalize these ROIs using CONN, or should I denormalize these myself and set them as a subject specific ROI manually?