Fmriprep connectivity analysis

Dear NeuroStars,

I am trying to get a correlation matrix of voxel time-series, that sit along the gray matter (GM) of a common template. So, I need an preprocessed EPI image and GM template (based on t1), that both lie on the same template.

My target template is MNI152_2mm_brain. Using fmriprep (1.1.2), I could get EPI images realigned to this target template (figure, right). And, it also produced a beautiful GM probabilistic map realigned to the MNI152NLin2009cAsym template (figure, left).

The question is: How to bring the GM map to my target template? Is there already some fmriprep options to specify MNI152_2mm_brain as the target space for anatomical preprocessing? Or, shall I go for nipype functions for the volume resampling?

Thanks a lot1

Hi @sheyma,

I would really recommend you to use the MNI152NLin2009cAsym template that comes with fMRIPrep. However, if for whatever reason that is not possible, you can:

  1. Use the GM map that comes with the MNI152 2mm linear directly. I’d say the registration error can be dismissed, as this map will be a lot smoother than that of MNI152NLin2009cAsym.

  2. Resample the GM map into your target MNI152 2mm linear. See Normalization of fmri data without changing voxel size for a similar problem). For improved alignment, you may run a nonlinear registration between MNI152NLin2009cAsym and MNI152 2mm linear.

  3. Work on the native space of the EPI

Hi @oesteban,

Thanks for your beautiful suggestions!

I will stick on this suggestion number 0, but combine it with suggestion number 2 :slight_smile: .

Resampling the GM map, which is beautifully aligned on MNI152NLin2009cAsym, to 2mm would make it aligned to EPI.

Thanks a lot!

Please check my answer to your other thread. Just to clarify: “mismatched sampling grid” != “misaligned”. Samples may be in different locations in space, but information may be indeed aligned (in register).

Perfect! I just checked your answer on the other thread. Cool!