Registering dHCP fMRI data

,
#1

Dear friends

I have a question regarding registering dHCP fMRI data (task-rest_desc-preproc_bold.nii) to another alas, e.g. UNCInfant012Atlase.
I already have downloaded the fMRI pipeline and anat_pipeline data. Would you please guide me in this matter?

Thank you,
Reza

#2

Hi Reza,

If you have a structural T1 scan for the subject, you can calculate an affine transform to your atlas space by registering the subject T1 to the atlas T1 image. You should use a subject preprocesssed T1 that has been skull stripped, and align it to the skull stripped T1 from the atlas (corresponding to the age that is relevant to your study neonatal, 1 y.o, or 2 y.o.). The quickest/easiest way to do that transformation would probably be FSL’s flirt or fnirt command, although antsRegistration from ANTs might perform better. Then, after you calculate the transform, apply it your fMRI image using flirt/fnirt (if using FSL), or antsApplyTransform (if using ANTs).

Hope this helps!
Steven

#3

Hi @rezanazari95 and @Steven

We already provide a high-quality QC’d nonlinear transform for each subject to the 40-week dHCP atlas space (extdhcp40wk). You could register the two atlas spaces (UNCInfant012Atlas and extdhcp40wk), and then combine the inter-atlas transform with each single subject transform.

Furthermore, if you already have a UNCInfant012Atlas to MNI (MNI152NLin6ASym or MNI152NLin2009aAsym) then we provide transform to these atlases here:

Just make sure that you combine the transforms appropriately before applying them so that you get a single resampling and do not accumulate errors due to multiple interpolations.

Hope this helps, Sean