Normalization to a pediatric template. Jump between SPM and FSL/

1

Dear all,

At the moment, I am preprocessing a children fMRI dataset using SPM in combination with the ArtRepair toolbox (Mazaika, Whitfield-Gabrieli, Reiss, & Glover, 2007) which improves the fMRI analysis for high motion pediatric subjects.
SPM automatically normalizes the children’s functional images into the space defined by the adult MNI152 template. Since children’s brains are a bit smaller, I would like to create an additional pipeline in which I normalize the functional images to a pediatric template instead of an adult template.
Therefore, my question is three-folded:

  1. Is normalization to a template other than MNI152 possibly within SPM12?
  2. If not, can I jump between SPM and FSL, with FSL doing the normalization to a pediatric template? The other preprocessing steps, or at least a large part of them would need to be done in SPM12 since the ArtRepair toolbox is Matlab/SPM based.
  3. If it is not possible with FSL, is there another method in which this could work (e.g., Python,…)?

Thank you in advance!

Sofie

2

Hi Sofie,

Is normalization to a template other than MNI152 possibly within SPM12?

It is, but you need tissue probability maps (TPM) for each tissue type. If you have a template then you would need to segment it and then feed these TPM to the normalization step in SPM.

Another way would be to use nipype and set up a pipeline to do all preprocessing in SPM except the normalization step which you could do with FSL, or better yet with ANTs. You can even create your own template with ANTs based on your subject’s brains and normalize that to the MNI152 or the pediatric template. Then use the transformations to warp each brain to fit this newly created template. Nipype is a bit of a hassle to learn, though. If you have someone around who’s worked with it before that would help you a lot.

Once you have the normalized images (regardless of method, i.e. SPM or FSL or ANTs) you can load them into SPM to run the statistics on them.

1 Like