Surface Template for Children

Hi,

We are carrying out an fMRI study for 3-year-olds (from a non-standard international population) and are currently evaluating our options for creating both surface and volumetric templates without any adult reference templates. We have thoroughly read the FreeSurfer wiki, and the template creation steps in FreeSurfer always seem to involve adult priors. Our problem with this is that we are unsure if it might bring bias into our analysis.

We are thinking about these questions and would appreciate any input:

  1. Is it possible to create the surface template for FreeSurfer without the adult reference image (and how complex, it terms of time and effort, would it be)?
  2. How large would you expect the bias to be if using a study specific template using the standard FreeSurfer process (using adult priors)?
  3. Is it possible to create a volumetric template for children without an adult reference image (ANTs template construction e.g. starts with an initial template)?
  4. What would the bias/effect be if creating a pediatric volumetric template by starting with an adult template?

Thank you,
Anne

I dont have prior experience with child, but I do creating templates.

For the volumetric approach, I would use ANTs tools. I have always obtained good results. They have a set of .sh scripts which include some priors. You can check this one and this other post. Also, if you dont have prior experience with ANTs, take a look to their handsout pdf.

For the surface template, I would propose two approaches:

  1. Segment the volumetric template with Freesurfer. I have seen this for MNI152, so, why not for child?
  2. Follow approach proposed in Development HCP surface paper: Construction of a neonatal cortical surface atlas using Multimodal Surface Matching in the Developing Human Connectome Project - ScienceDirect
  3. Brute-force the creation of template. Take a look to FSL’s FA-template creation. Basically, it register all the MRI (here surface ) to an specific individual, and choose the one that opt the registrations. Then you can create the average of those registered-surface using mri_average
  4. Try to follow last part of this tutorial

Extra refs that might help you?:

  1. MCRIBS/lib/sphere_templates at master · DevelopmentalImagingMCRI/MCRIBS · GitHub
  2. Development HCP surface paper: Construction of a neonatal cortical surface atlas using Multimodal Surface Matching in the Developing Human Connectome Project - ScienceDirect
  3. dHCP 3years template surface: Cortical Surface Template – Brain Development
2 Likes