A pre-adolescent imaging-based brain parcellation

[JessyD]

Research question(s)

• Can we create a brain parcellation using data from pre-adolescent subjects?
• How similar/dissimilar is this parcellation compared to an adult parcellation?

Description

This project aims to create a brain parcellation for pre-adolescents subjects using the ABCD dataset. When performing a neuroscience study, it is common to extract regions of interest using a parcellation scheme. Although many parcellation schemes exist, these parcellations have been derived from adult data and might lead to biases when applied to a different age group. In this project, we aim to create a pre-adolescent brain parcellation following the methods described by Yeo et al. (2011) using the ABCD dataset.

Tools and algorithms to be used

freesurfer, the original code for the parcellation is in Matlab (it is an open question if we want to translate it to Python or just adapt the original code for our dataset)

Skills we could use help with

any programming skills or familiarity with the ABCD dataset are welcome

General idea of how to structure the project

0. Explore the example code provided to replicate the paper

1. Establish the common spherical coordinates

2. Apply clustering approach to define the boundaries of the functionally distinct cortical regions (von Mises-Fisher Distribution to identify clusters in a sphere)

   1. Define the optimal number of clusters (7-17 networks)
   2. Compute the confidence for each spatial location belonging to the assigned network (Silhouette metric)

3. Validate results on an independent part of the dataset

Useful literature

Yeo et al. (2011) Review about parcellations of the human brain

Suggested keywords/tags

parcellations neuroimaging replication resting-state functional connectivity cluster analysis

[dnkennedy]

Hi @JessyD , nice project. Do you have any initial thought as to how to define 'optimal' in your step 2.i. "Define the optimal number of clusters (7-17 networks)"?

[JessyD]

Hi @dnkennedy, as I thought about replicating the Yeo paper using the ABCD dataset, I thought of using the same method that they did. They looked at the instability of the clustering approach using different numbers of clusters (Fig 6 on the Yeo et al. 2011 paper). But this was just a first idea; I would love to know if you have any other suggestions.

[hhopman]

Hi Jessy,

I would love to collaborate on this project. Since the code is already there, I thought maybe it is also interesting to compare baseline and follow-up data to examine the stability of the parcellation.

Maybe we can organize a Zoom meeting to further exchange ideas?

Best, Helene

[jessicaflannery]

Hello! I'm also very interested in this idea. I'm not sure how I'd best contribute here yet, but please let me know if you need more folks and/or if you have ideas of what type of help you are looking for? Thanks!