Skip to content

takwan/Network-modulity-based-parcellation

Repository files navigation

Project

Thalamocortical triple-network dysconnectivities in psychosis (Kim, M., Kim, T. et al. 2022, Schizophr Bull).

ref to Hwang, K. et al. 2017, J Neurosci: Modularity-based thalamic parcellation.


Atlas created using resting-state fMRI data

  • ./cortMod_parc_6mods/cortNet_mods.nii.gz: cortical modular networks
  • ./thalMod_parc_6mods/thalNet_mods.nii.gz: thalamic modular networks
  • data from 273 individuals with (high risk for) psychosis or without any psychiatric disorder (healthy volunteers).
  • data acquisition from Dept of Psychiatry at Seoul National University Hospital (PI: Prof. Jun Soo Kwon).

Codes

Cortical network identification

1. Construction of cortico-cortical connectivity matrix (subject-level)

  • used the Godon atals (333 regions).
  • 01_thr_conn.m: iteratively thresholded the connectivity map of each subject, ranging from 15% to 1% in steps of 0.1% connectivity density.
  • returns ./linkArray/Subj##/clink_thr###.net.

2. Identification of cortical network modules (subject-level)

  • 02_prepCortmod.sh: used the InfoMap algorithm.
  • returns ./linkArray/Subj##/modOut/clink_thr###.clu.
  • sample data: Subj01

3. Finding a consensus from the modularitiy data (subject-level)

  • cortmod_01_update.ipynb: updating a consensus matrix across the ranges (descending order).
  • returns ./linkArray/consMats/consMat_Subj##.csv.

4. Group-level cortical network mapping (repeat steps 1~3)

  • cortmod_02_group.ipynb: avaraging subject-level modularity matrices, returning ./linkArray/consMatDM_low.csv.
  • 03_cortmod_thrConn.m: thresholding group-level cortico-cortical connectivity matrix, returing ./linkArray/links_group/clink_thr###.net.
  • 04_runCortmod_grp.sh: identifying group-level cortical network modules, returing ./linkArray/links_group/modOut/clink_thr###.clu.
  • cortmod_02_group.ipynb: updating a group-level consensus matrix across the ranges, returning ./linkArray/links_group/modOut/cortmod_grp.csv.
  • 05_cortMod_parc_6mods.sh: merging nodes to creat modular networks.

Thalamic network parcellation

06_thal_parcel.m:

  • used the Morel thalamus atlas
  • partial correlation between thalamic voxels and mean timeseries of each cortical module.
  • controlling for signals from rest of the cortical modules.
  • a winnter-take-all parcellation approach: assigning network labels where having the largest corr coeff.