Hierarchical Brain Networks

Determining the Hierarchical Architecture of the Human Brain Using Subject-Level Clustering of Functional Networks
Teddy J. Akiki, Chadi G. Abdallah
Scientific Reports 9, 19290 (2019) doi: 10.1038/s41598-019-55738-y

Installation

The code relies on HierarchicalConsensus and GenLouvain.

Usage

This will help you generate a group-representative community organization starting from regional (parcellated fMRI) time series.

Mapping community organization at the subject level

Before running the hierarchical consensus clustering algorithm, we first need to map the community organization from individual subjects. Functional MRI time series can be loaded in a nindiv x dtpoints x nroi MATLAB array, where nindiv is the number of individuals, dtpoints is the number of datapoints in the time series, and nroi is the number of ROIs. The null model here is based on Random Matrix Theroy. After you make sure that all relevant files have been added to the path, you can run

for i=1:nindiv
    all_ci{ii,1}=RMT_com(squeeze(TS_all(i,:,:)),n);
end

This will generate a cell array all_ci containing subject-level hierarchical community organization. Note that the number of hierarchies can vary between subjects.

Optional: to reassign singletons, you can use

for i=1:nindiv
    tmptmp_ci=all_ci{i,1};
    newtmp_ci=ci_restoresingleton(tmptmp_ci);
    all_ci{i,1}=newtmp_ci;
end

To organize the data in the cell arrays into an easier to use contactenated array, you can use

all_ci_combined=all_ci{1,1};
for i=2:nindiv
    tmp_ci=all_ci{i,1};
    all_ci_combined=horzcat(all_ci_combined,tmp_ci);
end

Generating a group-level consensus

Now you can use HierarchicalConsensus algorithm (Jeub et al., 2018) to generate a co-classification matrix followed by the concensus clustering algorithm.

C=coclassificationMatrix(all_ci_combined);
[Sc,Tree]=hierarchicalConsensus(all_ci_combined,0.05);
[Sall,thresholds]=allPartitions(Sc,Tree);
consensusPlot(C,Sc,Tree);

Brain maps

You will find the hierarchical community partition solutions described in the study in the brainmaps folder.

Extracting A424 time series

Get_A424_TS.m can be used to extract the A424 bold time series from any fMRI.nii.gz in the standard space.