This is the documentation file for spatially-constrained exemplar-based node parcellation of the human brain at the individual- and state-specific level. For more details, please see Salehi et al., 2017 [1] and Salehi et al., 2018 [2].
The parcellation algorithm is implemented in C++, and is part of the open-source BioImage Suite Project. To run these files, you need to download BioImageSuite source code from here. After downloading BioImage Suite source code (under the name "bioimagesuite32_0b1_src"), you can run the individualized parcellation algorithm via the following:
-
Move the file 'vtkbisIndividualizeParcellation.cpp' to '/bioimagesuite32_0b1_src/Connectivity/'
-
Move the file 'bis_individualizeconnectivity.tcl' to '/bioimagesuite32_0b1_src/bioimagesuite/bis_algorithm/'
-
Build the package following the steps in the BioImageSuite mannual (see here)
-
Make sure you set the environment path, by running the following command:
source bioimagesuite32_0b1_src/build/setpaths.csh- Execute the the individualzied parcellation algorithm using the following command:
bis_individualizeconnectivity.tcl -inp $input -inp2 $parc -indiv_group 1 -blursigma $BW -num_exemplar $K-
$input is the .nii file containing the voxel-level time series.
-
$parc is the initial group-level parcellation that the algorihtm starts from. Please refer to the manuscript for more details.
-
indiv_group option should be set to 1 to generate individualized parcellations, as well as the functional connectivity matrices using the individualize partcellations. If set to 0, it will only output the functional connectivity matrices, using initial group-level parcellation.
-
$BW is the smoothing kernel's bandwidth. In the original work, we set BW=4.
-
$K is the number of nodes in the parcellation. K should match the group-level parcellation. Here we used the parcelation by Shen et al. [3], where K=268 for whole brain analysis, and K=188 for cortical analysis.
For further questions please raise an issue here.
[1] Salehi, M., Karbasi, A., Scheinost, D., & Constable, R. T. (2017, September). A submodular approach to create individualized parcellations of the human brain. In International Conference on Medical Image Computing and Computer-Assisted Intervention (pp. 478-485). Springer, Cham.
[2] Salehi, M., Greene, A. S., Karbasi, A., Shen, X., Scheinost, D., & Constable, R. T. (2018). There is no single functional atlas even for a single individual: Parcellation of the human brain is state dependent. bioRxiv, 431833.
[3] Shen, X., Tokoglu, F., Papademetris, X., & Constable, R. T. (2013). Groupwise whole-brain parcellation from resting-state fMRI data for network node identification. Neuroimage, 82, 403-415.