Repository accompanying the article "Investigating structural and functional aspects of the brain’s criticality in stroke.". Contents:
- code with simulation of Ising and HTC dynamics defined on arbitrary networks,
- various network generators, and
- code for performing cluster analysis
- numpy
- scipy
- numba
- networkx
- scikit-image
- jupyter (only for jupyter notebook tutorials)
- matplotlib (only for jupyter notebook tutorials)
Tested using:
- numpy (1.21.0)
- scipy (1.8.0)
- numba (0.55.1)
- scikit-image (0.19.3)
- networkx (2.8)
Before starting cluster analysis, create files containing simulation
results (of the HTC model and Ising model). Two executable scripts are
used for this purpose: simulation_htc.py and simulation_ising.py.
The simulation scripts require configuration files. Examples used in
tutorial-notebooks are located in cfgs/ directory. Then, the script
are run using following command:
./simulation_htc.py Path/to/Config/File.ini
or
./simulation_ising.py Path/to/Config/File.ini
For the structure of configuration files see sim_htc_config.ini and
sim_ising_config.ini which contain detailed information.
The simulation scripts saves all data in output.npz file
in a desired directory.
The standard Hagmann et al.'s connectome is located in hagmann_connectome.npz
file. It consists of a connection matrix and labels assigning regions-of-interest
(ROIs) to appropriate resting-state-networks (RNSs).
Examples for preparation of Ising grids (for Ising simulations) and modified
connectomes (for HTC model simulations) can be foud in gen_connectome.ipynb.
Examples of cluster analysis are prepared in cluster_analysis.ipynb notebook.
Clustering routines for both models can be found in clusters_htc.py and
clusters_ising.py. In the case of the HTC model clusters are found using graph
approach (networkx), while in the Ising model, clusters are computed using image-analysis
based methods (skicit-image).