This repository provides the code used to analyze the data of migraine patients in the paper: Involvement of the cerebellum in structural connectivity enhancement in episodic migraine
The overall pipeline can be seen in the picture below:
You'll need MRtrix, FSL, and MATLAB installed on the computer you'll use to run the code. I used MRtrix version 3.0.3, FSL version 6.0.5 and MATLAB 2023a version (but older versions of MATLAB should also work).
Then download this repository into a folder of your choosing by going to that folder and then cloning it through the terminal:
git clone https://github.com/anamatoso/connectivity-analysis-diffusion.git
The data directory structure should be the following:
.
├── 01_tractography.sh # Tractography script
├── 02_count_streamlines.sh # Count streamlines script
├── bvals.bval # b-values file
├── data # Data folder
├── sub-controlXXX_ses-[SESSION] # Folder with the dMRI files
│ ├── sub-controlXXX_ses-[SESSION]_clean.nii.gz # dMRI image (already preprocessed)
│ ├── sub-controlXXX_ses-[SESSION]_clean_mask.nii.gz # dMRI image mask
│ └── sub-controlXXX_ses-[SESSION]_rotated_bvecs.bvec # b-vectors
├── sub-controlXXX # Anatomic imge folder
└── sub-controlXXX_restored-MPRAGE_brain.nii.gz # T1-weighted image
└── ...
├── matrix_data # Output data folder (will be created automatically)
├── ALL116 # Folder with the connectivity matrices using the AAL116 atlas
└── schaefer100cersubcort # Folder with the connectivity matrices using the schaefer100cersubcort atlas
├── streamline_count # Output streamline count folder (will be created automatically)
└── streamline_count_JHUlabels
├── sub-controlXXX_ses-[SESSION]_JHUlabels.txt
└── ...
├── 03_connectivity_analysis.m # MATLAB script for the connectivity analysis
├── 04_streamline_count_analysis.m # Matlab script to analyse the streamline count
├── dados_clinicos_[GROUP].csv # CSVs that contain the clinical data of each group (patients and controls)
└── ... # Other files and folders
Run the file 01_tractography.sh
for every patient folder you have with dMRI data (sub-controlXXX_ses-[SESSION]).
Should your directory structure differ, change Step 1 accordingly so that the variables point to the correct files.
Then run the 02_count_streamlines.sh
file which will create the streamline_count folder that will store the number of streamlines in each region of the JHUlabels white matter atlas.
To make it easier and compartmentalized, create a Python environment using venv. In my case, pyenv is the name of the environment I chose, but you can name it whatever you want. However, but the rest of the code will assume that the name is pyenv, so change it appropriately if necessary.
python -m venv pyenv
Then activate it, install the packages from the requirements file, and deactivate it:
source ./pyenv/bin/activate
pip install -r requirements.txt
deactivate
You'll need to install the NBS toolbox and add it to MATLAB's PATH.
Then open MATLAB, go into the repository and run 03_connectivity_analysis.m
and 04_streamline_count_analysis.m
. If it yields an error due to the name of the files that can be easily solved by renaming the files.
Note: don't forget to add the auxiliary functions folder to the PATH in MATLAB before running the files.
If you want to cite our work please use:
Matoso, A., Fouto, A.R., Esteves, I. et al. Involvement of the cerebellum in structural connectivity enhancement in episodic migraine. J Headache Pain 25, 154 (2024). https://doi.org/10.1186/s10194-024-01854-8
@article{matoso2024involvement,
title={Involvement of the cerebellum in structural connectivity enhancement in episodic migraine},
author={Matoso, Ana and Fouto, Ana R and Esteves, In{\^e}s and Ruiz-Tagle, Amparo and Caetano, Gina and da Silva, Nuno A and Vilela, Pedro and Gil-Gouveia, Raquel and Nunes, Rita G and Figueiredo, Patr{\'\i}cia},
journal={The Journal of Headache and Pain},
volume={25},
number={1},
pages={1--11},
year={2024},
publisher={BioMed Central}
doi={10.1186/s10194-024-01854-8}
}
If you need any help reproducing this work or if you encounter any bugs or not expected behaviour, feel free to reach out or to submit an issue or a pull request.