To reproduct the results from the analysis of the Synchropioid project, you will need to go through each steps of this tutorial.
a/ To download the following tutorial:
Go to the public repository https://github.com/hcherkaoui/analysis_synchropioid
Click on the green 'Code' button
Click on the 'Download ZIP' button
After downloading the repository, go to your chosen downloading folder and extract the downloaded compressed folder with
unzip analysis_synchropioid-main.zip
To start the tutorial, go to its root folder:
cd analysis_synchropioid-main
a/ This step installs the tools to perform the analysis.
Install key packages needed for the tutorial:
sudo apt install python3-pip liblapack-dev docker.io
b/ Install the needed Python tools with the requirements.txt file:
python3 -m pip install -r requirements.txt
/!If pip struggles to install all package, re-run multiple time the command to force pip to resolve the problems.
Need steps 0/ and 1/
a/ Running generate_symbolic_dicom_dir.py and bidsify_synchropioid.py on 20 CPUs to make the Synchropioid dataset BIDS compliant:
cd 00_prepro_fmri/ ./bidsify_synchropioid.py -v -i /biomaps/acquisitions/dicom/SIGNA_PET/ -o /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/ --cpu 20
Need steps 0/, 1/ and 2/
a/ Launching fmriprep on a console:
sudo docker run -ti --rm -v /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/:/data:ro -v /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/:/derivatives:rw -v /biomaps/freesurfer/license.txt:/opt/freesurfer/license.txt:ro poldracklab/fmriprep:latest /data /derivatives/ participant --output-space MNI152Lin --fs-license-file /opt/freesurfer/license.txt --fs-no-reconall --random-seed 0 --dummy-scans 10 --nthreads 20
/!If you need to restart a failed preprocessing step, first you will need to delete the previous failed produced folder:
rm -rf /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/fmriprep/
Simply re-run the docker command after the deleting step.
b/ To match the BIDS format, rename the produced preprocessing folder
mv /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/fmriprep/ /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/derivatives/
Need steps 0/, 1/, 2/ and 3/
a/ Open the each fmri-prep .html report (one per subject) under the folder /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/derivatives/ to check each preprocessing step.
b/ Run MRI-quality-control to add a simple quality check:
mriqc /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/derivatives/ . participant
Need steps 0/, 1/, 2/, 3/ and 4/
a/ Running decomposition_multi_subjects and decomposition_multi_groups.py to estimation the HRFs for each subject:
cd 02_hrf_est/ python3 decomposition_multi_subjects.py --max-iter 100 --seed 0 --preproc-dir /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/derivatives/ --results-dir results_hrf_estimation_single --cpu 20 --verbose 1
Those commands will produced the folder results_hrf_estimation_single and results_hrf_estimation_group that store the produced results.
Need steps 0/, 1/, 2/, 3/ and 4/
a/ Running estimation_connectome:
cd 03_connectome/ python3 estimation_connectome.py --preproc-dir /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/derivatives/ --result-dir results_connectome --verbose 1
Those commands will produced the folder results_connectome that store the produced results.
Need steps 0/, 1/, 2/, 3/, 4/, 5/ and 7/
a/ Produce the silouette curve w.r.t the temporal regularization value plot for the HRF estimation:
cd 05_plotting_hrf_est/ python3 plot_silhouette_score_per_params_single.py --plots-dir plots --results-dir ../03_hrf_est/results_hrf_estimation_single/ --verbose 1
b/ Produce vascular maps for each subjects for the best temporal regularization value for the HRF estimation:
python3 haemodynamic_maps_per_subjects.py --bids-root-dir /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/ --results-dir ../03_hrf_est/results_hrf_estimation_single/ --best-params-file decomp_params/best_single_subject_decomp_params.json --output-dir output_dir --verbose 1
c/ Produce the haemodynamic region scatter plot and the mean haemodynamic evolution plot for the HRF estimation:
python3 plot_haemodynamic_delays_comparison_subjects.py --plots-dir plots --bids-root-dir /biomaps/synchropioid/dataset_synchropioid/fmri_nifti_dir/ --results-dir ../03_hrf_est/results_hrf_estimation_single/ --best-params-file decomp_params/best_single_subject_decomp_params.json --verbose 1
d/ Produce the T-test on the vascular maps plot for the HRF estimation:
python3 plot_t_test_per_run.py --vascular-maps-dir output_dir --plots-dir plots --verbose 1 eog plots/
e/ Produce the temgesic vs contro examples comparisons on the vascular maps plot for the HRF estimation:
python3 plot_temgesics_vs_control_group.py --vascular-maps-dir output_dir --bids-root-dir /media/veracrypt1/synchropioid/fmri_nifti_dir/ --plots-dir plots --task-filter only_hb_rest --verbose 1
All the plots are gathered under the plots folder.
Need steps 0/, 1/, 2/, 3/, 4/, 6/ and 7/
a/ Produce the norm plot for the Connectome:
cd 04_plotting_connectome/ python3 plot_connectome_norm_evolution.py --connectome-dir ../04_connectome/results_connectome/ --plots-dir plots --verbose 1
b/ Produce the temgesic/control prediction plot for the Connectome:
python3 plot_decoding_connectomes.py --connectomes-dir ../04_connectome/results_connectome/ --plots-dir plots --seed 0 --cpu 20 --verbose 1
c/ Produce the learning curve for the temgesic/control prediction task plot for the Connectome:
python3 plot_learning_curve_connectomes.py --connectomes-dir ../04_connectome/results_connectome/ --plots-dir plots --seed 0 --cpu 20 --verbose 1
d/ Produce the T-test on the connectome matrices plot for the Connectome:
python3 plot_t_test_per_run.py --connectome-dir ../04_connectome/results_connectome/ --plots-dir plots --verbose 1 eog plots/
e/ Produce the ANOVA on the connectome matrices plot for the Connectome:
python3 plot_anova_connectomes.py --connectomes-dir ../03_connectome/results_connectome/ --plots-dir plots --seed 0 --cpu 20 --task-filter only_hb_rest --verbose 1
All the plots are gathered under the plots folder.
To add a new subject to the Synchropioid dataset, simply edit the dicom_subjects_list.txt file by adding a newline with the corresponding DICOM directory name (e.g. add a new line S00...).