Analysis code for the manuscript "Characterizing spatiotemporal white matter hyperintensity pathophysiology in vivo to disentangle vascular and neurodegenerative contributions"
We share all analysis code, code used for generating plots, group-level raw results (e.g., 3D volume maps for spatial clusters, linear model outputs, etc.) and visualizations included in the paper.
To run the code, the user needs to assign the working directory to the directory of the script (e.g., analyses/1_norm_mod), load the required dependencies (see below), and then run each script in order.
No individual-level data from UKB or ADNI can be shared for confidentiality reasons.
For questions/comments, please reach out to Olivier Parent (olivier.parent@mail.mcgill.ca)
Description of the files made available in data/. All files are shared in both minc and nifti format.
Custom UK Biobank template space (T1w)
- UKB_template_T1: 1mm isotropic
- UKB_template_T1_2mm: 2mm isotropic
- UKB_template_mask: mask
Custom UK Biobank template space (FA)
- UKB_template_FA: 1mm isotropic
Final data-driven WMH parcellation based on pathophysiology
- WMH_parc_patho: in UKB space
- WMH_parc_patho_MNI: in MNI ICBM152 non-linear symmetric 09c space
Prevalence maps for WMH and NAWM
- NAWM_prevalence: normal-appearing white matter labels prevalence
- WMH_prevalence: white matter hyperintensity labels prevalence
- WMH_mask: mask of voxels with >1 WMH label
Other WMH parcellation used for comparisons (all in UKB space)
- WMH_parc_vascular: parcellation of arterial territories (from Liu et al., 2023, Scientific Data, https://doi.10.1038/s41597-022-01923-0)
- WMH_parc_lobar: lobar parcellation
- WMH_parc_pv_deep: periventricular/deep parcellation
- WMH_parc_fiber: fiber type atlas
- R/4.1.2
- python/3.9.8
- PCNtoolkit/0.35 (python package; https://github.com/amarquand/PCNtoolkit)
- Spectrum/v1.1 (R package; https://cran.r-project.org/package=Spectrum)
- pySuStaIn (python package; https://github.com/ucl-pond/pySuStaIn)
- neuromaps (python package; https://github.com/netneurolab/neuromaps)