compensatory_decline_shapes_pd_progression (under construction)

This repository will contain all analysis code used in "Parkinson’s disease progression is shaped by longitudinal changes in cerebral compensation" - Johansson et al. in preparation.

Repository folder structure

|_ Data |_ ROIs
|__ Johansson2024_BRAIN...
|__ n2_ROIs...
|_ scripts
|__ dwi
|___ dti
|__ freesurfer
|__ tb_fMRI
|___ 1st_level
|___ group_level
|____ clinical_correlations
|____ group_comparisons
|__ longitudinal...rRmd
|_ templates
|__ 50hc50pd_b0-avg
|__ FMRIB58_FA
|__ HCP1065_FA
|__ HCP1065_MD

Data supporting figures 1-5

All data used to generate plots can be found in the following link: https://surfdrive.surf.nl/files/index.php/s/O3ygV8gjUjx7oca. For Figures 3-4, statistial derivatives from imaging analyses are also included.

Clinical characteristics and demographics

Directory: /scripts/longitudinal_clinical_and_task-based_analysis.Rmd

• Demographics
• Clinical scores
• Symptom progression
  1. Longitudinal mixed effects modelling
  2. Co-variance with cognition and medication

Task performance

Directory: /scripts/longitudinal_clinical_and_task-based_analysis.Rmd

• Response times
  1. Longitudinal mixed effects modelling
  2. Clinical correlations
• Additional behavioral metrics: accuracy, misses, button variability, switching variability

MRI

BIDSification

Example code:
module load bidscoin/4.0.0
source activate /opt/bidscoin
cd /project/directory
bidsmapper raw bids
bidseditor bids
bidscoiner raw bids\

Resulting file structure:
|_ bids directory
|__ sub-0X
|___ ses-0Y
|____ anat
|_____ sub-0X_ses-0Y_acq-MPRAGE_T1w.json
|_____ sub-0X_ses-0Y_acq-MPRAGE_T1w.nii.gz
|____ func
|_____ sub-0X_ses-0Y_task-motor_acq-MB6_bold.json
|_____ sub-0X_ses-0Y_task-motor_acq-MB6_bold.nii.gz
|____ dwi
|_____ sub-0X_ses-0Y_acq-MB3_dir-AP_dwi.bval
|_____ sub-0X_ses-0Y_acq-MB3_dir-AP_dwi.bvec
|_____ sub-0X_ses-0Y_acq-MB3_dir-AP_dwi.json
|_____ sub-0X_ses-0Y_acq-MB3_dir-AP_dwi.nii.gz
|_____ sub-0X_ses-0Y_acq-MB3_dir-AP_sbref.json
|_____ sub-0X_ses-0Y_acq-MB3_dir-AP_sbref.nii.gz
|____ fmap
|_____ sub-0X_ses-0Y_acq-MB3_dir-PA_epi.json
|_____ sub-0X_ses-0Y_acq-MB3_dir-PA_epi.nii.gz
|____ beh
|_____ sub-0X_ses-0Y_task-motor_acq-MB6_events.json
|_____ sub-0X_ses-0Y_task-motor_acq-MB6_events.tsv

Task-based functional MRI

Preprocessing

Example code for fMRIPREP:
module load fmriprep/23.0.2
cd /project/directory/bids
/opt/conda/bin/fmriprep /project/directory/bids /project/directory/bids/derivatives/fmriprep_v23.0.2/motor participant -w /scratch/marjoh/50961740.dccn-l029.dccn.nl/sub-ID01 --participant-label ID01 --skip-bids-validation --fs-license-file /opt_host/fmriprep/license.txt --mem_mb 37500 --omp-nthreads 4 --nthreads 4 --task-id motor --use-aroma --echo-idx 1 --ignore fieldmaps flair --output-spaces MNI152NLin6Asym --longitudinal --skip_bids_validation

First-level

Directory: /scripts/tb_fMRI/1st_level
|_ rewrite_fmriprep_confounds_aroma2.m correlates task regressors against ICA-AROMA noise components and relabels those that pass a certain threshold (5% explained variance) to non-noise.
|__ generate_task_regressors.m generates regressors from task events, defined according to extract_onsets_and_duration_pm.m. |_ motor_1stlevel.m carries out the first-level analysis.
|__ extract_onsets_and_duration_pm.m extract onsets and durations from task performance data, and generates parametric modulations of them.
|__ non_gm_covariates_fmriprep.m extracts confound timeseries from fMRIPREP output and inserts them as covariates in the first-level analysis.
motor_copycontrasts.m prepares first-level contrasts for group-level analysis. Left-sided responders are flipped horizontally so that the responding side is the same across participants. ZipOrUnzip.m utility function used to compress unused beta images from the 1st-level analysis

Regions-of-interest for group-level analyses

Directory: /ROIs
Johansson2024_BRAIN_ROIs.sh describes how the main region-of-interest Johansson2024_BRAIN_ROIs.nii was constructed from the results of Johansson et al. 2024.

Statistical analysis: Longitudinal mixed effects modelling (AFNI, 3dLME)

Directory: /scripts/tb_fMRI/group_level/group_comparisons
dataTable_3dLMEr_multiple.R and dataTable_3dLMEr_single.R will prepare tables to be used as input for 3dLME.
3dLME.sh carries out the mixed effects modelling.
3dLME_ClusterizeResid.sh generates tables that determine the cluster-level threshold of significance.
ExtractClusters.sh and ExtractClusters_runcode.sh identifies significant clusters.
ExtractBetas.m and ExtractBetas_runcode.m extracts values from significant clusters.
longitudinal_clinical_and_task-based_analysis.Rmd visualizes results in graphs.

Statistical analysis: Clinical correlations (FSL, randomise)

Directory: /scripts/tb_fMRI/group_level/clinical_correlations
motor_2ndlevel_Differencing.m takes the delta between visits (follow-up - baseline) for each first-level contrast of interest.
fsl_randomise_covars.R generates files needed to perform by-session and complete-case correlation analyses with clinical scores. Relies on retrieve_resphand.R to find the responding hand of participants.
fsl_add_regressors_to_covs.m adds indices of basal ganglia dysfunction and posterior substantia nigra free water to covariate files.
fsl_randomise_imgs.sh concatenates images and produces scripts for running randomise-parallel.
fsl_randomise_jobsub.sh submites randomise-parallel scripts to the cluster.
fsl_extract_stats.sh and fsl_extract_stats_loop.sh extracts stats and produces visualization-friendly variants of statistical images from randomise.
longitudinal_clinical_and_task-based_analysis.Rmd visualizes results in graphs.

Diffusion-weighted MRI

Preprocessing

QSIprep

Example code for QSIprep:
module load qsiprep/0.19.0
cd /project/directory/bids
/usr/local/miniconda/bin/qsiprep /project/directory/bids /project/directory/bids/derivatives participant --output-resolution 2 -w /scratch/marjoh/50614976.dccn-l029.dccn.nl/sub-0X --participant-label 0X --skip-bids-validation --fs-license-file /opt_host/fmriprep/license.txt --mem_mb 30000 --omp-nthreads 1 --nthreads 1 --skip_bids_validation --dwi-only --denoise-method dwidenoise --unringing-method mrdegibbs --prefer_dedicated_fmaps --pepolar-method TOPUP --eddy-config /project/directory/bids/derivatives/qsiprep/eddy_params.json --write-graph

FreeSurfer longitudinal pipeline

Directory: /scripts/freesurfer
s0_copy_T1.sh prepares T1w anatomicals for recon-all.
|_ fs_submitJobs.sh provides utility for submitting steps 1-3 above to a torque cluster. |__ s1_fs_cross_processSubject.sh performs recon-all on all T1w images.
|__ s2_fs_base_processSubject.sh creates base output that occupies the halfway space between baseline and follow-up measurements.
|__ s3_fs_long_processSubject.sh generates recon-all outputs that are unbiased with respect to time.\

Diffusion metrics

Directory: /scripts/dwi/dti
|_ qsimeasure.py is a utility function for generating metrics from QSIprep-processed DWI data.
|__ dipy_b0.py uses DIpy to generate b0 images.
|__ dipy_fw.py uses DIpy to generate a variety of tensors with DIpy and FSL implementations.
|__ amico_noddi.py generates NODDI metrics. Currently not used, but included here for convenience if it becomes relevant at a later stage.

Study-specific templates

Directory: /templates and /scripts/dwi
ANTs_generate_template.sh IN PROGRESS\

• FA: HCP1065_FA and FMRIB58_FA directories contain 1mm and 2mm resolution templates in MNI-space based on FA images of 50 healthy controls and 50 patients (both baseline and follow-up), created using antsMultivariateTemplateConstruction.sh. FSL-standard templates in HCP1065- and FMRIB58-space were used as targets.\
• MD: HCP1065_MD. Same as FA, except the target image in MNI space was an MD template.\
• b0: 50hc50pd_b0-avg. This template was formed by taking the average across baseline and follow-up b0 images from the same 50 healthy controls and 50 patients that were included in the FA and MD template construction.

Normalization of diffusion metrics

Directory: /scripts/dwi
normalize_sub.sh and normalize2.sh will normalize diffusion metrics according to the following steps: 1) create an unbiased T1w template, 2) boundary-based DWI>T1w transformation, 3) non-linear transformation from DWI>T1w>MNI. normalize2.sh will additionally project selected metrics onto the base (i.e. halfway) cortical surfaces generated by the FreeSurfer longitudinal pipeline.

Substantia nigra free water

Directory: /scripts/dwi
n2_ROIs_HCP1065_1mm.nii.gz is a region-of-interest that was drawn on the 50hc50pd_b0-avg template using FSLeyes, following conventions explained in Archer et al. 2019.
n2_ROIs_HCP1065_1mm_labels.txt details the coordinates and labels of the SN roi.
extract_metrics_sub.sh and extract_metrics.sh extracts free water values from the SN mask.

Statistical analysis

Directory: /scripts/longitudinal_structural_analysis.Rmd

• Substantia nigra free water
  1. Longitudinal mixed effects modelling and clinical correlations
  2. Clinical correlations
• Surface-based mean diffusivity
  1. Longitudinal mixed effects modelling and clinical correlations
  2. Clinical correlations
  3. plot_surface_md.R visualizes cortical MD