PAM50 normalized metrics

This repository contains morphometric measures in the PAM50 anatomical dimensions computed from adults (healthy controls and patients) and children across multiple datasets.

👉 If you find this work and/or code useful for your research, please cite our paper:

Valošek, Bédard et al., (2024). A database of the healthy human spinal cord morphometry in the PAM50 template space. Imaging Neuroscience, 2 1–15, https://doi.org/10.1162/imag_a_00075

👉 For interactive figures, please visit the NeuroLibre preprint:

Valošek, Bédard et al., (2023). A database of the healthy human spinal cord morphometry in the PAM50 template space. NeuroLibre Reproducible Preprints, 17, https://doi.org/10.55458/neurolibre.00017

Usage

As part of SCT

The repository is downloaded automatically during the SCT installation.

CSV files from this repository are used by the following SCT functions:

• sct_process_segmentation -normalize-PAM50 1, example:

sct_process_segmentation -i sub-001_T2w_label-SC_seg.nii.gz -vertfile sub-001_T2w_label-SC_seg_labeled.nii.gz -perslice 1 -normalize-PAM50 1 -o sub-001_T2w_metrics_PAM50.csv

• sct_compute_compression -normalize-hc 1, example:

sct_compute_compression -i sub-001_T2w_label-SC_seg.nii.gz -vertfile sub-001_T2w_label-SC_seg_labeled.nii.gz -l sub-001_T2w_label-compression.nii.gz -normalize-hc 1 -o sub-001_T2w_compression_metrics.csv

Standalone usage

If you want to use the morphometric measures outside of SCT in your own research, you can simply download this repository and use the CSV files directly:

git clone https://github.com/spinalcordtoolbox/PAM50-normalized-metrics.git

Datasets Overview

The table below summarizes all datasets included in this repository. It is also available as datasets.tsv and can be regenerated by running python code/generate_datasets_summary.py.

metric	name	num_subjects	num_sessions	num_sites	population	sex (M/F/unknown)	age (mean±SD [min–max])	coverage	contrast	resolution	link
spinal_canal	spine-generic multi-subject	203	203	43	healthy adults	105/98/0	28.7±5.6 [19–52]	cervical spine	T2w	0.8mm iso	spine-generic/data-multi-subject
spinal_cord	spine-generic multi-subject	203	203	43	healthy adults	105/98/0	28.7±5.6 [19–52]	cervical spine	T2w	0.8mm iso	spine-generic/data-multi-subject
spinal_cord	whole-spine	58	58	2	healthy adults	30/18/10	27.2±6.7 [21–56]	whole spine	T2w	1.0mm iso	openneuro/ds005616
spinal_cord	MASiVar	96	144	n/a	healthy adults and children	55/41/0	9.6±8.7 [5–47]	cervical spine	T1w	1.0mm iso	openneuro/ds003416

Datasets Structure

PAM50-normalized-metrics/
├── datasets.tsv                          # master dataset overview (auto-generated)
├── spinal_canal/                         # spinal canal morphometric measures
│   └── spine-generic_multi-subject/
│       ├── dataset_description.json      # dataset metadata (coverage, contrast, resolution, ...)
│       ├── participants.tsv              # subject metadata
│       ├── sub-amu01_T2w_PAM50.csv
│       └── ...
└── spinal_cord/                          # spinal cord morphometric measures
    ├── spine-generic_multi-subject/
    │   ├── dataset_description.json
    │   ├── participants.tsv
    │   ├── sub-amu01_T2w_PAM50.csv
    │   └── ...
    └── whole-spine/
        ├── dataset_description.json
        ├── participants.tsv
        ├── sub-amuAL_T2w_PAM50.csv
        └── ...

Each CSV file contains per-slice morphometric measures for one subject, normalized to the PAM50 template space.

Adding a New Dataset

1. Create a new dataset folder — under the appropriate metric directory (e.g., spinal_cord/<new-dataset>/), add the per-subject CSV files, a participants.tsv with subject metadata, and a dataset_description.json with the dataset metadata:

   spinal_cord/
   └── <new-dataset>/
       ├── dataset_description.json  # dataset metadata
       ├── participants.tsv          # subject metadata (participant_id, sex, age, institution, ...)
       ├── sub-001_T2w_PAM50.csv
       └── ...
   

   The dataset_description.json should contain:

   {
       "name": "<human-readable dataset name>",
       "coverage": "<anatomical coverage>",
       "contrast": "<MRI contrast>",
       "resolution": "<voxel size>",
       "population": "<subject population>",
       "link": "<URL to raw dataset>",
       "link_text": "<display text for the link>"
   }

2. Regenerate datasets.tsv and update README — run the script from the repository root; it will auto-compute subject counts, sex, and age statistics from participants.tsv, fetch the dataset metadata from dataset_description.json, and automatically update the summary table in the Datasets Overview section above:

   python code/generate_datasets_summary.py