Datasets:
You need to agree to share your contact information to access this dataset
This repository is publicly accessible, but you have to accept the conditions to access its files and content.
This collection redistributes 10 public datasets, each under its own license: CC0 (Pixar, CNP, SOOP); CC BY 4.0 + the TCIA Data Usage Policy (UCSF-PDGM, UPENN-GBM, BraTS 2021); CC BY-SA 3.0 (IXI); CC BY-NC-SA 3.0 (ABIDE I, OpenBHB); CC BY-NC (ADHD-200). OpenBHB additionally asks users to accept the most restrictive data usage agreement of its source cohorts. Each dataset's README states its license, terms and required citations.
Log in or Sign Up to review the conditions and access this dataset content.
Brainmarks-sMRI
Ten public structural brain MRI datasets for evaluating sMRI foundation models, with classification, regression and segmentation targets. The images are the original releases, unmodified. Each dataset adds harmonized metadata tables and fixed train/val/test splits.
Datasets
| Dataset | Participants | Images | Targets | License | Size |
|---|---|---|---|---|---|
| ABIDE I | 1,112 | T1w | autism diagnosis, age, sex | CC BY-NC-SA 3.0 | 7.4 GB |
| ADHD-200 | 961 | T1w | ADHD diagnosis | CC BY-NC | 9.0 GB |
| BraTS 2021 | 1,477 | T1w, T1c, T2w, FLAIR, tumor mask | tumor segmentation, MGMT | CC BY 4.0 | 15.8 GB |
| CNP | 272 | T1w, DTI | psychiatric diagnosis (4-way) | CC0 | 13.4 GB |
| IXI | 584 | T1w, T2w, PD, DTI | age | CC BY-SA 3.0 | 17 GB |
| OpenBHB | 3,984 | T1w | age (with site debiasing) | CC BY-NC-SA 3.0 | 32 GB |
| Pixar | 155 | T1w | age | CC0 | 1.0 GB |
| SOOP | 1,715 | T1w, FLAIR, DWI, ADC, lesion mask | stroke lesion segmentation, discharge mRS, NIHSS | CC0 | 72 GB |
| UCSF-PDGM | 495 | T1w, T1c, T2w, FLAIR, DWI, ADC, tumor mask | IDH, MGMT, 1p/19q, grade, survival, tumor segmentation | CC BY 4.0 | 15.9 GB |
| UPENN-GBM | 630 | T1w, T1c, T2w, FLAIR, tumor mask | survival, IDH1, MGMT, tumor segmentation | CC BY 4.0 | 25.3 GB |
Each dataset folder's README.md has its source, version, license, citation, and split details.
Layout
<dataset>/
README.md # source, version, license, citation, splits
manifest.sha256 # checksums of source/; verify with `sha256sum -c manifest.sha256`
source/ # the original release, verbatim
tables/
images.tsv # one row per image: participant_id, session_id, modality, desc, path
samples.tsv # one row per scan session: participant_id, session_id, age, sex, site, targets...
samples.json # column descriptions, levels and units
splits.tsv # one row per participant: participant_id, split, official_split, rank, complete
- Paths in
images.tsvare relative to<dataset>/. - Splits are by participant. Official splits are kept where they exist; otherwise the split is 60/20/20, stratified, with a fixed seed.
completemarks participants with all core images and the primary targets.rankorders participants within each split so that the lowest-ranked N form a balanced subset; subsets are nested as N grows.
Usage
hf download medarc/brainmarks-smri --repo-type dataset --include "pixar/*" --local-dir brainmarks-smri
import pandas as pd
root = "brainmarks-smri/pixar"
images = pd.read_csv(f"{root}/tables/images.tsv", sep="\t")
samples = pd.read_csv(f"{root}/tables/samples.tsv", sep="\t")
splits = pd.read_csv(f"{root}/tables/splits.tsv", sep="\t")
# A balanced 50-participant training subset.
train = splits[(splits.split == "train") & splits.complete]
train_50 = train.nsmallest(50, "rank").participant_id
License and citation
Datasets and their derivatives are released under their original licenses. Non-commercial terms apply to ABIDE I, ADHD-200 and OpenBHB. If you use a dataset, use the citation given in the README and follow all dataset-specific acknowledgement conditions.
Reproducing
Every file here can be re-downloaded from its original source and checked against manifest.sha256. The download scripts and table-building code are at https://github.com/MedARC-AI/brainmarks-smri.
- Downloads last month
- 22