The Cancer Imaging Archive (TCIA) Collections
The Cancer Imaging Archive (TCIA) hosts de-identified cancer imaging datasets, organized by disease, modality, or research focus. Most collections share imaging in DICOM, sometimes accompanied by clinical, genomic, and expert annotation.
This repository provides curated, analysis-ready NIfTI derivatives for selected TCIA brain tumor collections.
Where possible, diffusion MRI derivatives (SZ / GQI / QSDR) are provided in formats compatible with DSI Studio.
Raw TCIA DICOM archives remain on TCIA servers; only derived files are mirrored here.
UPenn-GBM — Advanced MRI, Clinical, Genomics & Radiomics
Release tag: upenn-gbm
Source: University of Pennsylvania Glioblastoma (UPenn-GBM) TCIA Collection
This collection provides multi-parametric MRI, clinical outcomes, tumor segmentations, and genomic features for de novo glioblastoma (GBM).
Included data
- Imaging
- T1w, T2w, FLAIR, contrast-enhanced T1w
- Skull-stripped, co-registered NIfTI volumes
- Segmentations
- Tumor subregions (expert-corrected): enhancing, non-enhancing, edema
- Non-imaging (via TCIA)
- Clinical outcomes
- Genomics
- Radiomics features
- Applications
- Tumor segmentation benchmarking
- Radiogenomics
- Progression modeling and survival prediction
License
CC BY 4.0
Citation
Bakas S. et al., Scientific Data (2022) 9:453
The University of Pennsylvania glioblastoma (UPenn-GBM) cohort: Advanced MRI, clinical, genomics, & radiomics.
Linux / macOS — bash download
curl -s https://api.github.com/repos/data-nih/tcia/releases/tags/upenn-gbm | jq -r '.assets[].browser_download_url' | xargs -n1 -P4 curl -LO
Windows PowerShell 5.x download
(Invoke-RestMethod "https://api.github.com/repos/data-nih/tcia/releases/tags/upenn-gbm").assets | ForEach-Object { Invoke-WebRequest $_.browser_download_url -OutFile (Split-Path $_.browser_download_url -Leaf) }
UCSF-PDGM — Preoperative Diffuse Glioma MRI
Release tag: ucsf-pdgm
Source: UCSF Preoperative Diffuse Glioma MRI (TCIA)
This dataset contains preoperative MRI for 501 adult patients with WHO grade II–IV diffuse gliomas, including advanced diffusion and perfusion imaging.
Included data
-
Cohort
- 501 subjects
- WHO grade II–IV
- Genomic markers: IDH, MGMT, 1p/19q
-
Imaging
- 3D T2, 3D T2/FLAIR
- SWI
- DWI + 55-direction HARDI
- Pre/post-contrast T1w
- 3D ASL perfusion
-
Segmentations
- Enhancing tumor
- Non-enhancing / necrotic tumor
- FLAIR hyperintensity (edema)
- Radiologist-refined BraTS-ensemble AI segmentations
-
Applications
- AI segmentation
- Radiogenomics (IDH, MGMT prediction)
- Comparative glioma research (vs. BraTS, TCGA-GBM, UPenn-GBM)
License
CC BY 4.0
Citation
Calabrese, E. et al. (2022). The University of California San Francisco Preoperative Diffuse Glioma MRI (UCSF-PDGM). TCIA. DOI: 10.7937/tcia.bdgf-8v37
Linux / macOS — bash download
curl -s https://api.github.com/repos/data-nih/tcia/releases/tags/ucsf-pdgm | jq -r '.assets[].browser_download_url' | xargs -n1 -P4 curl -LO
Windows PowerShell 5.x download
(Invoke-RestMethod "https://api.github.com/repos/data-nih/tcia/releases/tags/ucsf-pdgm").assets | ForEach-Object { Invoke-WebRequest $_.browser_download_url -OutFile (Split-Path $_.browser_download_url -Leaf) }
Licensing and Use
- All derivatives follow the original TCIA collection licenses (mostly CC BY 4.0).
-
Users must comply with:
- TCIA data usage and citation requirements
- Citations of the primary publications for each dataset
- When using curated derivatives from this repository, please add:
“Derived imaging files were curated via the Pittsburgh Fiber Data Hub (
data-nih/tcia) from TCIA collections.”
Notes
- This repository does not include the original DICOM archives (hosted on TCIA).
- NIfTI and diffusion derivatives are provided for convenience and downstream analysis (tractography, radiomics, segmentation research).
- Users should perform their own QC and methodological validation before scientific reporting.