Skip to content

NSCLC-Radiomics (TCIA) as NIfTI: 422 lung-cancer CTs with organ and tumour segmentations

pl.load("hf_nsclc_radiomics") loads this dataset. The Lung1 cohort of NSCLC-Radiomics (Aerts et al.), converted from TCIA DICOM + DICOM-SEG to NIfTI by farrell236. Per patient one CT and the available structures: primary tumour (GTV-1), left and right lung, heart, oesophagus, spinal cord, plus a combined label map built by the repo's combine_segmentations.py (label values below come from that script).

import pardesline as pl

ds = pl.load("hf_nsclc_radiomics")
print(ds.describe())  # roles, frames, units, labels, licence, citation
s = ds[0]
obj = s.ct  # fetched now, cached under ~/.cache/pardesline/datasets/hf_nsclc_radiomics

Provenance and licence
Source Hugging Face Hub repo farrell236/NSCLC-Radiomics-NIFTI, paths NSCLC-Radiomics-NIFTI
Scheme hf
Size of the whole source 11.7 GB
Licence CC-BY-3.0
Commercial use allowed by the licence
Redistribution allowed
Licence note CC BY 3.0 on the Hub card and on the TCIA collection page; cite the data, the publication and TCIA (checked 2026-10-13).
Terms https://www.cancerimagingarchive.net/collection/nsclc-radiomics/
Homepage https://huggingface.co/datasets/farrell236/NSCLC-Radiomics-NIFTI
Cite Aerts H.J.W.L., Wee L., Rios Velazquez E., et al. Data From NSCLC-Radiomics (version 4). The Cancer Imaging Archive, 2019. doi:10.7937/K9/TCIA.2015.PF0M9REI. Aerts H.J.W.L. et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun 5:4006, 2014.
Samples grouped by id

Roles (what each sample contains)

Role Type Frame Units Labels / layout Pattern or file
ct volume (CT) from the file header from the file header NSCLC-Radiomics-NIFTI/{id}/image.nii.gz
Thorax CT (NIfTI, LPS mm from the header)
mask labels from the file header from the file header 1=esophagus, 2=gtv, 3=heart, 4=lung_left, 5=lung_right, 6=spinal_cord NSCLC-Radiomics-NIFTI/{id}/segmentation.nii.gz
Combined label map (values from combine_segmentations.py in the repo)
gtv labels from the file header from the file header 1=gtv NSCLC-Radiomics-NIFTI/{id}/seg-GTV-1.nii.gz
Primary gross tumour volume
lung_left labels from the file header from the file header 1=lung_left NSCLC-Radiomics-NIFTI/{id}/seg-Lung-Left.nii.gz
lung_right labels from the file header from the file header 1=lung_right NSCLC-Radiomics-NIFTI/{id}/seg-Lung-Right.nii.gz
heart labels from the file header from the file header 1=heart NSCLC-Radiomics-NIFTI/{id}/seg-Heart.nii.gz
esophagus labels from the file header from the file header 1=esophagus NSCLC-Radiomics-NIFTI/{id}/seg-Esophagus.nii.gz
spinal_cord labels from the file header from the file header 1=spinal_cord NSCLC-Radiomics-NIFTI/{id}/seg-Spinal-Cord.nii.gz

Standardized layout

Files are fetched on first access into the dataset cache and stay in their original format; ds.export(folder) writes every sample as standard files (PLY for points and meshes, MetaImage for volumes and label maps) with a .pardesline.json sidecar carrying frame, units, labels, licence and provenance, plus a manifest for the folder. Label maps become exact surfaces with to_mesh(label=k), DICOM series become one volume, raw arrays are decoded through the layout above.

ds = pl.load("hf_nsclc_radiomics").take(2)
ds.export("./standardized")        # PLY / MetaImage + sidecars + manifest.json

Algorithms and models that run on it

Generated from src/pardesline/data/manifests/hf_nsclc_radiomics.json on 2026-10-14.