4D-Lung (segmentation subset)

Longitudinal 4D (respiratory-gated, phase-resolved) fan-beam CT of 20 locally-advanced non-small-cell lung cancer (NSCLC) patients, with expert manual RTSTRUCT contours, from Data from 4D Lung Imaging of NSCLC Patients (4D-Lung) on The Cancer Imaging Archive (Hugo et al., VCU).

This is the segmentable subset of the full collection — read carefully. The full TCIA 4D-Lung collection is 183 GB and contains both 4D fan-beam CT (4D-FBCT, "4DCT") and 4D cone-beam CT (4D-CBCT). Only the 4DCT series carry public RTSTRUCT segmentations; the 4DCBCT series (134 GB) have no public masks. This mirror therefore contains only the 820 4DCT phase series + the 800 RTSTRUCT series (~46 GB) that form usable image/mask pairs. If you need the un-annotated 4DCBCT, download it from TCIA directly.

Dataset Details

Field	Value
Modality	CT (4D fan-beam, respiratory-gated; 10 breathing phases 0–90%) + RTSTRUCT contours
Body part	Lung / thorax — locally-advanced NSCLC
Task	3D tumor-target segmentation (GTV + involved nodes)
Patients	20 (100_HM10395 … 119_HM10395)
CT series (phase volumes)	820
RTSTRUCT series	800
Image/mask pairs	800 (each RTSTRUCT references exactly one CT phase volume)
DICOM files	93,830
Acquisition	longitudinal — each patient imaged repeatedly across the chemo-radiotherapy course
Format	DICOM (CT + RTSTRUCT)
License	CC BY 3.0

Segmentation Labels & Recommended Ground Truth

RTSTRUCT ROI names follow the convention {Structure}_c{PP}, where PP is the breathing-phase percentage (c00 = 0% … c90 = 90%). Each RTSTRUCT is phase-specific and references the matching phase CT.

Recommended gold-standard ground truth: the radiotherapy target volume —

Label	ROI names	Coverage
GTV (gross primary tumor)	Tumor_cPP	all 800 pairs (100%)
Involved lymph nodes	LN_cPP, LN2_cPP	720 / 800 pairs (90%)

All contours were delineated by a single experienced radiation oncologist (E. Weiss) under physician supervision — there is one annotation tier (no multi-rater / partial-vs-full split).

The RTSTRUCT files also contain, with partial coverage, organs-at-risk (RLung, LLung, Esophagus, Heart, Cord, Trachea — present on only ~101 RTSTRUCT) and landmarks/fiducials (Carina, Vertebra, MarkerA–D; markers in 7 patients only). These are NOT the gold target but are preserved intact in the raw RTSTRUCT for downstream use. A handful of ROI names have minor typos (Tumor_ c00, LN2_C10).

Cross-dataset Overlap

None known. 4D-Lung is a single-institution VCU cohort (all subjects share the HM10395 site suffix). It is disjoint from NSCLC-Radiomics (Maastricht Lung1), LIDC-IDRI, and the Medical Segmentation Decathlon Lung task — no shared patients or source archives are documented. The only identifier is the TCIA Subject ID (NNN_HM10395).

Structure

<PatientID>/<StudyInstanceUID>/CT_<SeriesInstanceUID>/*.dcm        # CT phase volume
<PatientID>/<StudyInstanceUID>/RTSTRUCT_<SeriesInstanceUID>/*.dcm  # contours
series_index.json                                                  # index + RT→CT pairing
LICENSE.txt

Each RTSTRUCT references its source CT phase series via ReferencedFrameOfReferenceSequence / RTReferencedSeriesSequence. The flat layout encodes the DICOM Modality as the series-folder prefix (CT_… / RTSTRUCT_…). series_index.json provides, for every series, its patient / study / modality / relative path / breathing phase, and for every RTSTRUCT its resolved ref_ct_uid, ROI names, and GTV/node/OAR coverage flags — plus a ready-made pairs list of the 800 RTSTRUCT→CT pairings.

Source & Citation

• TCIA collection: https://www.cancerimagingarchive.net/collection/4d-lung/
• Data DOI: 10.7937/K9/TCIA.2016.ELN8YGLE
• Official, author-deposited (VCU); fully public, CC BY 3.0, no registration.

@article{hugo2017longitudinal4dlung,
  author  = {Hugo, Geoffrey D. and Weiss, Elisabeth and Sleeman, William C. and
             Balik, Salim and Keall, Paul J. and Lu, Jun and Williamson, Jeffrey F.},
  title   = {A longitudinal four-dimensional computed tomography and cone beam
             computed tomography dataset for image-guided radiation therapy
             research in lung cancer},
  journal = {Medical Physics},
  volume  = {44},
  number  = {2},
  pages   = {762--771},
  year    = {2017},
  doi     = {10.1002/mp.12059}
}

@misc{hugo20164dlungtcia,
  author    = {Hugo, G. D. and Weiss, E. and Sleeman, W. C. and Balik, S. and
               Keall, P. J. and Lu, J. and Williamson, J. F.},
  title     = {Data from 4D Lung Imaging of NSCLC Patients (4D-Lung) [Data set]},
  year      = {2016},
  publisher = {The Cancer Imaging Archive},
  doi       = {10.7937/K9/TCIA.2016.ELN8YGLE}
}

@article{clark2013tcia,
  author  = {Clark, Kenneth and Vendt, Bruce and Smith, Kirk and others},
  title   = {The Cancer Imaging Archive (TCIA): Maintaining and Operating a
             Public Information Repository},
  journal = {Journal of Digital Imaging},
  volume  = {26},
  number  = {6},
  pages   = {1045--1057},
  year    = {2013},
  doi     = {10.1007/s10278-013-9622-7}
}