data-test/ holds a real patient - the whole of TCIA 4D-Lung patient
P102 (about 980 MB in 1840 files), so the viewer can be exercised on
clinical data and not only on the synthetic phantom: a ten-phase 4DFBCT
with an RT Structure Set per phase, and the ten-phase 4DCBCT acquired from
the same patient nineteen days later. An installed copy of the program,
which has no source tree, fetches the same folder from GitHub with
Tools ▶ 📥 Download test data
(export-and-tools.md):
data-test/TCIA_4D-LUNG/P102/
4DFBCT+RTS/ study "4DFBCT+RTS"
1_CT_4DFBCT__Gated__0.0_A/ 133 CT slices, CT_0000.dcm ..
1_CT_4DFBCT__Gated__10.0_A/ CT_0132.dcm
... ten phases, 0 % .. 90 %
1_CT_4DFBCT__Gated__90.0_A/
RS_RTS__0.0_A.dcm one RTSTRUCT per phase,
... "RTS, 0.0%A" .. "RTS, 90.0%A"
RS_RTS__90.0_A.dcm
4DCBCT/ study "4DCBCT"
500_CT_4DCBCT__Gated__0.0_A/ 50 CT slices each
... ten phases, 0 % .. 90 %
509_CT_4DCBCT__Gated__90.0_A/
Both studies are one patient: PatientID 4D-LUNG_TCIA_P102, name P102,
StudyID 1, and no StudyDate on either (TCIA strips it; the slices keep
a Content Date).
| 4DFBCT | 4DCBCT | |
|---|---|---|
| Series | 10, all SeriesNumber 1 |
10, SeriesNumber 500 .. 509 |
| Slices per phase | 133 | 50 |
| Matrix / in-plane | 512 × 512, 0.9766 mm | 512 × 512, 0.8789 mm |
| Slice thickness | 3 mm (399 mm of coverage) | 3 mm (150 mm) |
| Description | 4DFBCT, Gated, <p>%A |
4DCBCT, Gated, <p>%A |
| Scanner | ADAC Pinnacle3 | Varian Trilogy Cone Beam CT, 125 kVp |
| Content date | 1998-03-25 | 1998-04-13 |
| Frame of Reference | one, shared by all ten phases | one per phase |
The percent in every series description is what the 4D detection keys on
(motion-4d.md), so each study appears as a single ten-phase
group - 4DCT - 4DFBCT, Gated, A (10 phases) and 4DCT - 4DCBCT, Gated, A (10 phases) - rather than as twenty loose series.
The structure sets carry twelve ROIs each - spinal cord, both lungs, heart,
esophagus, carina, a lymph node (LN), the tumor and four implanted gold
fiducial markers - plus a vertebra contour that exists only on phase 0, so
thirteen there. Every name carries its phase (Tumor_c00, Tumor_c50,
_c90 …), and each set references the 133 images of its own phase.
The 4DCBCT is worth its own note: the ten phases sit on an identical image
grid (same position, spacing and size) but each carries a different
Frame of Reference UID - phase 0 keeps TCIA's, phases 10 % .. 90 % have
2.25.… ones. That is exactly the case the Frame-of-Reference hints and
Transfer by relationship exist for (contours.md), and it
means a naive contour copy between CBCT phases is refused until a
relationship is given.
cargo run --release -- data-test/TCIA_4D-LUNG/P102/4DFBCT+RTS data-test/TCIA_4D-LUNG/P102/4DCBCT
loads the planning 4DFBCT into workspace A and the 4DCBCT into workspace B,
which is the inter-study, inter-modality case: different geometry, different
scanner, different Frame of Reference, nineteen days apart. Equivalently,
open data-test/ alone and both studies appear in one workspace, to be sent
on with right-click ▶ Copy series to workspace ….
Inside one study there is real respiratory motion to work with: the tumor
and the four markers move visibly between the phases, the 4D module plays
them as a loop, and the deformable methods of Image registration have
something anatomically real to recover. With an RTSTRUCT on every phase, the
structure propagation can be checked against a real contour rather than
against itself - propagate 0 % ▶ 50 % and compare with RS_RTS__50.0_A. It
is also the data the auto-segmentation was validated on
(auto-segmentation.md).
The data is patient P102 from the public 4D-Lung collection on The Cancer Imaging Archive (TCIA), a longitudinal 4D fan-beam CT / 4D cone-beam CT collection of 20 locally advanced NSCLC patients treated with chemoradiotherapy:
https://www.cancerimagingarchive.net/collection/4d-lung/
It is redistributed here under CC BY 3.0, the license of the original collection. If you use it, cite the data and the associated publications:
Data. Hugo, G. D., Weiss, E., Sleeman, W. C., Balik, S., Keall, P. J., Lu, J., & Williamson, J. F. (2016). Data from 4D Lung Imaging of NSCLC Patients (Version 2) [Data set]. The Cancer Imaging Archive. https://doi.org/10.7937/K9/TCIA.2016.ELN8YGLE
Publication. Hugo, G. D., Weiss, E., Sleeman, W. C., Balik, S., Keall, P. J., Lu, J., & Williamson, J. F. (2017). A longitudinal four-dimensional computed tomography and cone beam computed tomography dataset for image-guided radiation therapy research in lung cancer. Medical Physics, 44(2), 762-771. https://doi.org/10.1002/mp.12059
TCIA. Clark, K., Vendt, B., Smith, K., Freymann, J., Kirby, J., Koppel, P., Moore, S., Phillips, S., Maffitt, D., Pringle, M., Tarbox, L., & Prior, F. (2013). The Cancer Imaging Archive (TCIA): Maintaining and Operating a Public Information Repository. Journal of Digital Imaging, 26(6), 1045-1057. https://doi.org/10.1007/s10278-013-9622-6
The copy here is TCIA's, byte for byte: the collection is already
de-identified - patient P102 under the pseudonymous ID
4D-LUNG_TCIA_P102, no birth date, no institution or station name, study
dates removed - and nothing was rewritten on top of it, so the UIDs,
scanner tags and acquisition dates are the real ones a clinical archive
would hand over. That makes it a truthful test of the loader: original
UIDs of full length, a CBCT whose phases disagree about their Frame of
Reference, and SeriesNumber 1 repeated across all ten 4DFBCT phases.
If you need a scrubbed copy - of this or of anything else - the built-in anonymizer (Tools ▶ 🔏 Anonymize DICOM folder…, see export-and-tools.md) rewrites a folder to minimal, readable identifiers while keeping pixel data, geometry, ROI names, colors, types and contour points untouched, and every RTSTRUCT image reference still resolving.