Image & data credits

Last updated: July 31, 2026

The patient anatomy in the RT Image Matching Trainer — and in the screenshots and animations of it shown on this site — is derived from de-identified medical-imaging datasets that researchers and institutions have published under open Creative Commons Attribution (CC BY) licences. This page credits those sources, with our thanks.

Imaging datasets

Trainer cases are built from the following collections, accessed through The Cancer Imaging Archive (TCIA) / the NCI Imaging Data Commons unless noted:

• Prostate Anatomical Edge Cases Prostate fiducial CBCT case and 2D/2D prostate fiducial-match case (planning CT with the patient's real implanted gold markers); and the 2D/2D Hip Prosthesis / Pelvis case from O-MAR planning CT patient Prostate-AEC-111 (real unilateral total-hip arthroplasty). AP and patient-left lateral teaching DRRs were rendered from the CT; support hardware was removed from the projection volume and changes were made. The Cancer Imaging Archive — doi:10.7937/qstf-st65 · CC BY 4.0
• Pediatric-CT-SEG 2D/2D and CBCT Pediatric Abdomen cases from exam Pediatric-CT-SEG-E1FF3C7E. The near-isotropic CT was converted into AP and patient-left lateral educational DRRs and an abdomen/pelvis isotropic CBCT atlas. The CBCT case preserves selected expert Skin, liver, kidney, stomach, bladder, spinal-canal and bone contours as a non-cancer organ-registration exercise; no target volume was added. Changes were made. The Cancer Imaging Archive — doi:10.7937/TCIA.X0H0-1706 · CC BY 4.0
• LIDC-IDRI Lung SBRT CBCT case (synthetic lung nodule added), Breast CBCT case (synthetic lumpectomy plan, cavity, and clips added), and the 2D/2D Breast L mono-isocentric supraclavicular + medial-tangent case (reference field shapes baked into the adapted images). The Cancer Imaging Archive — doi:10.7937/K9/TCIA.2015.LO9QL9SX · CC BY 3.0
• 4D-Lung Lung SBRT Full Treatment case (one gated phase — 50% / end-exhale — of patient 100_HM10395's planning 4DCT, cropped and resampled; the physician RTSTRUCT tumour/lung/cord/heart contours were rasterized and a synthetic PTV of GTV + 5 mm was added); and the Breast L · Opposed Tangents 3D Conformal Full Treatment case from the arms-up 0% planning phase of patient 106_HM10395. For the breast case, the source RTSTRUCT lungs, heart, cord, and esophagus were used for geometric review; TotalSegmentator-derived breast anatomy was used to author an explicitly synthetic whole-breast teaching target; Plastimatch rendered setup and portal projections; and the patient-specific equal-and-opposite collimator rotations plus exact two-field jaw/MLC geometry were instantiated and read back in SlicerRT. No source breast treatment plan or dose was used. Changes were made. The Cancer Imaging Archive — doi:10.7937/K9/TCIA.2016.ELN8YGLE · CC BY 3.0
• NSCLC Radiogenomics 2D/2D Thorax case (AP + patient-left lateral planning DRRs generated from a chest CT). The Cancer Imaging Archive — doi:10.7937/K9/TCIA.2017.7HS46ERV · CC BY 3.0
• TCGA-PRAD (The Cancer Genome Atlas — Prostate Adenocarcinoma) 2D/2D Pelvis case (bony DRRs ray-summed from a diagnostic chest/abdomen/pelvis CT of patient TCGA-VP-A878). This collection also supplied MV-style pelvis and lumbar portal renders that remain in the case-data file but are not currently exposed as trainer cases; they are credited here because they are still distributed. The separate 2D/2D Lumbar Spine case is not derived from this collection — see Spine-Mets-CT-SEG below. Changes were made. The Cancer Imaging Archive — doi:10.7937/K9/TCIA.2016.YXOGLM4Y · CC BY 3.0
• Soft Tissue Sarcoma Soft-tissue sarcoma CBCT case (real tumour contour), 2D/2D femur case, and 2D/2D Shoulder / Scapula case from left-shoulder patient STS_008. The shoulder planning references were centred on the real GTV and converted into educational DRRs. The 2D/2D Knee / Tibia case uses whole-leg patient STS_047: the contralateral leg was removed at the measured inter-leg gap and the field was centred on the femoral condyles, with the released GTV_Mass ROI used only to anchor that search. Changes were made. The Cancer Imaging Archive — doi:10.7937/K9/TCIA.2015.7GO2GSKS · CC BY 3.0
• Spine-Mets-CT-SEG Spine Metastasis T4 2D/2D, CBCT and Full Treatment surfaces from patient 13683 use the same 0.5 mm radiotherapy-simulation CT and released T4 vertebral DICOM-SEG. Their start-screen cards have been retired, so those surfaces are reachable only through the rotating challenge; the assets remain in the case data and are credited here because they are still distributed. The collection classification identifies an osteolytic T4 level extending into the pedicle and pars; no lesion contour or patient treatment plan is released. Same-patient AP/lateral and beam-angle planning references were generated with Plastimatch. A 3 mm BODY-clipped expansion of the source T4 structure, carved outside a 2 mm cord-avoidance envelope, was created and read back in 3D Slicer. Three posterior static fields, jaws and a 120-leaf geometric MLC aperture with a separate 5 mm allowance were instantiated and read back in SlicerRT. That PTV, all field geometry, MU and delivery are explicitly synthetic educational simulation; no optimization, prescription, dose calculation or deliverable patient plan is represented. The separate Instrumented Spine 2D/2D and simulated first-day workflow use patient 11471 with a real three-level thoracic pedicle-screw and rod construct; their T6–T8-derived target and SlicerRT-read-back AP/PA plan are likewise synthetic. The separate Lumbar Spine 2D/2D case uses patient 10456, a 0.5 mm skin-to-skin radiotherapy-simulation CT whose released vertebral DICOM-SEG spans T8–L5; the L3 segment supplies the isocentre for AP/lateral Plastimatch references and no target volume, margin or plan is derived from it. Changes were made. The Cancer Imaging Archive — doi:10.7937/KH36-DS04 · CC BY 4.0
• TCGA-THCA (The Cancer Genome Atlas — Thyroid Cancer) Head & neck 2D/2D and CBCT cases from patient TCGA-DE-A4MA. DRRs were ray-summed from the head-and-neck CT. For the research-only CBCT teaching case, TotalSegmentator 2.16.0 generated a thyroid-gland mask and both lobes were reviewed in 3D Slicer 5.10. The same-study PET40 focus is used only as a gross-disease hypothesis (GTV, 6.0 cc). A simplified primary-site CTV (27.0 cc) combines the reviewed thyroid with a Slicer +5 mm local GTV expansion, then uses a Slicer +7/−7 mm morphological close to restore a connected isthmus region. One connected PTV (44.8 cc) encompasses both lobes with a Slicer +3 mm setup expansion. No nodal CTV is invented. These are literature-informed research teaching volumes—not released physician contours, a prescription, dose input, optimisation or deliverable plan. Changes were made. The Cancer Imaging Archive — doi:10.7937/K9/TCIA.2016.9ZFRVF1B · CC BY 3.0
• CPTAC-UCEC (Clinical Proteomic Tumor Analysis Consortium — Uterine Corpus Endometrial Carcinoma) Gynae / uterus CBCT case from contrast pelvic CT patient C3N-00872. The collection’s whole-uterus radiology annotation is preserved as the source registration target; it is not represented as a clinical GTV. Bladder and rectum are derived teaching contours, and the moving view applies conservative synthetic filling-driven pelvic deformation rather than patient-specific DIR. Two nested BODY-clipped target volumes were made and read back with the 3D Slicer 5.10 Segment Editor Margin effect as requested 5 mm + 5 mm steps. They are explicitly synthetic CTV/PTV teaching geometry, not clinician contours, dose inputs, optimization, or a deliverable plan. Changes were made. The Cancer Imaging Archive — doi:10.7937/K9/TCIA.2018.3R3JUISW · CC BY 4.0
• CC-Tumor-Heterogeneity Cervical Cancer CBCT-mode case from patient CCTH-A03's pre-treatment Timepoint1 acquisition. A source cervical-tumour annotation drawn on the collection's T2-weighted MRI was mapped to its paired pelvic CT through the released rigid DICOM Spatial Registration object. The CT was body-masked, cropped and resampled for the trainer; the moving cone-beam view is simulated from that CT rather than an independent daily scan. Two nested BODY-clipped geometric teaching contours were generated from the unchanged source annotation with the 3D Slicer 5.10 Segment Editor Margin effect as requested 5 mm + 5 mm steps. They are explicitly synthetic and are not represented as GTV, CTV, ITV, PTV, clinician-delineated treatment margins, dose inputs, optimization, or a plan. Changes were made. The Cancer Imaging Archive — doi:10.7937/ERZ5-QZ59 · CC BY 4.0
• Mediastinal-Lymph-Node-SEG Mediastinal Lymph Nodes CBCT-mode case from fully annotated contrast CT patient case_0412. All four components of the source manual lymph-node DICOM-SEG were retained while the CT was body-masked, cropped and resampled; the moving cone-beam view is simulated from that CT rather than an independent daily scan. For off-bone registration practice, the complete node set receives a conservative, longitudinal-dominant synthetic rigid displacement relative to the thoracic skeleton; this is not measured patient motion or a deformable respiratory model. Two nested BODY-clipped geometric teaching contours were generated from the unchanged diagnostic annotation with the 3D Slicer 5.10 Segment Editor Margin effect as requested 5 mm + 5 mm steps. The displacement and expansions are explicitly synthetic and are not represented as GTV, CTV, ITV, PTV, clinician-delineated treatment margins, dose inputs, optimization, or a plan. Changes were made. The Cancer Imaging Archive — doi:10.7937/QVAZ-JA09 · CC BY 4.0
• CPTAC-SAR (Clinical Proteomic Tumor Analysis Consortium — Sarcoma) CSI Full Treatment case from patient C3N-00875's continuous whole-body CT. Planning AP and patient-left lateral teaching DRRs were generated at Brain, Superior Spine, and Inferior Spine stations exactly 200 mm apart; the patient support was removed from the projection volume and the images were cropped, ray-summed, contrast-windowed, and resampled. All setup errors, couch parameters, 0.5/0.7 cm scenario margins, treatment sequence, and delivery are synthetic educational simulation only; no source patient plan, target, dose, or clinical protocol is represented. Changes were made. The Cancer Imaging Archive — doi:10.7937/TCIA.2019.9BT23R95 · CC BY 3.0
• Pancreatic-CT-CBCT-SEG (Hong J. et al.) Pancreas CBCT case (breath-hold planning CT with gastrointestinal organ contours). The Cancer Imaging Archive — doi:10.7937/TCIA.ESHQ-4D90 · CC BY 4.0
• Colorectal-Liver-Metastases Liver SBRT CBCT and teaching-workflow case (contrast CT with radiologist liver and tumour segmentations). The displayed CTV/PTV layers are synthetic teaching overlays, not source contours, clinical margins, dose, or a deliverable plan. The Cancer Imaging Archive — doi:10.7937/QXK2-QG03 · CC BY 4.0
• UPenn-GBM (Bakas S. et al.) Glioblastoma MR case (post-contrast T1 planning MR with radiologist-corrected tumour segmentation). The Cancer Imaging Archive — doi:10.7937/TCIA.709X-DN49 · CC BY 4.0
• Adrenal-ACC-Ki67-Seg Adrenal off-bone CBCT case from a contrast abdominal CT with a radiologist-refined adrenal-mass DICOM-SEG. The source mass is preserved while the moving view gives it a conservative synthetic SI-dominant rigid displacement relative to bone; this is not measured patient motion or a deformable respiratory model. Two nested BODY-clipped geometric teaching contours were generated from the unchanged source annotation with the 3D Slicer 5.10 Segment Editor Margin effect as requested 5 mm + 5 mm steps. The displacement and expansions are explicitly synthetic and are not represented as GTV, CTV, ITV, PTV, clinician-delineated treatment margins, dose inputs, optimization, or a plan. Changes were made. The Cancer Imaging Archive — doi:10.7937/1FPG-VM46 · CC BY 4.0
• CPTAC-CCRCC + CPTAC-CCRCC-Tumor-Annotations Renal Cell Carcinoma CBCT case from patient C3N-03018. The contrast CT was cropped and resampled into an isotropic teaching atlas; the radiologist-reviewed right-kidney tumour annotation is preserved as a source lesion outline. It is not represented as a clinical GTV, CTV or PTV, and no treatment margin was inferred. Changes were made. Source CT — The Cancer Imaging Archive — doi:10.7937/K9/TCIA.2018.OBLAMN27 · CC BY 3.0 Tumour annotation — The Cancer Imaging Archive — doi:10.7937/SKQ4-QX48 · CC BY 4.0
• EAY131 + EAY131-Tumor-Annotations v2 Esophageal Cancer CBCT case from patient EAY131-8537265. The thoracic CT was gap-filled from its released non-uniform slice positions, cropped and resampled into an isotropic teaching atlas; the expert esophageal-primary annotation is preserved as a source lesion outline. The 3D Slicer 5.10 Segment Editor Margin effect added requested +5 mm and +10 mm-total BODY-clipped synthetic teaching contours with SlicerRT verified. The outer contour is shown in the trainer’s synthetic PTV row, but neither expansion is a clinician-delineated GTV, CTV, ITV or PTV, clinical margin, dose input, optimization result, or treatment plan. Changes were made. Source CT — The Cancer Imaging Archive — doi:10.7937/C5KE-YX42 · CC BY 4.0 Tumour annotation — The Cancer Imaging Archive — doi:10.7937/Q9RN-M510 · CC BY 4.0
• A Paired Head CT-MRI Dataset for Cross-Modality Image Synthesis 2D/2D Brain and Acoustic Neuroma SRS cases use the same patient sub-19: the Brain DRRs come from its 0.5 mm bone-kernel CT, while Acoustic SRS adds the exactly co-gridded CE-T1W and CT-BRAIN. The real paired CT supplies CBCT/kV anatomy and CT-derived BODY while MRI localises a direct synthetic teaching PTV at the patient-right IAC/CPA. The source has no RTSTRUCT/SEG; no source GTV, CTV, cochlea contour, clinical margin, dose, or patient treatment plan is claimed. Volumes were de-identified by the source, cropped in-plane, resampled, display-normalised, projected, and converted into educational imaging; changes were made. Zenodo — doi:10.5281/zenodo.17486320 · CC BY 4.0

The Cancer Imaging Archive

TCIA-hosted collections above are made available thanks to the archive itself: Clark K, Vendt B, Smith K, et al. The Cancer Imaging Archive (TCIA): Maintaining and Operating a Public Information Repository. Journal of Digital Imaging. 2013;26(6):1045–1057. doi:10.1007/s10278-013-9622-7. Data were accessed via the NCI Imaging Data Commons.

Tools

Some organ contours (e.g. heart, ribs, and vertebral structures) were generated with TotalSegmentator (Apache-2.0): Wasserthal J, et al. TotalSegmentator: Robust Segmentation of 104 Anatomic Structures in CT Images. Radiology: Artificial Intelligence. 2023. doi:10.1148/ryai.230024.

The simulator's optional patient figure was built from a MakeHuman base mesh via the MPFB Blender extension. MakeHuman's generated mesh output is released under CC0 1.0; it was re-posed, re-topologised at the eyes, split into bands, and calibrated onto the trainer's landmark anchors — changes were made. The patient gown uses the "Fabric Pattern 07" normal map from Poly Haven (CC0 1.0), downscaled and converted to WebP. Neither CC0 source requires attribution; both are credited here for completeness.

The site is set in Inter (© The Inter Project Authors) and JetBrains Mono (© The JetBrains Mono Project Authors), both self-hosted under the SIL Open Font License 1.1; the licence text is served alongside the font files at /assets/fonts/LICENSE.txt.

Machine simulation & trademarks

The interactive treatment-machine model, console graphics, and animations were independently authored for this educational simulator. They use approximate geometry derived from publicly available dimensional literature and general treatment-machine features; no manufacturer CAD, software, screenshots, logos, or product decals are included.

This independent project is not affiliated with, sponsored by, or endorsed by any equipment manufacturer. Product names and trademarks belong to their respective owners and are referenced only where needed for factual attribution.

Licence notes

The CC BY licences above apply to the source imaging datasets, which remain available from their publishers at the links given. Attribution here does not imply that the dataset authors, TCIA, the NCI, Zenodo, or any equipment manufacturer endorses this site. The trainer's original software, design, and content are © Craig Utter; licensed and third-party materials remain subject to their respective terms — see our Terms of Service.

Contact

Questions about a credit or a source: support@rtimagematch.com.