TCIA HNSCC Collection (MD Anderson)
Overview
The TCIA HNSCC collection is a clinical imaging archive of 215 head and neck squamous cell carcinoma (HNSC) patients treated with curative-intent radiotherapy at MD Anderson Cancer Center (October 2003 – August 2013). It includes pre- and post-treatment PET-CT and RT simulation CT scans, RT planning DICOMs (RTPLAN, RTSTRUCT, RTDOSE), and a clinical spreadsheet with body composition measurements, HPV status, staging, and survival outcomes. Published by Grossberg et al. 2018 as TCIA collection DOI 10.7937/K9/TCIA.2017.umz8dv6s (433,384 DICOM files across 3,225 series). PMID:30179230
Composition
- Cancer type: HNSC, n=215 (screened from 2,840 consecutive patients).
- Subsite: 73% oropharyngeal primaries (base of tongue 51%, tonsil 43%); 85.5% male.
- Treatment: 59% (127/215) received concurrent chemotherapy, 98% of whom received platinum-based regimens (cisplatin); 13% postoperative RT. Mean RT dose 68.66 Gy (28–40 fractions).
- Imaging: pre-treatment whole-body PET-CT in 212/215 (98.6%); follow-up PET-CT in 213/215 (99.1%); RT simulation CT for all. Multiple scanner makes (GE, Philips, Picker, Marconi).
- HPV status: assessed by in situ hybridisation for high-risk HPV subtypes; AJCC 7th edition TNM staging.
- Body composition: pre- and post-RT skeletal muscle and adipose cross-sectional area at L3 vertebra measured in Pinnacle v9.6 using HU thresholds; lumbar skeletal muscle index (SMI) and adipose index (ADI) normalised to height².
- Scale: 433,384 DICOM files, 3,225 series, 765 studies. Each patient contributes pre-treatment CT/PET-CT, simulation CT, RTPLAN, RTSTRUCT, RTDOSE, and post-treatment imaging. PMID:30179230
Assays / panels (linked)
- PET-CT imaging — FDG PET-CT pre- and post-treatment.
- CT imaging — RT simulation CT and body composition analysis at L3.
- DICOM RT planning — RTPLAN, RTSTRUCT, RTDOSE.
Papers using this cohort
- PMID:30179230 — Grossberg et al. 2018, Scientific Data: primary dataset descriptor; 215 HNSCC patients with multi-timepoint PET-CT, RT planning DICOMs, and body composition annotations at L3.
- PMID:37397861 — Kim et al. 2023, Radiotherapy and Oncology: MDACC subset (n=627 oropharynx) used as external validation; validated externally by Dana-Farber Cancer Institute collaborators using pretrained challenge models.
Notable findings derived from this cohort
- Unique body composition annotations: pre- and post-RT skeletal muscle (HU −29 to 150) and adipose (HU −190 to −30) CSA at L3, with skeletal-muscle depletion defined at SMI <52.4 cm²/m² (men) / <38.5 cm²/m² (women). Total lean body mass and fat mass derived via published formulae. PMID:30179230
- Technical validation required 2–6 curation revisions per subject (mean 3.17) to resolve DICOM inconsistencies; 90.7% of subjects had series/study inconsistency issues on initial intake. PMID:30179230
- On the MDACC subset, only the top combined model (MTLR + EMR + volume) beat tumor volume alone, and by a small margin; external generalizability was limited compared with the RADCURE internal test set. PMID:37397861
Notes
- Kim et al. 2023 (PMID:37397861) uses a 627-patient oropharynx subset curated from the same TCIA HNSCC collection as Grossberg 2018; exact subset provenance is not reconciled in the paper. The canonical collection size (n=215) reflects the Grossberg 2018 descriptor.
Sources
- TCIA collection: HNSCC — TCIA DOI 10.7937/K9/TCIA.2017.umz8dv6s
- PMID:30179230 — Grossberg et al. 2018, Scientific Data, DOI 10.1038/sdata.2018.173.
- PMID:37397861 — Kim et al. 2023, Radiotherapy and Oncology.
This page was processed by entity-page-writer on 2026-04-15.