QIN-BREAST
QIN-BREAST (Quantitative Imaging Network breast cancer neoadjuvant treatment response PET/CT and MRI)
Longitudinal prone FDG PET/CT and quantitative 3 T breast MRI (DWI, DCE, multi-flip-angle T1 mapping) of 68 breast cancer patients from Vanderbilt, scanned before, during and after neoadjuvant therapy, with pathological response labels.
Overview
QIN-BREAST is a longitudinal imaging collection from Vanderbilt University, contributed by the group of Thomas E. Yankeelov within the NCI Quantitative Imaging Network and published by The Cancer Imaging Archive. It follows patients with breast cancer through neoadjuvant therapy, so that quantitative PET and MRI measures taken early in treatment can be developed and tested as predictors of response. A second collection, QIN-BREAST-02, extends the protocol to a multi-center study and is listed separately on TCIA.
Composition
The Data Access table lists 68 subjects, 216 studies, 536 DICOM series and 102,451 images (11.41 GB), while the summary line at the top of the page gives 67 subjects. In the open metadata digest, 43 subjects have FDG PET/CT and 41 have MRI. Imaging was planned at three time points: before therapy, after the first cycle, and either after the second cycle or at the end of therapy before surgery. A separate spreadsheet gives a pathological response label (pCR or non-pCR) for 59 patient IDs: 20 pCR and 39 non-pCR.
Acquisition
PET/CT was performed on a GE Discovery STE about 60 minutes after weight-scaled FDG injection, with a low-dose CT for attenuation correction. Research images were taken prone on a support device built in-house, which eases registration to MRI; standard-of-care supine whole-body images were added at the first and last time points. MRI used a Philips 3 T Achieva with a 16-channel bilateral breast coil. Each MRI session holds diffusion-weighted EPI in three directions (b values varied between patients), a multi-flip-angle 3D gradient echo series (2 to 20 degrees) for T1 mapping, and a 25-frame DCE series with gadopentetate dimeglumine injected after the third frame.
Annotations
The collection itself has no segmentations. Its labels are the per-patient pathological response classes. The page links to tumor segmentations produced by a third party under the AIMI annotations initiative, published separately on Zenodo.
Known limitations
- Single site, and not every subject has both PET/CT and MRI.
- Diffusion parameters changed during the study, as described by Li et al. (2015).
- The number of subjects differs by one between the two tables on the collection page.
- Age and sex are missing for some subjects in the DICOM headers, and no clinical table beyond the response label is provided.
Cohort
Aggregate numbers from the sources below. Bars are relative to the 68 subjects.
Sex
- Female 53 100%
Covers 53 of 68 subjects.
Contrast combinations
How many subjects have exactly each set of contrasts.
| T1map | dwi | dce | Subjects with exactly this set |
|---|---|---|---|
41 |
Modality
subjects
- CT 43 63%
- PET 43 63%
- MRI 41 60%
Contrast / sequence
subjects, values can overlap
- Diffusion-weighted 41 60%
- DCE perfusion 41 60%
- T1 map 41 60%
PET tracer
subjects
- [18F]FDG 43 63%
Scanner vendor
subjects
- GE HealthCare 43 63%
- Philips 41 60%
License and access
Our reading of the license, not legal advice. Before you use the data, read the original license and confirm that your use is allowed. We take no responsibility for how you use a dataset. Full disclaimer
Download without an account
DICOM images are downloaded with the TCIA Data Retriever from a manifest file and are also available in the NCI Imaging Data Commons. The treatment response spreadsheet is a direct download. No account is needed.
Creative Commons Attribution 3.0 Unported
Older version of CC BY. Use, share and adapt the data for any purpose, including commercial use, as long as you credit the creators.
The Data Access table lists CC BY 3.0 for both the images and the treatment response file. The page also requires the data citation under the TCIA Data Usage Policy.
What you can do
- Yes
- Yes
- Yes
- Yes
What you can share
- Yes
- Yes
- Yes
What you must do
- Yes
- Share alike No
- No
- No
- Manuscript review No
- Release code No
- Return results No
- Delete after use No
Limits
- No
- Location limits No
Citation
Li, X., Abramson, R. G., Arlinghaus, L. R., Chakravarthy, A. B., Abramson, V. G., Sanders, M., & Yankeelov, T. E. (2016). Data From QIN-BREAST (Version 2) [Data set]. The Cancer Imaging Archive. https://doi.org/10.7937/K9/TCIA.2016.21JUEBH0
All numbers
Every number on this page, as stored in stats.csv, with its source.
| Measure | Breakdown | Value | Source |
|---|---|---|---|
| Subjects | total the summary table at the top of the page lists 67 subjects; the metadata digest has 68 Patient IDs | 68 | tcia-qin-breast Data Access (Version 2) |
| Studies | total | 216 | tcia-qin-breast Data Access (Version 2) |
| Scans | total DICOM series | 536 | tcia-qin-breast Data Access (Version 2) |
| Images | total DICOM images | 102,451 | tcia-qin-breast Data Access (Version 2) |
| Subjects | modality=CT low-dose CT for PET attenuation correction | 43 | tcia-qin-breast-digest Patient ID, Modality columns |
| Subjects | modality=PT | 43 | tcia-qin-breast-digest Patient ID, Modality columns |
| Subjects | modality=MR | 41 | tcia-qin-breast-digest Patient ID, Modality columns |
| Subjects | tracer=fdg all PET series; FDG named on the collection page | 43 | tcia-qin-breast-digest Patient ID, Modality columns |
| Scans | modality=CT | 110 | tcia-qin-breast-digest Series Instance UID, Modality columns |
| Scans | modality=PT | 110 | tcia-qin-breast-digest Series Instance UID, Modality columns |
| Scans | modality=MR | 316 | tcia-qin-breast-digest Series Instance UID, Modality columns |
| Subjects | contrast=dwi series descriptions containing DWI | 41 | tcia-qin-breast-digest Patient ID, Series Description columns |
| Subjects | contrast=dce series descriptions containing dynamic | 41 | tcia-qin-breast-digest Patient ID, Series Description columns |
| Subjects | contrast=T1map multi-flip-angle series for T1 mapping (series descriptions containing T1-map); no computed maps | 41 | tcia-qin-breast-digest Patient ID, Series Description columns |
| Subjects | contrast_set=dce+dwi+T1map every MR subject has all three | 41 | tcia-qin-breast-digest Patient ID, Series Description columns |
| Subjects | vendor=ge GE Discovery STE PET/CT | 43 | tcia-qin-breast-digest Patient ID, Manufacturer columns |
| Subjects | vendor=philips Philips Achieva MRI | 41 | tcia-qin-breast-digest Patient ID, Manufacturer columns |
| Subjects | sex=female 13 subjects have no sex value and 2 are recorded as O | 53 | tcia-qin-breast-digest Patient ID, Patient Sex columns |
Sources
The keys used in the table above.
- tcia-qin-breast TCIA QIN-BREAST collection page (version 2, updated 2020/01/10) website
- tcia-qin-breast-digest TCIA QIN-BREAST radiology metadata digest (QIN-BREAST_2015-09-04-nbia-digest.xlsx, open download, CC BY 3.0), counted per unique Patient ID computed
- li2015 Li et al. 2015, Investigative Radiology (PMC4471951) paper