COVID-19-NY-SBU
Stony Brook University COVID-19 Positive Cases
Chest radiographs and CT of PCR-positive COVID-19 patients linked to clinical outcomes such as death, ICU stay and ventilation, used to build diagnostic and prognostic models. 1,384 patients from Stony Brook Medicine, New York, with a few MRI, PET and nuclear medicine studies.
Overview
COVID-19-NY-SBU is a collection of routine clinical images from adults who tested PCR positive for COVID-19 at Stony Brook Medicine in New York, released through The Cancer Imaging Archive in 2021. Each patient is linked to a row of clinical data drawn from the hospital's COVID Data Commons, so the collection is mainly used to build diagnostic and prognostic models that combine chest imaging with outcomes such as death, ICU admission or mechanical ventilation.
Composition
The release covers 1,384 patients, 7,361 studies and 17,950 series. Chest radiographs dominate: 1,343 patients have computed radiography and 157 have digital radiography, most of them portable AP chest views, with some abdominal films. 458 patients have CT, mainly chest CT with and without contrast, CT pulmonary angiography and abdomen-pelvis CT. A handful of patients have abdominal MRI, FDG PET/CT or nuclear medicine lung perfusion scans.
The clinical table has one row per patient for a single encounter, chosen as the most severe COVID-positive visit. It holds demographics, comorbidities, symptoms at admission, first lab and vital values, ICU stay, ventilation and discharge status, coded to the OMOP Common Data Model with LOINC codes for labs. In that table, 183 patients died and 1,201 were discharged; 260 were in the ICU and 213 were ventilated. Ages are only given in three bins (18 to 59, 60 to 74, 75 to 90).
Acquisition
Images were exported from the Stony Brook radiology PACS and de-identified with POSDA. Dates are shifted, and the clinical encounter date uses the same shift so images can be aligned with the selected visit. Radiographs come from Carestream systems; CT scanners are from GE, Toshiba and Siemens.
Annotations
There are no image-level labels or segmentations. Targets come from the clinical table (outcome, ICU, ventilation, lab values).
Known limitations
- Single center, and only COVID-positive patients, so there is no negative control group.
- Imaging is whatever was ordered in care, so the number and timing of studies differ widely between patients.
- Clinical values cover one selected encounter only; many fields are missing (NA).
- Sex differs slightly between the clinical table and the DICOM headers.
- The collection summary mentions brain MRI, but the MR series in the release metadata are all abdominal.
Cohort
Aggregate numbers from the sources below. Bars are relative to the 1,384 subjects.
Sex
- Female 592 43%
- Male 760 55%
- Unknown 32 2%
gender_concept_name column in COVID-19-NY-SBU clinical data deidentified_overlap_tcia.csv.cleaned.csv (open download, CC BY 4.0), one row per patient
Modality
- X-ray (computed radiography) 1,343 97%
- CT 458 33%
- X-ray 157 11%
- SPECT / nuclear medicine 5 <1%
- MRI 4 <1%
- PET 4 <1%
Modality column, Patient ID column in COVID-19-NY-SBU manifest NBIA digest (open download, CC BY 4.0), one row per series, counted per Patient ID, Study Instance UID and series
Age
age.splits column in COVID-19-NY-SBU clinical data deidentified_overlap_tcia.csv.cleaned.csv (open download, CC BY 4.0), one row per patient
Condition
Groups can overlap
- COVID-19 1,384 100%
Summary in TCIA collection page COVID-19-NY-SBU (version 1, 2021-08-11)
Scanner vendor
- Carestream 1,365 99%
- GE HealthCare 363 26%
- Canon Medical 132 10%
- Siemens Healthineers 10 <1%
Manufacturer column, Patient ID column in COVID-19-NY-SBU manifest NBIA digest (open download, CC BY 4.0), one row per series, counted per Patient ID, Study Instance UID and series
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
Images download with the TCIA Data Retriever; the clinical CSV and its field template download directly.
Creative Commons Attribution 4.0 International
Use, share and adapt the data for any purpose, including commercial use, as long as you credit the creators.
TCIA lists the images, the clinical data and the clinical data template under CC BY 4.0. Users must also agree to the TCIA Data Usage Policy, which says users must "Not use the requested datasets, either alone or in concert with any other information, to identify or contact individual participants" and requires citing the dataset DOI (https://www.cancerimagingarchive.net/data-usage-policies-and-restrictions/).
What you can do
- Yes
- 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
Saltz J, Saltz M, Prasanna P, Moffitt R, Hajagos J, Bremer E, Balsamo J, Kurc T. (2021). Stony Brook University COVID-19 Positive Cases [Data set]. The Cancer Imaging Archive. https://doi.org/10.7937/TCIA.BBAG-2923
Sources
Every number on this page comes from one of these documents. Each chart names the table or page it is taken from. The raw numbers are in stats.csv.
- TCIA collection page COVID-19-NY-SBU (version 1, 2021-08-11) website
- COVID-19-NY-SBU manifest NBIA digest (open download, CC BY 4.0), one row per series, counted per Patient ID, Study Instance UID and series computed
- COVID-19-NY-SBU clinical data deidentified_overlap_tcia.csv.cleaned.csv (open download, CC BY 4.0), one row per patient computed