NIH ChestX-ray14
Frontal chest radiographs with 14 text-mined thoracic finding labels (such as effusion, nodule, pneumothorax) plus No Finding
Open the source page ↗ See it in the directory
At a glance
- What it is
- Frontal chest radiographs with 14 text-mined thoracic finding labels (such as effusion, nodule, pneumothorax) plus No Finding
- Area
- Chest
- Modality
- DX
- Format
- PNG
- Size
- 112,120 images from 30,805 patients
- Normals included
- Yes
- License
- NIH Clinical Center terms (no use restrictions; attribution required)
- Commercial use
- Allowed by the license, with its attribution terms
- Access
- Open download, no account
- Status
- Verified: source page reviewed by Hounsfield Lab
- Added
- 2026-10-26
- Last checked
- 2026-11-11
- Source
- nihcc.app.box.com/v/ChestXray-NIHCC
How to access
Open the NIH Clinical Center Box folder and download the image archives, the label CSV, and the bounding-box CSV. No account is needed. Google Cloud also hosts a copy as a public dataset.
Limitations
Labels were text-mined from reports by NLP and are estimated to be more than 90% accurate, so expect label noise. The reports themselves are not released. Only a small subset has bounding boxes. Images are PNG, not DICOM. Single institution. Many images are follow-up studies of the same patients, so split by patient ID.
Citation required?
Yes. The terms ask you to link the NIH download site, cite Wang et al., ChestX-ray8, IEEE CVPR 2017, and acknowledge the NIH Clinical Center as the data provider.
Related open-source tools
- TorchXRayVision: Library of chest X-ray dataset loaders and pretrained classification, segmentation and autoencoder models
- CXAS: Chest X-ray anatomy segmentation (Mask2Former models: 187 frontal and 177 lateral classes) plus 132 named landmarks and landmark-based measurements such as cardiothoracic ratio, costophrenic angle, Cobb angle and the Haller index, in mm for DICOM input; cxas_idc writes one multi-segment DICOM SEG and a TID 1500 SR per radiograph; trained on PAXRay++ CT-projected radiographs
- CheXagent: Foundation model for chest X-ray interpretation
- MedRAX: Reasoning agent for chest X-rays that orchestrates multiple CXR models (ICML 2025)
More chest datasets
- CheXpert Plus: DX, CR · Stanford AIMI research use agreement (non-commercial; commercial license sold separately)
- COCA: CT · Stanford AIMI research use agreement (non-commercial; commercial license sold separately)
- INSPECT: CT · Stanford AIMI research use agreement (non-commercial; commercial license sold separately)
- LDCT-and-Projection-data: CT · Chest, liver, and phantom images CC BY 4.0; head images under the NIH Controlled Data Access Policy
- LIDC-IDRI: CT, CR, DX · CC BY 3.0
- Medical Segmentation Decathlon: CT, MR · CC BY-SA 4.0
Hounsfield Lab links to the source and does not host images. The license and access summary reflects the source page on the last-checked date; terms can change, so read the source before you download. Not legal advice.