Protege Discovery Platform
Viewing as USER

NIH Chest X-ray Sample IMAGE patient-level

NIH Clinical Center · NIH Open Access (see dataset terms)
✓ access granted — preview enabled

Chest X-ray images organised into folders by NIH Finding Label (Pneumonia, Effusion, Atelectasis, Infiltration, No_Finding), each image linked to CSV metadata and a JSON sidecar. A quick starting point for chest-X-ray classification and multimodal imaging experiments.

Potential training uses
  • Chest X-ray classification (e.g. pneumonia)
  • Weakly-labelled medical image pretraining
  • Image↔metadata linkage and multimodal tasks
Description
Sample of NIH ChestX-ray14 PNG images organised into folders by source Finding Label, with per-image JSON sidecars and CSV metadata.
Source
https://nihcc.app.box.com/v/ChestXray-NIHCC
Tags
radiologyimagechest-xraypatient-level

Snapshot & provenance

Snapshot kind
TRUSTED_LOCAL read-only reference (not copied)
State
PUBLISHED
Files
34
Package SHA-256
— (trusted local seed)
Published
2026-07-11T08:58:35.603762+00:00

Quality overview

0 error · 1 warning · 0 info

Assets (14)

CSV_TABLE · 2

NameSource pathFilesRowsQuality
BBox_List_2017.csv metadata/BBox_List_2017.csv 1 984 WARNING
Data_Entry_2017_v2020.csv metadata/Data_Entry_2017_v2020.csv 1 OK

DOCUMENT · 1

NameSource pathFilesRowsQuality
README.md metadata/README.md 1 OK

IMAGE_COLLECTION · 5

NameSource pathFilesRowsQuality
Atelectasislabel: Atelectasis images/Atelectasis 3 3 OK
Pneumonialabel: Pneumonia images/Pneumonia 3 3 OK
Infiltrationlabel: Infiltration images/Infiltration 3 3 OK
Effusionlabel: Effusion images/Effusion 3 3 OK
No_Findinglabel: No_Finding images/No_Finding 3 3 OK

JSON_SIDECAR_COLLECTION · 6

NameSource pathFilesRowsQuality
Effusion (json sidecars) images/Effusion 3 3 OK
No_Finding (json sidecars) images/No_Finding 3 3 OK
images (json sidecars) images 1 1 OK
Atelectasis (json sidecars) images/Atelectasis 3 3 OK
Infiltration (json sidecars) images/Infiltration 3 3 OK
Pneumonia (json sidecars) images/Pneumonia 3 3 OK

Annotations

HUMAN_CURATED summary · APPROVED
Chest X-ray PNG images grouped by folder-derived Finding Label. Folder labels are source-derived and weak — not a clinical diagnosis for each image.
HUMAN_CURATED limitation · APPROVED
Folder names such as 'Pneumonia' come from Data_Entry_2017 labels and are not per-image clinical ground truth.
HUMAN_CURATED customer_summary · APPROVED
Chest X-ray images organised into folders by NIH Finding Label (Pneumonia, Effusion, Atelectasis, Infiltration, No_Finding), each image linked to CSV metadata and a JSON sidecar. A quick starting point for chest-X-ray classification and multimodal imaging experiments.
HUMAN_CURATED training_uses · APPROVED
Chest X-ray classification (e.g. pneumonia) Weakly-labelled medical image pretraining Image↔metadata linkage and multimodal tasks