# Pediatric chest X-ray pneumonia dataset (Kermany et al., Guangzhou)

## Facts

- Provider: Daniel Kermany, Kang Zhang, Michael Goldbaum (Guangzhou Women and Children's Medical Center / UCSD); repackaged by hf-vision
- License: CC-BY-4.0 (https://data.mendeley.com/datasets/rscbjbr9sj/2)
- Size: 11 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/hf-vision/chest-xray-pneumonia
- Access: public, readable without an account
- DOI: https://doi.org/10.17632/rscbjbr9sj.2
- Category: Medicine
- Source: https://doi.org/10.1016/j.cell.2018.02.010
- Page: https://scigantic.com/archive/8318ffad-a5dd-5d7f-970a-c3522bdecdf6
- Keywords: Medicine, Parquet, Hugging Face

## About

This is the chest X-ray portion of the Kermany et al. dataset 'Labeled Optical Coherence Tomography (OCT) and Chest X-Ray Images for Classification' (Mendeley Data V2), released with the Cell 2018 paper 'Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning'. It contains 5,863 anterior-posterior chest radiographs (JPEG) from retrospective cohorts of pediatric patients aged one to five at Guangzhou Women and Children's Medical Center, labelled Normal or Pneumonia. Radiographs were screened for quality, graded by two expert physicians, and the evaluation set was checked by a third expert. The repository gives train (5,216), validation (16) and test (624) splits as Parquet files with image and label columns. As measured 2026-11-14 the current files are nine Parquet shards under data/ totalling about 1.2 GB; load with datasets (load_dataset('hf-vision/chest-xray-pneumonia')) or pyarrow. Widely used as a baseline for pneumonia classification, transfer learning, explainability and federated learning tutorials in medical imaging. The authors publish the data on Mendeley Data under CC BY 4.0; cite Kermany et al. 2018.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("hf-vision/chest-xray-pneumonia")
print(ds)
```

## Data structure

File types in sampled files: Parquet

Example file: `data/test-00000-of-00001.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/8318ffad-a5dd-5d7f-970a-c3522bdecdf6
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/8318ffad-a5dd-5d7f-970a-c3522bdecdf6
- Over MCP: `get_schema_card` with id `8318ffad-a5dd-5d7f-970a-c3522bdecdf6` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Medicine datasets](https://scigantic.com/topics/medicine)
