---
license: mit
pretty_name: TCGA-mini
size_categories:
- 1K<n<10K
tags:
- pathology
- whole-slide-imaging
- tcga
- medical-imaging
---

# TCGA-mini

`TCGA-mini` is a curated whole-slide imaging (WSI) dataset composed of TCGA `.svs` pathology slides.

- Number of slides: `1319` (`.svs`)
- Location in repo: `slides/`
- Storage footprint on the Hub: about `1.42 TiB` (`1450.94 GiB`, Git LFS-backed)

## Dataset structure

```text
TCGA-mini/
├── README.md
└── slides/
    ├── TCGA-xxxx-xxxx-xxZ-00-DX1.svs
    └── ...
```

## Data format

- File type: Aperio whole-slide image files (`.svs`)
- Naming: Original TCGA-style slide identifiers are preserved
- Labels/annotations: Not included in this repo

## Download

Download all slides:

```bash
hf download W8Yi/TCGA-mini \
  --repo-type dataset \
  --include "slides/*.svs" \
  --local-dir ./TCGA-mini
```

Download a single slide:

```bash
hf download W8Yi/TCGA-mini \
  --repo-type dataset \
  --include "slides/TCGA-06-0743-01Z-00-DX1.svs" \
  --local-dir ./TCGA-mini
```

## Quick usage example (OpenSlide)

```python
from openslide import OpenSlide

slide = OpenSlide("TCGA-mini/slides/TCGA-06-0743-01Z-00-DX1.svs")
print("levels:", slide.level_count)
print("dimensions:", slide.dimensions)
```

## Intended use

This dataset is intended for research and educational use in computational pathology and WSI pipelines.
It is not intended for clinical diagnosis or direct patient care.

## Source and attribution

Slides originate from The Cancer Genome Atlas (TCGA) program.
Users are responsible for complying with all applicable TCGA/GDC data-use and publication policies.

## Limitations

- No train/val/test split is provided
- No labels or clinical tables are included in this repo
- Whole-slide files are very large and require substantial storage and I/O throughput

## Citation

If you use this dataset, please cite:

1. This Hugging Face dataset repository: `W8Yi/TCGA-mini`
2. The TCGA program and relevant data source publications