# Schürch CRC: colorectal cancer H&E crops with CODEX-derived cell regions and cell types at the invasive front

## Facts

- Provider: Schürch et al.; curated by SPORA
- License: CC-BY-4.0 (https://data.mendeley.com/datasets/mpjzbtfgfr/1)
- Size: 14 files
- Format: Parquet, PNG, JSON (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/csmp-hub/schurch-crc
- Access: public, readable without an account
- Category: Health & Biomedical
- Source: https://pmc.ncbi.nlm.nih.gov/articles/PMC7479520/
- Page: https://scigantic.com/archive/ba1665cb-6872-5880-9eed-f7e3e19cb0f7
- Keywords: Health & Biomedical, histopathology, image segmentation, Parquet, PNG, JSON, Hugging Face

## About

H&E images of colorectal cancer tissue cores from Schürch et al. (Cell 2020), who used CODEX multiplexed imaging to map cellular neighborhoods at the colorectal cancer invasive front and relate them to antitumoral immunity. The typed cohort has 140 tissue cores from 35 patients. Cores are cut into 512x512 RGB crops at about 0.377 micrometers per pixel with native pixel values, paired with instance masks carrying the original cell-region IDs (generated from DRAQ5 nuclear staining with the study's segmentation, so regions can extend beyond the visible nucleus) and cell-type masks built from the original per-cell table: epithelial tumor cells, immune populations, stromal populations, and the source's ambiguous and artifact categories. Files: train.parquet and test.parquet crops, crop-instances.parquet, cell-annotations.parquet, classes.json, instance counts and the original reference.bib; about 2.4 GB (measured 2026-11-17). Images and masks are stored as PNG or TIFF bytes inside Parquet; read them with polars or pandas plus Pillow, or with Hugging Face datasets. Use for H&E-only cell typing trained on CODEX labels and for neighborhood analysis. Curated and packaged on Hugging Face by the SPORA project (csmp-hub), which keeps the original license and citation files. The original Mendeley Data release (mpjzbtfgfr) is CC BY 4.0; cite Schürch et al., Cell 2020.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("csmp-hub/schurch-crc")
print(ds)
```

## Data structure

File types in sampled files: Parquet, PNG, JSON

Example file: `cell-annotations.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/ba1665cb-6872-5880-9eed-f7e3e19cb0f7
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/ba1665cb-6872-5880-9eed-f7e3e19cb0f7
- Over MCP: `get_schema_card` with id `ba1665cb-6872-5880-9eed-f7e3e19cb0f7` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [JSON datasets](https://scigantic.com/topics/json)
- [Image segmentation datasets](https://scigantic.com/topics/image-segmentation)
- [Health & Biomedical datasets](https://scigantic.com/topics/health-and-biomedical)
- [PNG datasets](https://scigantic.com/topics/png)
- [Histopathology datasets](https://scigantic.com/topics/histopathology)
