# Hoefflin glioma: IDH-mutant glioma H&E crops with CODEX-derived cell regions and glioma cell states

## Facts

- Provider: Rouven Hoefflin et al.; curated by SPORA
- License: CC-BY-4.0 (https://zenodo.org/records/21335411)
- Size: 18 files
- Format: Parquet, PNG, JSON (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/csmp-hub/hoefflin-glioma
- Access: public, readable without an account
- Category: Health & Biomedical
- Source: https://doi.org/10.1016/j.ccell.2026.08.005
- Page: https://scigantic.com/archive/e97ce904-d885-5f9e-88ef-1dbcc7f0e679
- Keywords: Health & Biomedical, histopathology, image segmentation, Parquet, PNG, JSON, Hugging Face

## About

H&E images of IDH-mutant gliomas from Hoefflin et al. (Cancer Cell 2026), who paired H&E with CODEX multiplexed imaging to study how malignant cells and their microenvironment change across glioma grades. The collection has 512x512 crops at 0.5 micrometers per pixel from 30 slides of 25 patients; H&E was aligned to DAPI by the source creators and keeps original RGB values. Instance masks (uint32 TIFF) keep the original cell-region IDs (nuclear segmentation expanded by 2 micrometers) and cell-type masks carry six glioma cell states (cAC1, cAC2, cOPC, cUndiff, cMES, cMESHyp) plus astrocytes, oligodendrocytes, neurons, vascular cells, macrophages, CD4 and CD8 T cells, neutrophils and B cells, covering 2,448,212 distinct source cells. Files: train and test crop Parquet tables, crop-instances.parquet, cell-annotations.parquet, crop quality flags, classes.json and reference.bib; about 9.2 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-based prediction of glioma cell states and brain tumor microenvironment mapping. Curated and packaged on Hugging Face by the SPORA project (csmp-hub), which keeps the original license and citation files. The authors' Zenodo record 21335411 (IDH-mut glioma CODEX dataset) is CC BY 4.0; cite Hoefflin et al., Cancer Cell 2026 (doi 10.1016/j.ccell.2026.08.005).

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("csmp-hub/hoefflin-glioma")
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/e97ce904-d885-5f9e-88ef-1dbcc7f0e679
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/e97ce904-d885-5f9e-88ef-1dbcc7f0e679
- Over MCP: `get_schema_card` with id `e97ce904-d885-5f9e-88ef-1dbcc7f0e679` 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)
