# Phillips CTCL: cutaneous T-cell lymphoma skin biopsy H&E with CODEX cell regions and 23 cell types

## Facts

- Provider: Phillips et al.; curated by SPORA
- License: CC-BY-4.0 (https://data.mendeley.com/datasets/3gmvy3bcmk/1)
- Size: 17 files
- Format: Parquet, PNG, JSON (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/csmp-hub/phillips-ctcl
- Access: public, readable without an account
- DOI: https://doi.org/10.17632/3gmvy3bcmk.1
- Category: Health & Biomedical
- Source: https://doi.org/10.1038/s41467-021-26974-6
- Page: https://scigantic.com/archive/096021ca-18e4-5cf3-a879-2cb2bad7eeaf
- Keywords: Health & Biomedical, histopathology, Parquet, PNG, JSON, Hugging Face

## About

H&E images of skin biopsies from patients with cutaneous T-cell lymphoma (CTCL) in the study of Phillips et al. (Nature Communications 2021), which used CODEX multiplexed imaging to show that immune cell spatial arrangement predicts response to PD-1 blockade with pembrolizumab. The dataset has 69 paired tissue regions from 14 patients. Each 1,440x1,920-pixel core is tiled into 108 overlapping 512x512 RGB crops at 0.377 micrometers per pixel (128-pixel stride), with instance masks that keep the original cell-region IDs and cell-type masks for 23 types from the source table: T cell subsets, B cells, plasma cells, macrophages, dendritic cells, Langerhans cells, mast cells, neutrophils, stromal cells, epithelium and tumor cells (here malignant T cells, not epithelial tumor). Files include crop Parquet tables, cell-annotations.parquet, crop-quality.parquet, a reviewed training index, classes.json and reference.bib; about 3.5 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. 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 (3gmvy3bcmk) is CC BY 4.0; cite Phillips et al., Nature Communications 2021, doi 10.1038/s41467-021-26974-6.

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("csmp-hub/phillips-ctcl")
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/096021ca-18e4-5cf3-a879-2cb2bad7eeaf
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/096021ca-18e4-5cf3-a879-2cb2bad7eeaf
- Over MCP: `get_schema_card` with id `096021ca-18e4-5cf3-a879-2cb2bad7eeaf` at https://api.scigantic.com/mcp

## Related topics

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