# SLiMIA: Spheroid Light Microscopy Image Atlas, 7,984 3D cell culture spheroid images with masks from 9 microscopes

## Facts

- Provider: Eva Blondeel, Filippo Piccinini, Olivier De Wever et al.; mirrored by MedOtter
- License: CC-BY-4.0 (https://figshare.com/articles/dataset/27390402)
- Size: 59 files
- Format: Parquet (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/MedOtter/SLiMIA
- Access: public, readable without an account
- DOI: https://doi.org/10.6084/m9.figshare.c.7486311.v1
- Category: Life Sciences
- Source: https://doi.org/10.1038/s41597-025-04441-x
- Page: https://scigantic.com/archive/e73c31db-664e-5eb9-bbee-c57a97aeea95
- Keywords: Life Sciences, image segmentation, Parquet, Hugging Face

## About

SLiMIA (Spheroid Light Microscopy Image Atlas) is a multi-lab collection of transmitted-light images of three-dimensional cell culture spheroids, tumor and non-tumor, made for morphometric analysis such as spheroid size, area and shape. This mirror holds 7,984 image and manual mask pairs acquired on nine microscopes (five phase-contrast and four brightfield) and covering 48 cell lines, 8 culture media and 4 spheroid formation methods. Masks are binary (0 or 255) with a single spheroid foreground class; image sizes range from 800x600 to 2592x1944 pixels. There is no upstream train/test split, so all pairs are in train, and the microscope and originating lab are recorded per row, which matters because they introduce batch effects. Twelve unpaired upstream files are listed in unpaired.json and scale-bar images are in scalebars/. Parquet under data/, about 8.4 GB (measured 2026-11-21); read with Hugging Face datasets or pandas plus Pillow. Use for spheroid segmentation, organoid and tumor spheroid size measurement and cross-microscope generalization. The nine figshare data articles of collection DOI 10.6084/m9.figshare.c.7486311.v1 are CC BY 4.0 (the companion paper itself is CC BY-NC-ND); cite Blondeel et al., Scientific Data 2025 (doi 10.1038/s41597-025-04441-x).

## Load the data

```python
from datasets import load_dataset
ds = load_dataset("MedOtter/SLiMIA")
print(ds)
```

## Data structure

File types in sampled files: Parquet

Example file: `data/train-00000-of-00042.parquet`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/e73c31db-664e-5eb9-bbee-c57a97aeea95
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/e73c31db-664e-5eb9-bbee-c57a97aeea95
- Over MCP: `get_schema_card` with id `e73c31db-664e-5eb9-bbee-c57a97aeea95` at https://api.scigantic.com/mcp

## Related topics

- [Parquet datasets](https://scigantic.com/topics/parquet)
- [Image segmentation datasets](https://scigantic.com/topics/image-segmentation)
- [Life Sciences datasets](https://scigantic.com/topics/life-sciences)
