Skip to content

Dataset · MahmoodLab

Spatial transcriptomics with histology

MahmoodLab/hest

1,276 spatial transcriptomics profiles aligned with whole-slide images for multimodal tasks.

What's inside

HEST-1k is a collection of 1,276 spatial transcriptomics profiles, each linked and aligned with a whole histological slide image (pixel less than 1.15 µm/px) and metadata. Gathered from 180 public and internal cohorts: 26 organs, 2 species (human and mouse), 398 cancer samples of 25 types. Processing yielded over 1.5 million expression-morphology pairs and over 76 million data points. Tasks include image classification, feature extraction, and segmentation; language is English.

What you can do with it

  • Train a model to map histology and gene expression
  • Verify feature extraction on multimodal medical data
  • Train tissue segmentation on histological slides
Tasks
Image classification, Embeddings, Image segmentation
Languages
English
Size
100B–1T rows
Downloads this month
32.2K
Likes
146
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("MahmoodLab/hest")