Computational Pathology · owkin
nct-crc-he
Colorectal cancer histopathology patch dataset across nine tissue classes for image classification benchmarking.
What nct-crc-he is
NCT-CRC-HE is a computational pathology dataset compiled for image classification tasks in oncology. It consists of human tissue slide patches stained with haematoxylin and eosin (H&E), encompassing both cancerous and non-cancerous tissue specimens from multiple anatomical sites.
The dataset provides standardised image patches categorised into nine separate tissue classes, serving as a standard reference benchmark for training and evaluating computer vision models in digital pathology.
How it was made
Tissue specimens were sourced from two clinical archives: the National Center for Tumor Diseases (NCT) biobank and the University Medical Center Mannheim (UMM) pathology archive.
Following slide preparation, the image patches were processed using Macenko normalisation to standardise colour and staining variations across histological samples.
Key facts
- Modality
- H&E-stained histological images
- Task
- Image classification
- Classes
- 9 tissue classes
- Normalisation
- Macenko normalisation
- Source institutions
- NCT biobank and UMM pathology archive
- Data format
- Parquet
What you can use it for
- Pathology Model Benchmarking
- Comparing and evaluating machine learning architectures on multi-class histological image classification.
- Colorectal Tissue Classification
- Training automated classifiers to identify colorectal cancer and distinguish between nine distinct tissue types.
Explore the data
The schema, real rows and a SQL filter — read in your browser, nothing downloaded.
Explore
Read the real rows without downloading anything
Reading rows…
LiveRead from owkin/nct-crc-he at the moment you asked. Nothing is cached or stored — every row above came from that request.
How the columns are distributed
Over 99 rows of default/nct_crc_he_100
image
image- min
- 224
- median
- 224
- mean
- 224
- max
- 224
label
class_label- ADI11%
- BACK11%
- DEB11%
Strengths and limits
Good at
- Provides nine standardised tissue classes covering cancer and non-cancer tissue types.
- Applies Macenko normalisation to minimise staining inconsistencies across samples.
- Constructed from established clinical repositories at the NCT biobank and UMM pathology archive.
Watch out for
- Operates strictly at the patch level rather than providing whole-slide imaging context.
- Restricted to nine pre-defined histological classes originating from specific tissue archives.
Questions
Can I look inside nct-crc-he without downloading it?
Yes. This page reads its schema, real rows and per-column statistics in the browser and a SQL filter. Nothing is downloaded to your machine.
What licence is nct-crc-he under?
Its authors declare cc-by-sa-3.0. Read the terms on the original page before reusing the data.
What staining technique is used for the tissue slides?
The images consist of tissue sections stained with haematoxylin and eosin (H&E).
Where were the underlying tissue samples collected?
The specimens originate from the National Center for Tumor Diseases (NCT) biobank and the University Medical Center Mannheim (UMM) pathology archive.
How are variations in slide staining handled?
The dataset images are normalised using the Macenko colour normalisation algorithm.
Written from nct-crc-he's own dataset card. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.