Colorectal Cancer Histology Textures with 5,000 Image Tiles

A collection of 5,000 histological image tiles of 150x150 pixels each, representing eight tissue categories from human colorectal cancer. The data includes 10 larger 5000x5000 pixel images containing multiple tissue types, digitized at 20x magnification. It was created by researchers at Heidelberg University from anonymized, formalin-fixed paraffin-embedded tumor samples.

Use Cases

• Training image classifiers for tissue type recognition based on the eight specified categories.
• Developing texture analysis algorithms for histological images based on the provided multi-class texture data.
• Benchmarking segmentation models on the larger 5000x5000 pixel images containing multiple tissue types.

Strengths

• Contains 5,000 individual image tiles for model training.
• Images are high-resolution, digitized at 20x magnification with a scale of 0.495 µm per pixel.
• Includes 10 larger composite images useful for segmentation tasks.
• Dataset creation was approved by an institutional ethics board (approval 2015-868R-MA).

Limitations

• Row count for the image tiles is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Institute of Pathology, University Medical Center Mannheim, Heidelberg University.

Collection Method

Retrospective analysis of anonymized, formalin-fixed paraffin-embedded human colorectal adenocarcinoma samples from a pathology archive.

Geography

Mannheim, Germany.