Nctcrche100K: 100,000 H&E Stained Colorectal Cancer Tissue Patches

100,000 non-overlapping image patches from hematoxylin & eosin (H&E) stained histological images of human colorectal cancer and normal tissue. All images are 224x224 pixels at 0.5 microns per pixel and are color-normalized using Macenko's method. The dataset was manually extracted from 86 H&E stained slides from the NCT Biobank and UMM pathology archive, and last updated on HuggingFace in October 2023.

Use Cases

• Training image classifiers for nine distinct tissue classes, including adenocarcinoma epithelium and cancer-associated stroma.
• Developing models for automated tissue segmentation in colorectal cancer histology slides.
• Benchmarking color normalization techniques on H&E stained images.
• Researching feature extraction for distinguishing cancerous from normal colon mucosa.

Strengths

• 100,000 image patches provide a substantial sample size for model training.
• Images are standardized to 224x224 pixels at a consistent 0.5 microns per pixel resolution.
• Color normalization using a cited method (Macenko's) reduces staining variation.
• Data originates from two established medical institutions: the NCT Biobank and UMM pathology archive.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last updated 2023-10-04 19:37:15; freshness should be verified.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

NCT Biobank (National Center for Tumor Diseases, Heidelberg, Germany) and UMM pathology archive (University Medical Center Mannheim, Mannheim, Germany).

Collection Method

Manually extracted from N=86 H&E stained human cancer tissue slides from formalin-fixed paraffin-embedded (FFPE) samples.

Freshness

Last updated 2023-10-04 19:37:15.

Geography

Germany (Heidelberg and Mannheim).