Automated Nuclear Pleomorphism Scoring: 118 Breast Cancer Whole-Slide Images

118 digital pathology whole-slide images of breast cancer surgical resections from Radboud University Medical Centers, stained with hematoxylin and eosin. The dataset was used as a test set in a 2022 NPJ Breast Cancer paper for developing automated nuclear pleomorphism scoring models. The images are provided in TIFF format with a maximum spacing of 0.5 um/px and are released under a CC BY-NC 4.0 license.

Use Cases

• Benchmarking automated nuclear pleomorphism scoring algorithms based on whole-slide images.
• Training or validating computer-aided diagnosis systems for breast cancer based on digital pathology.
• Comparing algorithmic predictions against a panel of four pathologists' opinions as described in the paper.

Strengths

• Includes 118 whole-slide images from a specific medical center, providing a focused test set.
• Accompanied by a web-based evaluation platform for standardized algorithm comparison.
• Images are converted to a broadly accessible TIFF format from a proprietary MRXS format.

Limitations

• Row count and specific image dimensions are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect geographic and institutional bias inherent to the single-source collection.

Provenance

Source

Radboud University Medical Centers, Nijmegen, The Netherlands.

Collection Method

Scanned from surgical resection samples using a 3DHistech P1000 scanner.

Geography

Nijmegen, The Netherlands.