BreCaHAD: Breast Cancer Histopathology Images with Cell-Level Annotations

BreCaHAD is a dataset for breast cancer histopathological annotation and diagnosis. It contains original images and JSON files providing ground-truth annotations for six cell and tissue classes, including mitosis, apoptosis, tumor nuclei, and tubules. The dataset was created by Alper Aksaç.

Use Cases

• Training object detection models to identify tumor nuclei based on blue circle annotations.
• Developing classifiers for mitosis detection based on orange circle annotations.
• Segmenting tubule structures based on light blue circle annotations for lumen.
• Analyzing cell death patterns using apoptosis annotations marked with green circles.
• Comparing tumor and non-tumor nuclei distributions using the provided Excel spreadsheet.

Strengths

• Provides annotations for six distinct histopathological classes: mitosis, apoptosis, tumor nuclei, non-tumor nuclei, tubule, and non-tubule.
• Includes multiple data formats: original images, annotated example images, JSON ground truth files, and an Excel summary.
• Annotations include centroid coordinates normalized to [0,1] for spatial analysis.

Limitations

• The total number of images and rows is unknown, limiting assessment of dataset scale.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Alper Aksaç