Metadata for refining Breast Cancer Immunohistochemistry image generation challenge data at patch-level

# Refined HER2 Annotations for BCI Dataset This repository contains a metadata file listing paired HE/IHC image filenames from the original **BCI Dataset**, along with **refined HER2 annotations** based on expert pathological review. ## 📁 File Contents **Filename**: `BCI_metadata_pairs_refined.csv` | Column | Description ||--------------------|-----------------------------------------------------------------------------|| `pair_id` | Unique identifier for each image pair || `partition` | Data split (`train` or `test`) || `file_name_he` | Filename of the HE image (as in original BCI dataset) || `file_name_ihc` | Filename of the IHC image (as in original BCI dataset) || `her2_level` | Refined HER2 category (`0`, `1+`, `2+`, `3+`) based on pathologist revision of the patch | ## 🧠 Purpose This metadata provides updated HER2 annotations for a subset of the BCI Dataset, filtered and reviewed by a professional pathologist. It is intended to improve training, benchmarking, and reproducibility in research using BCI-based histopathological image analysis. ## ⚠️ Important Notes - **No image data is included** in this repository.- Filenames refer to the original BCI Dataset, which must be obtained separately.- HER2 levels included here are revised annotations at patch-level, not the original labels from the dataset (at WSI-level). ## 📚 Citation Requirements If you use this metadata, you **must also cite**: 1. **The original dataset**: ```bibtex @article{liu2022bci, title={BCI: Breast Cancer Immunohistochemical Image Generation through Pyramid Pix2pix}, author={Liu, Shengjie and Zhu, Chuang and Xu, Feng and Jia, Xinyu and Shi, Zhongyue and Jin, Mulan}, journal={arXiv preprint arXiv:2204.11425}, year={2022} } ``` 2. **This metadata resource** (Zenodo DOI), as the source of the refined patch-level HER2 annotations. Please include the DOI of this Zenodo record in any publication or derived work. ## 📝 License This metadata is provided under the **Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)** license. It is intended for **non-commercial academic use only**. Redistribution or use of the original image data must comply with the license of the original BCI Dataset. ## ✉️ Contact For questions or collaboration inquiries, please contact the maintainers via the Zenodo page or institutional email.