Breast and Lymph Node Ultrasound Image Dataset for Lesion Detection and Characterization

This dataset includes a curated, privacy-preserving subset of <b>100</b> breast and axillary lymph node ultrasound images collected at Taichung Veterans General Hospital for public illustration and methodological reference. In addition, annotation files corresponding to the curated dataset (approximately 3,800 images) are provided in JSON (Labelme) format. These annotation files contain expert-defined lesion annotations and associated metadata, allowing users to examine the annotation structure and use the dataset in detection benchmarking and related image analysis workflows.All images were anonymized prior to release in accordance with institutional privacy requirements, and no patient-identifying information was preserved. Access to the complete curated image set is available under controlled conditions through Figshare in accordance with the IRB-approved Data Use Agreement.

本数据集包含100张乳腺及腋窝淋巴结超声图像的精选隐私保护子集，该数据集采集自台中荣民总医院，用于公开展示与方法学参考。此外，本次还提供了对应该精选数据集（覆盖约3800张图像）的标注文件，格式为JSON（Labelme）。这些标注文件包含专家定义的病灶标注及关联元数据，可供使用者研究标注结构，并将该数据集应用于检测基准测试及相关图像分析工作流中。所有图像在发布前均已按照机构隐私要求完成匿名化处理，未留存任何可识别患者的信息。完整精选图像集的访问需通过Figshare平台，并遵循经机构审查委员会（Institutional Review Board, IRB）批准的数据使用协议，在受控条件下获取。