A early-stage Breast Cancer Ultrasound Dataset with Paired Images and Videos for Axillary Lymph Node Metastasis Prediction

Axillary lymph node (ALN) metastasis is a critical determinant of treatment strategies and prognosis in early-stage breast cancer. Ultrasound imaging has emerged as a non-invasive tool for ALN metastasis assessment, but the absence of a standardized ultrasound benchmark dataset has hindered the advancement and validation of computer-assisted diagnostic (CAD) techniques—particularly in underserved regions where access to high-quality diagnostic services remains limited. To address this gap, we present ALN-Ultra, the first large-scale open-access dataset comprising paired ultrasound images and videos from 257 breast cancer patients, supplemented with expert diagnostic evaluations and histopathological biopsy results. This dataset supports the development of machine learning algorithms based on 2D images and 3D videos, with three core objectives: improving diagnostic accuracy for ALN metastasis, elucidating the contribution of different data modalities to predictive performance, and facilitating cross-modal comparison. By enabling systematic evaluation of data modality efficacy, ALN-Ultra serves as a foundational resource for advancing CAD methodologies and promoting equitable detection of ALN metastasis in clinical practice.