Dataset No. 71
Pathology-Confirmed Breast Cancer Mammography with Localization and Classification
Annotation: Localization (bounding boxes of suspected breast cancer), Classification (lesion type, breast density, BI-RADS category), Radiology Reports, Pathology Reports
Clinical Information: Age, Sex, Days from mammography to pathology diagnosis
Location: Okayama in Japan
Notes
This dataset is rigorously de-identified and excludes cases with excessive noise. It includes cases with breast lesions assessed as BI-RADS category 3 or higher, with at least one lesion per case confirmed as breast cancer by histopathological diagnosis. Lesions were annotated by a physician certified in mammography screening interpretation and independently double-checked by a second certified physician. All annotations were generated for lesions categorized as BI-RADS 3 or higher. For each lesion, the corresponding findings on the craniocaudal (CC) and mediolateral oblique (MLO) views were assigned the same bounding box number to ensure cross-view consistency. Each bounding box label is linked to the corresponding pathological diagnosis, such as invasive ductal carcinoma, ductal carcinoma in situ, invasive lobular carcinoma, mucinous carcinoma, or no evidence of malignancy.
Need a custom dataset?
Tell us what you're looking for.