Mammogram Image Dataset for Breast Cancer Detection

Published: 15 February 2026| Version 2 | DOI: 10.17632/88vzgys5vg.2
Contributors:
Jamolbek Mattiev,
,

Description

The dataset was retrospectively compiled following approval from the relevant Institutional Review Board (IRB) or Ethics Committee. Data were sourced from de-identified patient biopsy records at a tertiary hospital, encompassing a defined collection window from January 1, 2023, to October 31, 2023. The dataset comprises 614 original (361 benign and 253 malignant cases) breast X-ray images (mammograms) sourced exclusively from a single institution: a hospital in Polokwane, Limpopo, South Africa. All images have been clinically validated to establish the ground truth by a panel of radiologists and specialized medical practitioners. The acquisition process utilized a dedicated breast X-ray system equipped with specialized accessories to minimize radiation exposure, adhering to safety protocols. All data is standardized in the PNG format with a uniform resolution of 1920 × 1080 pixels. The Breast Cancer Augmented directory contains augmented versions of the original mammography images, generated to support data-driven model development and evaluation. The augmented images preserve the original diagnostic labels (benign or malignant) and maintain spatial alignment with their corresponding source images.

Files

Categories

Medical Imaging, Image Processing, Breast Cancer, X-Ray, Early Cancer Detection, Deep Learning

Licence