乳腺癌组织病理学数据库

BreaKHis v1 是乳腺癌组织病理学图像分类的公开数据集，包含82名患者在40X、100X、200X和400X放大倍数下采集的7,909张RGB图像（2,480张良性，5,429张恶性），分辨率700×460，PNG格式。数据由巴西P&D实验室构建，图像命名包含活检方法、肿瘤类别、类型、患者ID和放大倍数。该数据集支持乳腺癌良恶性分类研究，适用于医学图像分析与医疗大模型训练，具备多倍率、标注清晰、来源权威等特点，便于算法基准测试与临床辅助诊断应用。

BreaKHis v1 is a public dataset for breast cancer histopathological image classification. It contains 7,909 RGB images (2,480 benign, 5,429 malignant) collected from 82 patients at 40×, 100×, 200× and 400× magnification factors, with a resolution of 700×460 and stored in PNG format. The dataset was constructed by the P&D Laboratory of Brazil. Image filenames include biopsy method, tumor category, type, patient ID and magnification factor. This dataset supports research on benign and malignant classification of breast cancer, and is suitable for medical image analysis and medical large language model training. It features multiple magnification options, clear annotations and authoritative sources, facilitating algorithm benchmark testing and clinical auxiliary diagnosis applications.

搜集汇总

背景与挑战

背景概述

该数据集是乳腺癌组织病理学图像分类的公开数据集（BreaKHis v1），包含82名患者的7,909张RGB图像，涵盖良性和恶性类别，图像在多个放大倍数下采集，分辨率为700×460。数据集发布于2015年，总容量3.99GB，采用CC-BY-4.0协议，适用于医疗科研中的乳腺癌病理分析任务。