Categorized Digital Database for Low energy and Subtracted Contrast Enhanced Spectral Mammography images

Deep learning (DL) has a promising potential in reducing the workload of radiologists and helping them provide a more accurate diagnosis. However, fully annotated and large-sized datasets are required. This dataset is a collection of 2,006 high-resolution Contrast-enhanced spectral mammography (CESM) images with annotations and medical reports.
Acquisition protocol:

CESM is done using the standard digital mammography equipment, with additional software that performs dual-energy image acquisition. Two minutes after intravenously injecting the patient with non-ionic low-osmolar iodinated contrast material (dose: 1.5 mL/kg), craniocaudal (CC) and mediolateral oblique (MLO) views are obtained. Each view comprises two exposures, one with low energy (peak kilo-voltage values ranging from 26 to 31kVp) and one with high energy (45 to 49 kVp). Low and high-energy images are then recombined and subtracted through appropriate image processing to suppress the background breast parenchyma. A complete examination is carried out in about 5-6 minutes.

Image preprocessing:

The images were converted from DICOM to JPEG using RadiAnt with best 100% image quality (lossless). They have an average of 2355 x 1315 pixels.

Supporting data:

Full medical reports are also provided for each case (DOCX) along with manual segmentation annotation for the abnormal findings in each image (CSV file). Each image with its corresponding manual annotation (breast composition, mass shape, mass margin, mass density, architectural distortion, asymmetries, calcification type, calcification distribution, mass enhancement pattern, non-mass enhancement pattern, non-mass enhancement distribution, and overall BIRADS assessment) is compiled into 1 Excel file.

https://www.robots.ox.ac.uk/~vgg/software/via/via.html was used for the segmentation annotation. It can be used to show the annotations on the images by clicking on Annotation--> import annotations (from csv), and then proceeding to upload any image to view the annotations drawn over it. Moreover, a helper repository is created to help with pre-processing, model training, model evaluation, and segmentation annotation loading: https://github.com/omar-mohamed/CDD-CESM-Dataset

Regarding the tabs on the annotations Excel file, these are commonly used radiological descriptors as defined by the American College of Radiology 2013 lexicon.

深度学习（Deep Learning, DL）在减轻放射科医师工作负担、辅助其提供更精准诊断方面具有可观潜力。然而，该领域的研究依赖于大规模高质量标注数据集。本数据集收录了2006幅高分辨率对比增强光谱乳腺摄影（Contrast-enhanced spectral mammography, CESM）图像，且均配有标注信息与完整医学报告。

采集协议：
CESM采用标准数字化乳腺X线设备完成采集，搭配支持双能图像采集的专用软件。经静脉注射非离子型低渗碘对比剂（剂量：1.5 mL/kg）2分钟后，获取头尾位（craniocaudal, CC）与内外斜位（mediolateral oblique, MLO）图像。每个投照体位包含两次曝光：一次为低能量模式（管电压峰值范围26~31 kVp），另一次为高能量模式（45~49 kVp）。随后通过适配的图像处理算法对高低能量图像进行重组与减影，以抑制乳腺实质背景信号。完整检查流程耗时约5~6分钟。

图像预处理：
所有图像均通过RadiAnt软件从DICOM格式转换为JPEG格式，转换过程采用100%最优无损质量设置。图像平均分辨率为2355×1315像素。

配套支持数据：
每例样本均附带完整医学报告（DOCX格式），以及针对图像中异常病灶的手动分割标注（CSV格式文件）。每幅图像及其对应的手动标注项（包括乳腺构成、肿块形态、肿块边缘、肿块密度、结构扭曲、不对称征象、钙化类型、钙化分布、肿块增强模式、非肿块样增强模式、非肿块样增强分布以及整体乳腺影像报告和数据系统（Breast Imaging Reporting and Data System, BIRADS）评估结果）均整合至单个Excel文件中。

本数据集的分割标注采用VGG图像标注工具（https://www.robots.ox.ac.uk/~vgg/software/via/via.html）完成。使用者可通过点击菜单栏"Annotation"→"Import annotations (from csv)"导入标注文件，随后上传任意图像即可查看叠加于图像之上的标注结果。此外，项目组还搭建了辅助工具仓库，用于辅助图像预处理、模型训练、模型评估及分割标注加载，仓库地址为：https://github.com/omar-mohamed/CDD-CESM-Dataset。

关于标注Excel文件中的标签项，均采用美国放射学会2013版术语词典中定义的通用放射学描述规范。