BIMCV COVID-19

BIMCV-COVID19+ 数据集是一个大型数据集，其中包含 COVID-19 患者的胸部 X 光图像 CXR（CR、DX）和计算机断层扫描（CT）成像以及他们的放射学发现、病理学、聚合酶链反应（PCR）、免疫球蛋白 G（ IgG) 和免疫球蛋白 M (IgM) 诊断抗体测试和来自巴伦西亚地区医学影像库 (BIMCV) 医学影像数据库的影像学报告。这些发现被映射到标准的统一医学语言系统 (UMLS) 术语中，它们涵盖了广泛的胸部实体，与以前数据集中注释的实体数量大大减少形成鲜明对比。图像以高分辨率存储，实体使用医学成像数据结构 (MIDS) 格式的解剖标签进行本地化。此外，一组放射科专家对 23 幅图像进行了注释，以包括放射学发现的语义分割。此外，还提供了广泛的信息，包括患者的人口统计信息、投影类型和成像研究的采集参数等。数据库的这些迭代包括 7377 CR、9463 DX 和 6687 CT 研究。

The BIMCV-COVID19+ Dataset is a large-scale medical imaging repository containing chest X-ray (CXR) images (including CR and DX modalities) and computed tomography (CT) scans of COVID-19 patients, alongside their radiological findings, pathology results, polymerase chain reaction (PCR) test results, immunoglobulin G (IgG) and immunoglobulin M (IgM) diagnostic antibody assay results, and radiological reports sourced from the BIMCV (Valencian Regional Medical Imaging Repository) medical imaging database. These findings are mapped to standard Unified Medical Language System (UMLS) terminology, covering a broad spectrum of thoracic entities—a stark contrast to the significantly fewer annotated entities found in prior datasets. All images are stored at high resolution, with entities localized using anatomical labels formatted in accordance with the Medical Imaging Data Structure (MIDS) standard. Additionally, a panel of board-certified radiologists annotated 23 images to provide semantic segmentation of their corresponding radiological findings. Furthermore, comprehensive supplementary metadata is provided, including patient demographic information, projection types, acquisition parameters of imaging studies, and other relevant clinical details. This iteration of the dataset encompasses 7,377 CR, 9,463 DX, and 6,687 CT studies.

搜集汇总

背景与挑战

背景概述

BIMCV COVID-19数据集包含7377 CR、9463 DX和6687 CT研究，涵盖了COVID-19患者的胸部X光图像和CT图像，以及相关的放射学发现和诊断测试结果。数据集提供了丰富的标注信息和高分辨率图像，适用于COVID-19诊断和研究。