Statistical survey of open source medical image databases on the Internet
Statistical survey of open source medical image databases on the Internet
复制标题
DOI：
10.4103/digm.digm_1_19
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Hongkai Wang;Xinlei Ma;Haoyu Zhai;Yuhao Liao;Yi Wu;Na Chen;Shaoxiang Zhang;Bin Zhang
中科院分区：
文献类型：
--
作者：
Hongkai Wang;Xinlei Ma;Haoyu Zhai;Yuhao Liao;Yi Wu;Na Chen;Shaoxiang Zhang;Bin Zhang
背景和目标：如今，越来越多的开源医学影像数据库发布在互联网上，用于医学教学，算法开发和医学研究。然而，仍然缺乏对这些数据库的统计调查。在这项调查中，我们总结了互联网上的开源医学图像数据库的现状。其目的是让每个人都能更容易地找到和使用开源医学图像数据。方法：通过检索科学论文和互联网搜索引擎收集有关公共医学图像数据库的信息。根据收集的信息，对不同疾病、身体部位、成像模式和国家的数据库数量和图像数量进行了统计。结果：癌症，特别是乳腺癌和肺癌，在所有疾病中的数据库数量排名第一。就数据库数量而言，乳房、大脑、肺部和胸部是排名前四的身体部位。计算机断层扫描、磁共振成像和X射线是开源数据集中最常见的成像方式。美国和荷兰是拥有数据库最多的两个国家。结论：疾病和身体部位的排名与一个国家的疾病发病率和卫生保健费用密切相关。医学影像数据库的开放源的数量仍在增长，有必要在未来几年对其存在状态进行持续的统计研究。在互联网上打开所有收集的数据库列表（https：//docs.qq.com/sheet/DQWF0QlZKVHpHU1Za）。
Background and Objectives: Nowadays, more and more open source medical imaging databases are published on the Internet for medical teaching, algorithm development, and medical research. However, a statistical survey of these databases is still lacking. In this survey, we summarize the current status of open source medical image databases on the Internet. The aim is to make it easier for everyone to find and use open source medical image data. Methods: Information about publicly available medical image databases was collected by searching for scientific papers and Internet search engines. Based on the collected information, the number of databases and the number of images were counted for different diseases, body parts, imaging modalities, and countries. Results: Cancer, particularly breast cancer and lung cancer, ranked top in database numbers among all diseases. The breast, brain, lung, and chest are the top four body parts in terms of database numbers. Computed tomography, magnetic resonance imaging, and X-ray are the most common imaging modalities in the open source datasets. The USA and the Netherlands are the top two countries who own the most databases. Conclusions: The rankings for diseases and body parts were closely related to the diseases morbidity and the health-care expenditure of a country. The number of open sources of medical imaging databases is still growing; there is a need for continuous statistical research on their existence status in the coming years. The list of all the collected databases is opened on the Internet (https://docs.qq.com/sheet/DQWF0QlZKVHpHU1Za).