MEDICAT
收藏资源简介:
MEDICAT是由艾伦人工智能研究所创建的一个包含217,060张医学图像的数据集,来源于131,410篇开放获取的生物医学论文。该数据集不仅包括图像,还包含图像的标题、内联引用以及手动标注的子图像和子标题,其中75%的图像为复合图像。MEDICAT的创建旨在解决医学图像检索和图像与文本对齐的挑战,特别关注子图像与子标题的匹配问题。数据集的应用领域包括医学图像分析和学术搜索引擎的图像检索功能,旨在提高医学图像与文本关系的理解和应用。
MEDICAT is a dataset developed by the Allen Institute for Artificial Intelligence, which contains 217,060 medical images sourced from 131,410 open-access biomedical papers. Beyond the raw images, the dataset also provides image captions, inline citations, as well as manually annotated sub-images and sub-captions, with 75% of the images being composite images. MEDICAT was created to address the challenges of medical image retrieval and image-text alignment, with a particular focus on the matching task between sub-images and their corresponding sub-captions. Its application areas include medical image analysis and image retrieval functions for academic search engines, aiming to improve the understanding and practical application of the relationships between medical images and text.

- 1MedICaT: A Dataset of Medical Images, Captions, and Textual References艾伦人工智能研究所 · 2020年



