Radiology Objects in COntext (ROCO)
收藏资源简介:
ROCO数据集是一个大规模的医学和多模态成像数据集,包含从PubMed Central Open Access FTP镜像自动检测的非复合图像。每张图像都附有下载链接和标题,以及从图像标题中提取的关键词、相应的UMLS语义类型和UMLS概念唯一标识符。该数据集可用于构建图像标题生成模型、图像分类和标记的分类模型或基于内容的图像检索系统。
The ROCO dataset is a large-scale medical and multimodal imaging dataset, comprising non-compound images automatically detected from the PubMed Central Open Access FTP mirror. Each image is accompanied by a download link and a title, along with keywords extracted from the image title, corresponding UMLS semantic types, and UMLS concept unique identifiers. This dataset can be utilized to construct image caption generation models, classification and tagging models for image classification, or content-based image retrieval systems.
数据集概述
名称: Radiology Objects in COntext (ROCO)
类型: 大型医学和多模态影像数据集
来源: 来自PubMed Central Open Access FTP镜像的出版物图像
内容:
- 图像下载链接及其标题
- 图像标题中提取的关键词
- 对应的UMLS Semantic Types (SemTypes) 和 UMLS Concept Unique Identifiers (CUIs)
用途:
- 构建图像标题生成模型
- 图像分类和标记的分类模型
- 基于内容的图像检索系统
相关活动: 用作ImageCLEF 2019中概念检测任务的开发数据
引用:
- 引用文献: "Radiology Objects in COntext (ROCO): A Multimodal Image Dataset"
- 作者: O. Pelka, S. Koitka, J. Rückert, F. Nensa, C.M. Friedrich
- 发表于: MICCAI Workshop on Large-scale Annotation of Biomedical Data and Expert Label Synthesis (LABELS) 2018
- DOI: 10.1007/978-3-030-01364-6_20




