MedPromptX-VQA
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/BioMedIA-MBZUAI/MedPromptX
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个全新的视觉问题回答数据集，它将来自MIMIC-IV和MIMIC-CXR数据库的交错图像和电子健康记录（EHR）数据融合在一起。该数据集通过将图像与结构化的临床数据相结合，旨在提高胸部X光诊断的准确性，并已显示出在模型性能上的显著提升。其所涉及的任务是视觉问题回答。
This dataset is a novel visual question answering (VQA) dataset that integrates interleaved image and electronic health record (EHR) data sourced from the MIMIC-IV and MIMIC-CXR databases. By combining images with structured clinical data, this dataset aims to improve the accuracy of chest X-ray diagnosis and has demonstrated significant improvements in model performance. The task involved in this dataset is visual question answering (VQA).
提供机构：
MIMIC-IV and MIMIC-CXR
