ReXVQA

该数据集名为ReXVQA，是胸部放射学领域内视觉问答（VQA）的最大基准测试集，包含了大约69.6万个问题与16万个胸部X光研究案例的配对。该数据集覆盖了反映放射学核心推理技能的多样化临床任务，并采用多选题的形式设计，以体现医学教育实践中的放射学推理核心。其规模约为69.6万个问题和16万个胸部X光研究案例，旨在评估胸部放射学中的视觉问答任务（Vqa）。

The dataset named ReXVQA is the largest benchmark dataset for visual question answering (VQA) in the field of chest radiology. It comprises approximately 696,000 paired question-chest X-ray study instances, corresponding to 160,000 unique chest X-ray studies. The dataset covers a diverse set of clinical tasks that reflect the core reasoning skills of radiology, and is structured in a multiple-choice format to embody the core of radiological reasoning in medical education practice. With a corpus of 696,000 questions and 160,000 chest X-ray study cases, this dataset is designed to evaluate visual question answering (VQA) tasks in chest radiology.