GEMeX
收藏资源简介:
GEMeX是一个大规模、可解释的医学视觉问答(VQA)基准数据集,专门用于胸部X光诊断。该数据集由香港理工大学等机构创建,包含151,025张放射图像和1,605,575个问题-答案对,涵盖开放式、封闭式、单选和多选四种问题类型。数据集通过与放射科医生合作,对Chest ImaGenome数据集进行精细处理,利用GPT-4o生成多样化的问题,并提供详细的视觉和文本解释。GEMeX旨在解决现有医学VQA数据集缺乏解释性和多样性问题,适用于医疗诊断、医学教育和临床决策支持等领域。
GEMeX is a large-scale, interpretable medical visual question answering (VQA) benchmark dataset specifically designed for chest X-ray diagnosis. Developed by institutions including The Hong Kong Polytechnic University, this dataset contains 151,025 radiological images and 1,605,575 question-answer pairs, covering four question types: open-ended, closed-ended, single-choice, and multiple-choice. Developed through fine-grained processing of the Chest ImaGenome dataset in collaboration with radiologists, this dataset leverages GPT-4o to generate diverse questions and provides detailed visual and textual explanations. GEMeX aims to address the lack of interpretability and diversity in existing medical VQA datasets, and is applicable to fields such as medical diagnosis, medical education, and clinical decision support.

- 1GEMeX: A Large-Scale, Groundable, and Explainable Medical VQA Benchmark for Chest X-ray Diagnosis香港理工大学 · 2024年



