PathVQA
收藏资源简介:
是否有可能培养“AI病理学家”通过美国病理学委员会的董事会认证考试?为了实现这一目标,我们构建了一个医学视觉问答(VQA)数据集,其中AI代理将获得病理图像和问题,并被要求给出正确答案。该数据集包含来自 32,799 张病理学图像的 4,998 个开放式问题。 该数据集由两本公开可用的病理学教科书构建:“病理学教科书”和“基础病理学”,以及一个公开的数字图书馆:病理学教育信息资源 (PEIR)。 图像和说明的版权属于这两本书的出版商和作者,以及 PEIR 数字图书馆的所有者。
Is it feasible to train an "AI pathologist" to pass the board certification examination administered by the American Board of Pathology? To this end, we constructed a medical visual question answering (VQA) dataset, where the AI Agent is provided with pathological images and corresponding questions, and tasked with generating accurate answers. This dataset comprises 4,998 open-ended questions sourced from 32,799 pathological images. This dataset is compiled from two publicly available pathology textbooks: *Textbook of Pathology* and *Basic Pathology*, as well as a public digital library, the Pathology Education Informational Resource (PEIR). The copyrights of the images and their accompanying descriptions belong to the publishers and authors of these two textbooks, as well as the owners of the PEIR digital library.

- PathVQA数据集首次发表,专注于病理图像上的视觉问答任务,为医学图像分析领域提供了新的研究方向。
- PathVQA数据集首次应用于多项研究,展示了其在病理图像理解和诊断辅助中的潜力。
- PathVQA数据集被广泛用于国际会议和期刊,推动了医学图像处理技术的进步。
- 1PathVQA: 30000+ Questions for Medical Visual Question AnsweringUniversity of Warwick, University of California, San Diego · 2020年
- 2Medical Visual Question Answering: A SurveyUniversity of Science and Technology of China · 2021年
- 3A Comprehensive Review of Visual Question Answering: Datasets, Algorithms, and ChallengesUniversity of Electronic Science and Technology of China · 2021年
- 4A Survey on Visual Question Answering: Datasets and ApproachesUniversity of California, Los Angeles · 2020年
- 5Visual Question Answering: A Survey of Methods and DatasetsUniversity of California, Los Angeles · 2020年



