INSPECT
收藏资源简介:
INSPECT数据集是由斯坦福大学创建的一个大规模多模态医学数据集,专注于肺栓塞的诊断与预后评估。该数据集包含19,402名患者的结构化纵向电子健康记录(EHR),包括诊断/程序代码、实验室检查、药物、生命体征和人口统计信息,以及与这些记录对应的23,248次CT扫描及其相应的放射学报告印象部分。INSPECT数据集的创建旨在解决现有医学数据集在多模态数据整合方面的不足,特别是在3D医学影像与EHR数据结合方面的缺失。通过提供丰富的多模态数据,INSPECT数据集支持开发和评估用于肺栓塞诊断和预后的新型多模态融合方法,从而推动AI在医学诊断、预后和治疗规划中的应用。
The INSPECT dataset is a large-scale multimodal medical dataset developed by Stanford University, focusing on the diagnosis and prognostic assessment of pulmonary embolism. This dataset contains structured longitudinal electronic health records (EHRs) from 19,402 patients, including diagnostic/procedural codes, laboratory tests, medications, vital signs, and demographic information, along with 23,248 CT scans corresponding to these records and their corresponding impression sections from radiology reports. The development of the INSPECT dataset aims to address the gaps in existing medical datasets regarding multimodal data integration, particularly the absence of integrated 3D medical imaging and EHR data. By providing rich multimodal data, the INSPECT dataset supports the development and evaluation of novel multimodal fusion approaches for pulmonary embolism diagnosis and prognosis, thereby advancing the application of AI in medical diagnosis, prognosis, and treatment planning.

- 1INSPECT: A Multimodal Dataset for Pulmonary Embolism Diagnosis and Prognosis斯坦福大学 · 2023年



