PANORAMA2024

PANORAMA 研究主要由两个子研究组成：AI 研究——一个带注释的数据集，由回顾性、多中心、富含癌症的常规上腹部 CECT 扫描组成，向所有参与团队和整个研究界公开。该数据集旨在通过包括 PDAC 和非 PDAC 患者（健康胰腺患者和非 PDAC 胰腺病变患者）来代表临床现实。阅片研究——PANORAMA 阅片研究将包括来自多个机构的约 400 个案例，这些案例将由 40 多名具有不同专业水平的国际读者小组阅读。最终，PANORAMA 旨在对参与阅片研究的腹部放射科医生在大挑战中开发的最先进的 AI 算法进行基准测试——以评估现代胰腺 AI 解决方案在 CECT 中的 PDAC 检测和诊断方面的临床可行性。The PANORAMA study primarily comprises two sub-studies: the AI Study, an annotated dataset composed of retrospective, multi-center, cancer-enriched routine upper abdominal contrast-enhanced computed tomography (CECT) scans, which is publicly available to all participating teams and the entire research community. This dataset is designed to reflect real-world clinical scenarios by enrolling patients with PDAC, non-PDAC pancreatic conditions (including those with healthy pancreases and patients with non-PDAC pancreatic lesions). The PANORAMA Image Reading Study will include approximately 400 cases from multiple institutions, which will be reviewed by a panel of over 40 international readers with varying levels of professional expertise. Ultimately, the PANORAMA study aims to benchmark the state-of-the-art AI algorithms developed for the grand challenge by abdominal radiologists participating in the reading study, in order to evaluate the clinical feasibility of modern pancreatic AI solutions for PDAC detection and diagnosis using CECT scans.

搜集汇总

背景与挑战

背景概述

该数据集是PANORAMA挑战赛的公共训练和开发数据集第一部分，包含来自荷兰两个医疗中心的2238例匿名增强CT扫描，以及来自其他公开数据集的额外病例，专注于胰腺导管腺癌的检测和诊断研究。数据集旨在用于验证人工智能算法与放射科医生在胰腺癌诊断性能上的对比，以支持临床转化研究，文件总大小为48.2 GB。