owkin/plism-dataset-tiles

PLISM数据集是一个用于评估AI模型在机构间领域迁移中鲁棒性的病理图像数据集。它包含了东京大学医院1955年至2018年间诊断和手术的患者组织样本的数字化切片，这些切片在不同的扫描仪和13种H&E染色条件下进行处理。数据集中的每个样本包含来自各种器官的46个组织微阵列，总共包含了310,947个图像块，用于图像特征提取和图像分类任务。

The PLISM (Pathology Images of Scanners and Mobilephones) dataset is a collection of histopathological specimens created to evaluate the robustness of AI models against inter-institutional domain shifts. It includes digitized slides from patients diagnosed and operated at the University of Tokyo Hospital between 1955 and 2018, processed under different scanners and 13 H&E staining conditions. Each sample in the dataset encompasses 46 Tissue Micro Arrays (TMA) from various organs, with a total of 310,947 image tiles for image feature extraction and classification tasks.