swiss-ai/patho-ssl-data-curation

本研究提供了数字病理学中用于视觉基础模型的自动数据筛选的数据集。数据集包含了3500万个瓦片的层次聚类结果，以及用于可视化的瓦片图像和相关的元数据。这些数据可以帮助研究人员在瓦片级别上进行无监督的数据筛选，以改善视觉基础模型在下游临床任务中的表现。

This study provides a dataset for automatic data curation for vision foundation models in digital pathology. The dataset includes hierarchical clustering results for 35 million tiles, as well as tile images and associated metadata for visualization. These data can help researchers perform unsupervised data curation at the tile level to improve the performance of vision foundation models on downstream clinical tasks.