kidney WSI patch embeddings

The dataset contains embeddings extracted from PAS-stained kidney whole slide images. The foreground of the whole slide image data is extracted by U-Net, a convolutional neural network model for medical image segmentation. Then the foreground are cut into  patches of the size 224*224, and  the patches are converted to embeddings with  the dimension of 1*768 by CtransPath, a self-supervised learning model for histopathology image data. The label form records the survival days of each whole slide image, indexed by the name of the whole slide image.