Deep Learning Virtual Staining Dataset: Breast, Mouse, and Organoids

This dataset contains comprehensive image data for training and evaluating deep learning models dedicated to virtual staining of histopathological images. The goal is to generate H&E-like pathology images from autofluorescence (AF) or multi-channel fluorescence images. It covers multiple tissue types including human breast tissue, mouse tumors, and organoids, facilitating research in label-free imaging and computational pathology.

Most processed images (patches and evaluation sets) are in standard TIF format.  The files inside Initial_data.zip are in 3DHISTECH Mirax proprietary format (.mrxs). To open MRXS files: Please use 3DHISTECH SlideViewer (formerly CaseViewer) or the open-source software QuPath.