NuCorpus-15M: A Large-Scale DAPI-Stained Cell Nucleus Image Dataset-Part 1

Dataset Description NuCorpus-15M is a large-scale dataset specifically created for nuclear morphology analysis in biomedical research. It consists of 15.52 million high-quality DAPI-stained cell nucleus images, representing a diverse range of human and mouse tissues, both diseased and non-diseased. The dataset spans 15 different organs or tissue types and has been meticulously curated to ensure high consistency and quality for various downstream tasks. Dataset Details Total Images: 15.52 million Species: Human (78.9%) and Mouse (21.1%) Tissues Covered: Breast, colon, brain, tonsil, and more. Data Source: Images are derived from 56 standardized high-resolution tissue sections provided by 10x Genomics' Xenium platform. Sub-image size: 112x112 pixels, each centered around a cell nucleus. File format: HDF5 for both images and metadata, facilitating efficient large-scale data handling. Preprocessing: All images in NuCorpus-15M have been pre-processed for grayscale conversion and Normalization. Data Distribution: The dataset is divided into 20 parts, and this is the first batch. Dataset Structure Each entry in the dataset consists of a DAPI-stained image and its corresponding metadata, stored in HDF5 format to facilitate efficient access. The images have been pre-processed to normalize fluorescence intensity and remove suboptimal images.