CSRefiner: A lightweight framework for fine-tuning cell segmentation models with small datasets

This dataset release includes two demonstration collections from mouse brain tissue sections generated with the STOmics Stereo-seq platform: FFPE DAPI-stained sections and fresh frozen H&E-stained sections, and a dataset of fresh frozen H&E-stained mouse lung slices from a 10x VisiumHD platform. Each dataset contains whole-slide images, cell-level expression matrices, and manually curated annotations. These resources were produced for the CSRefiner study, a lightweight framework for fine-tuning cell segmentation models with small datasets. They are intended to support benchmarking, method development, and downstream analysis in spatial transcriptomics.