Accurate, scalable and cross-platform cell identification for high-resolution spatial transcriptomics.

Recent advances in spatial transcriptomics (ST) have brought unprecedented insights into cellular diversity and cell-cell interactions within their spatial context. High-resolution ST techniques, including barcoding-based and imaging-based platforms, have achieved remarkable subcellular resolution. However, precise cell segmentation remains a major challenge, hampering effective single-cell spatial analysis. Existing methods are often platform specific and lack scalability for datasets with large fields of view. Here we introduce Cellist, a new, multi-modal, cell-segmentation method that combines image and expression information, enabling comprehensive cell-level analyses. Applied to mouse brain Stereo-seq data, Cellist improves within-cell transcriptomic coherence compared to existing approaches. It further enhances spatial domain identification and cell-type annotation. Importantly, Cellist is compatible with various ST techniques including Seq-Scope, seqFISH+, STARmap and 10x Xenium, exhibiting robust performance and high computational efficiency across diverse ST platforms and biological systems. Finally, application to post-neoadjuvant immunotherapy, nonsmall cell lung-cancer samples reveals the spatial heterogeneity of tumor clones and identifies therapy response-related myeloid subtypes and structures. These findings highlight the potential of Cellist in enhancing the power of high-resolution ST techniques for characterizing intricate tissue architectures. Cellist is publicly available at https://github.com/wanglabtongji/Cellist .