AI-driven large-scale electron microscopy enables whole-tissue subcellular digitization

Abstract: The distribution and interactions of organelles critically mediate cellular physiology and pathology. Large-scale electron microscopy enables visualization of subcellular networks at the tissue level with nanometer resolution, but robust and efficient computational tools are lacking. Here, we present DeepOrganelle, a deep learning tool for high-throughput, cell-resolved mapping and digitization of organelle distribution and interactions in large-scale 2D/3D electron microscopy. When applied to spermatogenesis across 22 differentiation status of the germ cells, DeepOrganelle uncovered previously unrecognized, stage-dependent dynamics of membrane contact sites within one subphase during meiosis. It also revealed coordinated organelle redistribution in Sertoli cells towards the blood–testis barrier, digitizing the remodeling dynamics of the tissue. This study establishes DeepOrganelle as a powerful framework for capturing subcellular dynamics at the whole-tissue level.