FOCUS-3D: Robust, generalizable volumetric cell segmentation for three-dimensional fluorescence microscopy

FOCUS-3D is a cutting-edge volumetric cell segmentation framework designed for three-dimensional fluorescence microscopy. This innovative tool addresses the challenge of extracting quantitative cellular insights from complex volumetric datasets captured by advanced microscopy techniques. By leveraging a comprehensive annotated cell resource and state-of-the-art artificial intelligence designs, FOCUS-3D demonstrates remarkable versatility and robustness in handling diverse species, tissues, fluorescent markers, and imaging modalities.

One of the key strengths of FOCUS-3D lies in its ability to uncover distinct phases of cellular morphogenesis during zebrafish (Danio rerio) development. The framework successfully differentiates early motility-driven rearrangements from subsequent cell shape remodeling and tissue repacking processes. Furthermore, it establishes a connection between these morphological states and corresponding spatial and developmental transcriptional programs across multiple datasets.

FOCUS-3D's multi-faceted approach, which includes volumetric representation learning, multi-scale feature extraction, and query-based mask prediction, sets it apart from previous segmentation methods. By integrating these advanced AI techniques, the framework achieves state-of-the-art performance and provides researchers with a powerful tool to unravel the intricate spatial organization of cells within tissues.