TopoStitcher: A Geometric–Topological Structure-Guided Stitching Framework for Single-Molecule Localization Microscopy

Panoramic super-resolution localization imaging is revolutionizing the multiscale analysis paradigm of cell biology by bridging nanoscale ultrastructure (e.g., neuronal synaptic) with macroscale architectures (e.g., intercellular communication). Although localization data preserve the original structural information, existing stitching methods based on such data still face critical limitations, including registration error caused by geometric-topological structural discontinuities at seam regions and computational inefficiency in processing large-scale datasets. To address these challenges, we propose TopoStitcher, a stitching framework guided by geometric and topological structures for single-molecule localization microscopy. This framework directly performs stitching based on reconstructed images at the seam regions and then converts image-based registration offsets into localization offsets. This strategy avoids point-to-point computation of massive data, significantly improving the computational efficiency. Additionally, TopoStitcher leverages the inherent geometric-topological structure of biological samples to guide the registration process, effectively mitigating registration errors caused by structural discontinuities. In this way, TopoStitcher achieves high stitching performance while preserving the original information on localization data. Experimental results demonstrate that TopoStitcher outperforms existing stitching methods, particularly in handling biological samples with geometric-topological structural discontinuities. The study provides a critical advancement for the development of super-resolution cell biology.