Fast and robust drift correction for single-molecule localization microscopy

Abstract

Abstract Owing to its gradual accumulation of molecular positions, single-molecule localization microscopy (SMLM) depends on the proper correction of sample drifts that occur during data acquisition. However, current data-based drift-correction approaches for SMLM are often unreliable and time-consuming, limiting the achieved resolution and throughput. Here we report nearest paired cloud (NP-Cloud), a fast and robust SMLM drift-correction method. By pairing the nearest molecules in SMLM data segments and calculating their displacements within a small search radius, NP-Cloud efficiently utilizes the continuously valued positions of each super-localized molecule while drastically reducing the computational cost. With both simulated and experimental SMLM data, we thus demonstrate substantially improved robustness and fidelity for drift correction in three dimensions, as well as speeds >100-fold faster over traditional single-referenced approaches and >104 faster over traditional cross-referenced redundant approaches. Excellent drift corrections are achieved for diverse samples within seconds. We thus provide a robust, fast, and practical solution to SMLM drift correction.