Self-supervised denoising for structured illumination microscopy enables long-term super-resolution live-cell imaging

Abstract

Detection noise significantly degrades the quality of structured illumination microscopy (SIM) images, especially under low-light conditions. Although supervised learning based denoising methods have shown prominent advances in eliminating the noise-induced artifacts, the requirement of a large amount of high-quality training data severely limits their applications. Here we developed a pixel-realignment-based self-supervised denoising framework for SIM (PRS-SIM) that trains an SIM image denoiser with only noisy data and substantially removes the reconstruction artifacts. We demonstrated that PRS-SIM generates artifact-free images with 10-fold less fluorescence than ordinary imaging conditions while achieving comparable super-resolution capability to the ground truth (GT). Moreover, the proposed method is compatible with multiple SIM modalities such as total internal reflective fluorescence SIM (TIRF-SIM), three-dimensional SIM (3D-SIM), lattice light-sheet SIM (LLS-SIM), and non-linear SIM (NL-SIM). With PRS-SIM, we achieved long-term super-resolution live-cell imaging of various bioprocesses, revealing the clustered distribution of clathrin coated pits and detailed interaction dynamics of multiple organelles and the cytoskeleton.