Principal component analysis for three-dimensional structured illumination microscopy (PCA-3DSIM)

Abstract

Three-dimensional structured illumination microscopy (3DSIM) is an essential super-resolution imaging technique for visualizing volumetric subcellular structures at the nanoscale, capable of doubling both lateral and axial resolution beyond the diffraction limit. However, high-quality 3DSIM reconstruction is often hindered by uncertainties in experimental parameters, such as optical aberrations and fluorescence density heterogeneity. Here, we present PCA-3DSIM, a novel 3DSIM reconstruction framework that extends principal component analysis (PCA) from two-dimensional (2D) to three-dimensional (3D) super-resolution microscopy. To further compensate spatial nonuniformities of illumination parameters, PCA-3DSIM can be implemented in an adaptive tiled-block manner. By segmenting raw volumetric data into localized subsets, PCA-3DSIM enables accurate parameter estimation and effective interference rejection for high-fidelity, artifact-free 3D super-resolution reconstruction, with the inherent efficiency of PCA supporting the tiled reconstruction with limited computational burden. Experimental results demonstrate that PCA-3DSIM provides reliable reconstruction performance and improved robustness across diverse imaging scenarios, from custom-built platforms to commercial systems. These results establish PCA-3DSIM as a flexible and practical tool for super-resolved volumetric imaging of subcellular structures, with broad potential applications in biomedical research.

This article developed PCA-3DSIM, a mathematically grounded enhancement to 3D structured illumination microscopy that improves robustness by integrating physical modeling with statistical analysis.