Applying a Modified Wavelet Shrinkage Filter to Improve Cryo-Electron Microscopy Imaging.

Abstract

Cryo-electron microscopy (Cryo-EM) imaging has the unique potential to bridge the gap between cellular and molecular biology by revealing the structures of large macromolecular assemblies and cellular complexes. Therefore, cryo-EM three-dimensional (3D) reconstruction has been rapidly developed in recent several years and applied widely in life science research; however, it suffers from reduced contrast and low signal-to-noise ratios with a high degree of noise under low electron dose conditions, resulting in failures of many conventional filters. In this article, we explored a modified wavelet shrinkage filter (with optimal wavelet parameters: three-level decomposition, level-1 zeroed out, subband-dependent threshold, soft thresholding, and spline-based discrete dyadic wavelet transform) and extended its application in the cryo-EM field in two aspects: single-particle analysis and cryo-electron tomography. Its performance was assessed with simulation data and real cryo-EM experimental data. Compared with the undenoised results and conventional denoising techniques (e.g., Gaussian, median, and bilateral filters), the modified wavelet shrinkage filter maintained the resolution and contrast but reduced the noise, leading to higher quality images and more accurate measures of the biological structure. We expect that our study can provide benefits to cryo-EM applications: 3D reconstruction, visualization, structural analysis, and interpretation. All these data and programs are available.