Cryogenic electron tomography (cryoET) allows visualization of cellular structures in situ. However, anisotropic resolution arising from the intrinsic “missing-wedge” problem has presented major challenges in visualization and interpretation of tomograms. We developed IsoNet (Isotropic Reconstruction of Electron Tomography), a deep learning-based software package that iteratively reconstructs missing-wedge information and increases signal-to-noise ratio, using knowledge learned from raw tomograms. Without sub-tomogram averaging, IsoNet generates tomograms with significantly reduced resolution anisotropy and enables functional interpretation. Applications of IsoNet to cryoET data demonstrate greatly improved structural interpretability and identification of differently oriented complexes, as tested on tomograms of Saccharomyces cerevisiae 80S ribosomes. We also devise a novel mathematical method to quantify the reliability of recovered missing-wedge information. Such quantification provides a measure of the recovery quality of IsoNet and informs optimization of parameters.
Back to Computational Microscopy