3D surface reconstruction of cellular cryo-soft X-ray microscopy tomograms using semisupervised deep learning

Abstract

Cryo-soft X-ray tomography (cryo-SXT) is a powerful method to investigate the ultrastructure of cells, offering resolution in the tens of nanometer range and strong contrast for membranous structures without requiring labeling or chemical fixation. The short acquisition time and the relatively large field of view leads to fast acquisition of large amounts of tomographic image data. Segmentation of these data into accessible features is a necessary step in gaining biologically relevant information from cryo-soft X-ray tomograms. However, manual image segmentation still requires several orders of magnitude more time than data acquisition. To address this challenge, we have here developed an end-to-end automated 3D segmentation pipeline based on semisupervised deep learning. Our approach is suitable for high-throughput analysis of large amounts of tomographic data, while being robust when faced with limited manual annotations and variations in the tomographic conditions. We validate our approach by extracting three-dimensional information on cellular ultrastructure and by quantifying nanoscopic morphological parameters of filopodia in mammalian cells.

Protein nanobarcodes enable single-step multiplexed fluorescence imaging

“…Our deep learning approach adds to a rapidly growing body of work in the imaging field. Similar deep learning methods, for example, image segmentation [45][46][47][48] and feature detection [49,50], are among the most sought-after applications in imaging. Other prominent applications include resolution enhancement [51][52][53] and increasing the signal-to-noise ratio [54].…”

Soft X-ray Microscopy in Cell Biology: Current Status, Contributions and Prospects

“…1 ); therefore, software packages for working with EM are suitable for reconstructing the volume from tilt series, for segmentation, and for subsequent data analysis [ 12 , 13 , 20 ]. Specialized tools are also being developed for working with cryo-SXT data, performing image restoration, and increasing their information yield [ 21 ], thus lightening the most operator-dependent steps: segmentation of three-dimensional data, as well as isolation of the contours and surfaces of organoids from the array of “voxels” [ 22 , 23 ].…”

Automated 3D cytoplasm segmentation in soft X-ray tomography

“… 8 While segmentation pipelines for light and electron microscopy are firmly established, automatic analysis of SXT data is limited. 9 …”

“… 16 implemented a trainable Weka segmentation machine learning tool, 17 accessible in Fiji. Furthermore, neural network-based algorithms, such as convolutional neural networks or U-Net, have been used to extract membranous organelles in the study by Dyhr et al., 9 Francis et al., 18 and Egebjerg et al. 19 A segmentation method based on the combination of 2D U-Nets was used to automatically segment whole β-cells anatomy.…”