Electron Tomography (ET) is an important method for studying cell ultrastructure in three-dimensional (3D) space. By combining cryo-electron tomography of frozen-hydrated samples (cryo-ET) and a sub-tomogram averaging approach, ET has recently reached sub-nanometer resolution, thereby realizing the capability for gaining direct insights into function and mechanism. To obtain a high-resolution 3D ET reconstruction, alignment and geometry determination of the ET tilt series are necessary. However, typical methods for determining geometry require human intervention, which is not only subjective and easily introduces errors, but is also labor intensive for high-throughput tomographic reconstructions. To overcome these problems, we have developed an automatic geometry-determination method, called AutoGDeterm. By taking advantage of the high-contrast re-projections of the Iterative Compressed-sensing Optimized Non-Uniform Fast Fourier Transform (NUFFT) reconstruction (ICON) and a series of numerical analysis methods, AutoGDeterm achieves high-precision fully automated geometry determination. Experimental results on simulated and resin-embedded datasets show that the accuracy of AutoGDeterm is high and comparable to that of the typical “manual positioning” method. We have made AutoGDeterm available as software, which can be freely downloaded from our website http://ear.ict.ac.cn.

Electron tomography (ET) plays an important role in studying in situ cell ultrastructure in three-dimensional space. Due to limited tilt angles, ET reconstruction always suffers from the “missing wedge” problem. With a validation procedure, iterative compressed-sensing optimized NUFFT reconstruction (ICON) demonstrates its power in the restoration of validated missing information for low SNR biological ET dataset. However, the huge computational demand has become a major problem for the application of ICON. In this work, we analyzed the framework of ICON and classified the operations of major steps of ICON reconstruction into three types. Accordingly, we designed parallel strategies and implemented them on graphics processing units (GPU) to generate a parallel program ICON-GPU. With high accuracy, ICON-GPU has a great acceleration compared to its CPU version, up to 83.7×, greatly relieving ICON’s dependence on computing resource.