Electron tomography (ET) is a combination of electron microscopy technique and tomography technique to study three-dimensional (3D) structure of subcellular macromolecular objects at a nanometer resolution. ET generates 3D images from a series of two-dimensional (2D) projections at different orientations, obtained by tilting specimens in small angular steps around axes. In the process of 3D reconstruction of ET, aligning the series of tilted images and computing the 3D reconstruction are the key steps to attain the high-quality reconstructed results.
Reconstruction of a 3D volume relies on accurate alignment of the tilt series of 2D projections. The commonly used alignment methods are divided into two categories: fiducial (or marker) and marker-free alignment. The vast amount of tomographic data alignment is done using marker alignment and excellent softwares for marker alignment are useful packages presently available, such as IMOD and Inspect3D (FEI). It is not, however, possible to have gold markers in every sample, and the gold markers may be poorly distributed so as to hide the information of sample structures. Alternatively, marker free alignment adopts a cross-correlation technique that tracks similar image features that are present in several micrographs and explores them implicitly as substitutes for markers. The marker-free method is performed in some ET softwares, e.g. protomo. However, cross-correlation is highly sensitive to changes such as viewpoint change or non-rigid deformations which are just the changes between ET images.
Due to physical limitations of microscopes, the angular tilt range is limited and, as a result, tomographic tilt series have a wedge of missing data corresponding to the limited angular range. Furthermore, in order to obtain high resolution reconstructions, specimens are imaged at very low electron doses, which makes EM images extremely noisy (signal-to-noise ratio in the order of 0.1). As a consequence, high-resolution electron tomography requires a method of “3D reconstruction from projections” able to deal with limited angle conditions and extremely low signal-to-noise ratios of the projection images. Weighted backprojection (WBP) has been one of the most popular methods in the field of 3D reconstruction of ET, due to its algorithmic simplicity and computational efficiency. The major disadvantage of WBP, however, is that the results may be strongly affected by limited-angle data and noisy conditions. Iterative methods, such as algebraic reconstruction techniques (ART), simultaneous iterative reconstruction technique (SIRT), component averaging methods (CAV), block-iterative CAV (BICAV), and simultaneous algebraic reconstruction technique (SART), constitute one of the main alternatives to WBP in 3D reconstruction of ET, owing to their good performance in handling incomplete, noisy data. However, these traditional iterative methods are computationally expensive in order to acquire satisfactory reconstructed results.
Furthermore, 3D reconstruction of ET demands huge computational costs and resources that derive from the computational complexity of the reconstruction algorithms and the size and number of the projection images involved. Traditionally, high-performance computing has been used to address such computational requirements by means of parallel computing on supercomputers, large computer clusters and multicore computers. Recently, graphics processing units (GPUs) offer an attractive alternative platform to cope with the demands in ET in terms of the high peak performance, cost effectiveness, and the availability of user-friendly programming environments, e.g. NVIDIA CUDA. Several advanced GPU acceleration frameworks have been proposed to allow 3D ET reconstruction to be performed on the order of minutes.
Various software for ET image alignment, three-dimensional (3D) reconstruction, and 2D/3D image processing or for one of the mentioned steps have been developed (first IMOD, then TOM, EM3D, UCSF tomography and TomoJ). The major limitation of these freeware is that they lack the accurate marker free alignment methods and high performance reconstruction algorithms and have not effectively implemented them both. Here, we developed new software for electron tomography named AuTom that provides an integrated scheme for ET image alignment, reconstruction, and processing. The main merits of AuTom lie in three aspects: first, an accurate marker-free ET image alignment method that employs scale invariant feature transform (SIFT) feature-extraction and robust parameter estimation techniques; second, high performance iterative reconstruction methods–such as ART, SIRT and iterative methods like adaptive simultaneous algebraic reconstruction technique (ASART); finally, efficient parallel 3D reconstruction using iterative methods on GPU platform. As a result, AuTom is an open source software that is specifically designed for ET to meet the requirements of marker-free alignment, high-performance iterative reconstruction methods and huge computational costs. AuTom can run on any platform that supports Qt and CUDA libraries. It has a friendly graphic user interface along with detailed manuals.