Improvements on marker-free images alignment for electron tomography

Abstract

Electron tomography is a technique to obtain three-dimensional structural information of samples. However, the technique is limited by shifts occurring during acquisition that need to be corrected before the reconstruction process. In 2009, we proposed an approach for post-acquisition alignment of tilt series images. This approach was marker-free, based on patch tracking and integrated in free software. Here, we present improvements to the method to make it more reliable, stable and accurate. In addition, we modified the image formation model underlying the alignment procedure to include different deformations occurring during acquisition. We propose a new way to correct these computed deformations to obtain reconstructions with reduced artifacts. The new approach has demonstrated to improve the quality of the final 3D reconstruction, giving access to better defined structures for different transmission electron tomography methods: resin embedded STEM-tomography and cryo-TEM tomography. The method is freely available in TomoJ software.

Keywords: Electron tomography; Image processing; Tilt series alignment.

Graphical abstract

Fig. 1

Comparison of the reconstructions of the phantom with the different method of correction of deformation. Each image corresponds to the sum of XZ planes to better see the artifacts: (A) no correction of deformations, (B) correction using 2D affine transforms, (C) correction using 2D non-linear mapping, (D) optimal reconstruction if no deformation exists, (E) correction using 3D projector integrated in reconstruction, (F) correction using 3D projector integrated in reconstruction with deformations computed from landmarks. White arrows point to features. Near the central plane (thin arrows), the feature are well reconstructed with all methods of correction of deformation. The features above (dotted arrows) or far (large arrows) from central plane display correction of the in-plane deformation, but show triangular shapes unless the 3D projector is employed.

Fig. 2

Visualization of landmarks chains found with the gold bead-like optimization. A) The landmarks found are displayed with green circles on the projection image near $0°$. B) zoom on a region with gold bead. C) zoom on a region with no gold bead, but sufficiently spherically symmetric, distinctive features.

Fig. 3

Comparison of reconstructions obtained with different alignment models. The first column corresponds to alignment using our old alignment model (Sorzano et al., 2009), the second column displays the result of the new model with only shifts and in-plane rotation (this experiment shows the improvements not due to the 3D deformation model), and third column shows the result of the new model with deformation correction. The top row corresponds to the central XY plane. The second and third rows correspond to the XZ plane number 49 and 315, respectively. White arrows points to features with visible change between algorithms. Black arrows points the plastic of holey grid where change in shape is visible. The last column corresponds to the comparison of a feature with visible differences between the three reconstructions. The scale bar corresponds to 500 nm.

Fig. 4

Comparison of the methods of application of alignment with deformations. The top row corresponds to the central XY plane. The second and thirds row correspond to the XZ plane number 49 and 315, respectively. The first column corresponds to method 1 with linear mapping. The second column corresponds to method 2 with non-linear mapping. The third column corresponds to method 3 with application during reconstruction process. White arrows point to features with visible change between algorithms. The last column corresponds to the comparison of a feature with visible differences between the three methods of application of alignment with deformations.Scale bar corresponds to 500 nm.

Fig. 5

Comparison of the reconstructions of Trypanosoma Brucei cilia obtained with or without deformation. A and C: Central slice in XY and XZ direction respectively from the reconstruction with deformation correction, B and D: Central slice in XY and XZ direction respectively from the reconstruction without correction of deformations. a, b, c, d, e zoomed-in region corresponding to the areas delimited by white squares on central slices from reconstruction with deformation correction. a’, b’, c’, d’, e’ zoomed-in region from reconstruction without correction of deformations corresponding to the same areas as a, b, c, d, e respectively (delimited by white squares on central slices). White arrows points to details improved in the reconstruction with deformation correction. The scale bar corresponds to 200 nm.

Fig. 6

Reconstruction of bacteriophage T5. A) plane from the reconstruction. B) to D) zoom in virus either filled with DNA (B and C) or empty except for a brand of DNA going through the tail (D, white arrow). E) to G) zoom in tail extremity. When viruses are filled with DNA the tail fibers are clearly identifiable in E) and F): central fiber is pointed with a white solid arrow, and lateral with a dotted arrow. When viruses are empty, the central fiber is lost and the tail is open. H) zoom on tail, I) average of a tail, and I’) its Fourier transform: spots correspond to the helical structure and the pitch angle (green lines) presents a value of 38°. J) and J’) gold bead view in XZ plane. Scale bar corresponds to 200 nm.