Biological serial block face scanning electron microscopy at improved z-resolution based on Monte Carlo model

Abstract

Serial block-face electron microscopy (SBEM) provides nanoscale 3D ultrastructure of embedded and stained cells and tissues in volumes of up to 10⁷ µm³. In SBEM, electrons with 1-3 keV energies are incident on a specimen block, from which backscattered electron (BSE) images are collected with x, y resolution of 5-10 nm in the block-face plane, and successive layers are removed by an in situ ultramicrotome. Spatial resolution along the z-direction, however, is limited to around 25 nm by the minimum cutting thickness. To improve the z-resolution, we have extracted depth information from BSE images acquired at dual primary beam energies, using Monte Carlo simulations of electron scattering. The relationship between depth of stain and ratio of dual-energy BSE intensities enables us to determine 3D structure with a ×2 improvement in z-resolution. We demonstrate the technique by sub-slice imaging of hepatocyte membranes in liver tissue.

Figure 1

Diagram of sub-slice imaging in the SBEM. (a) Surface of the sample block is imaged by scanning a focused electron probe from a field emission source across a region of interest and collecting backscattered electrons (BSEs) with an annular detector. Sample block is raised by height ≥ 25 nm, and a section of that thickness is shaved off using a diamond knife mounted in the SBEM’s in situ microtome. The newly exposed surface is then re-imaged. This process is repeated until an image stack is collected from the desired sample volume, 70 slices in this illustration. (b,c) Series of backscattered electron images captured at energy E₁ and E₂, respectively. (d,e) 3D reconstruction of single-energy image series from (b,c), respectively. (f) Dual-energy reconstruction gives sub-slice resolution.

Figure 2

Monte Carlo simulation model and results. (a) Three-dimensional geometrical model for 50 nm × 50 nm × 12.5 nm cuboids stained with 3% lead located at different depths in an 800 nm × 800 nm × 800 nm epoxy block, together with views in the x-z plane, y-z plane and x-y plane. Centers of cuboids are located 6.25 nm, 18.75 nm, 31.25 nm and 43.75 nm, respectively, from the top surface of the block from left to right. Dimension in z is not drawn to scale. (b) Simulated backscattered images at primary beam energies from 0.8 keV to 2.2 keV with an interval of 0.1 keV. (c) Corresponding intensity profiles for cuboids shown in (b). These plots provide a guide of primary voltage selection. Scale bar, 50 nm.

Figure 3

Effect of stain composition on accuracy of the methods. (a) Three-dimensional geometrical model for two 625 nm × 625 nm × 12.5 nm cuboids (green) stained with 3% lead and containing 2,500 voxels of dimensions 12.5 nm × 12.5 nm × 12.5 nm, and one 625 nm × 625 nm × 25 nm cuboid (yellow) stained with 1.5% lead and containing 5,000 voxels of the same dimension, located at different depths in a 10 µm × 10 µm × 10 µm epoxy block, with views in the x-z, y-z and x-y planes. Centers of cuboids are located at 6.25 nm, 18.75 nm and 12.5 nm, respectively, from the top surface of the block (left to right). Dimension in z is not drawn to scale. (b) Simulated images at primary beam energies of 1.0 keV and 1.4 keV; and (c) calculated sub-slice specimen structures with a nominal z-resolution of 12.5 nm. (d) Line profile of the features in the top and bottom sub-slices are indicated by dashed lines in (c). (e) Histograms for cuboids 1 to 3 and their corresponding backgrounds in the top and bottom sub-slice at a fluence of 1000 e/pixel. Scale bar, 625 nm.

Figure 4

Reconstructed volume in x-z plane from liver sample cut at increments of 50 nm and imaged with primary energies of 1.4 keV and 2.4 keV. (a,c,e) Average of BSE image stacks acquired at beam energies of 1.4 keV and 2.4 keV, shown in three different regions of hepatocyte. (b,d,f) Same views as in (a,c,e) but calculated with sub-slice z-resolution. (g,i,k) Views at higher magnification from the rectangular regions indicated in (a,c,e), respectively. (h,j,l) Views at higher magnification from the rectangular regions indicated in (b,d,f), respectively. Arrows indicate membranes of endoplasmic reticulum that show higher resolution in the sub-slice reconstruction. Scale bar, 500 nm.

Figure 5

Reconstructed volume in x-y and y-z planes from liver sample cut at increments of 25 nm and imaged with primary energies of 1.0 keV and 1.4 keV. (a) Low-magnification BSE image of hepatocyte in the x-y plane acquired at primary energy of 1.4 keV. (b) Region indicated by square in (a) at higher magnification showing membranes of endoplasmic reticulum. (c) Average of 1.0 keV and 1.4 keV BSE image intensities displayed in the y-z plane. (d) Calculated sub-slice stack in the y-z plane showing improved spatial resolution relative to (c), but with additional noise. (e) Five-slice average in the x-direction of y-z view in (c). (f) Five-slice average in the x-direction of y-z view in (d). Arrows in (c–f) denote areas with improved resolution. Scale bar in (a), 5 µm; and in (b–f), 500 nm.

Figure 6

3D rendered membranes with and without sub-slice reconstruction. (a,b) Views of region of hepatocyte in the y-z orthoslices of Fig. 5(d,f) show clear separation of membranes after sub-slice reconstruction. (c,e) Views of three membranes indicated by arrow in (a,b) obtained from the averaged image stacks acquired at beam energies of 1.0 keV and 1.4 keV, i.e., without sub-slice reconstruction, showing that it is not possible to segment the green and red membranes accurately due to the limited z-resolution of 25 nm in the individual datasets acquired at each beam energy. (d,f) Same views as in (c,e) but after sub-slice reconstruction, again showing a clear separation of the green and red membranes. Scale bar, 500 nm.