By combining backscatter image sequence acquisition and deconvolution, a new high-isotropic resolution imaging method has been developed for cell biologists.
Cell biologists need high-resolution 3-D imaging to understand the structure-function relationships of organelles and other structures in cells, and the connectivity and organization of cells in tissues. Many of these structures are too small to be seen clearly in a light microscope and require the higher resolving power of an electron microscope.
Scanning electron microscopes (SEM) are becoming increasingly popular for 3-D studies in neurobiology due to new physical slicing techniques. When using these approaches, the section thickness often limits the resolution in the z-direction. For instance, Drosophila brain tissue exhibits neurites less than 40 nm, which could cause a misinterpretation of connectivity if cuts were made at the same order of magnitude.
Serial slicing methods based on diamond-knife cutting are, however, reaching practical limitations in terms of achievable z-resolution and voxel isotropy. This is mainly due to the difficulty of cutting sections thinner than 15 nm with the desired consistency. While focused ion beam (FIB) serial block face imaging can improve the z-resolution to 5 nm, this technology is restricted due to the total volume of material that can be processed.
Scientists at FEI, Hillsboro, Ore., invented the new ThruSight method that achieves high-isotropic resolution by a combination of backscatter image sequence acquisition and deconvolution (DC) (Figure 1). ThruSight is built on improved understanding of beam-sample interaction for classically prepared resin-embedded samples. Monte Carlo simulation tools and experimental observations show that these materials exhibit high linearity, allowing for the use of image formation models based on linear convolution. Furthermore, the point spread function (PSF) of backscatter electrons (BSE) in these materials appears to be well confined laterally for the typically used primary energies.
As the range of penetration in the sample is dependent on the energy of the primary beam, acquiring an image sequence with increasing landing energies leads to the acquisition of images from increasing depth. These images contain overlapping volume information from which 3-D layers can be extracted using DC algorithms (Figure 2).
The good lateral confinement of PSFs allows for restricting the DC to the z-axis making it similar to a source separation task. As the structure of the PSF is difficult to obtain experimentally, it is a variable in the resulting blind DC problem. Efficient iterative methods allow for the recovery of both the depth layers and PSFs.
To verify the reconstruction results FEI combined this technique with classical FIB-SEM serial block imaging using the through-the-lens detector in BSE mode on a Helios NanoLab 650 DualBeam. The z-resolution was controlled by switching the primary beam energy in small steps. The comparison with a high-resolution FIB reconstruction showed identical structures on the studied samples proving the reliability of the 3-D technique. From this comparison, the reached depth at 5 kV was estimated to be around 200 nm. The combination of virtual slicing using BSE DC with physical cutting offers the possibility of reconstructing very large high-resolution data sets with isotropic voxels.
The method has been developed around typical life science specimens (dehydrated, heavy metal-stained and resin-embedded), prepared for serial sectioning techniques. On resin-embedded biological samples the technique can create 3-D models of the region 100 to 200 nm below the surface with isotropic resolution as good as 4 nm throughout. When combined with serial sectioning, it breaks the dependence of z-resolution on section thickness, permitting faster, higher-resolution composite models of larger volumes from fewer, thicker physical sections.
The method also allows for the collection of high-quality isotropic data, even when the specimen does not allow the physical removal of thin enough layers.