Comparing Array Tomography with Serial Block-face Electron Microscopy

Using a fig and a tomato

Array tomography (AT) and Serial Block-face Electron Microscopy (SBEM) are two microscopy techniques that work in a similar way and are hard to understand. We playfully explained and compared these techniques using a fig (AT) and a tomato (SBEM).

After having recorded the sections of the fruits, we uploaded the data to webknossos, aligned it, and segmented it.

Check out the results here:
Tomato: https://wkl.ink/8913
Fig: https://wkl.ink/9112

Watch the video to understand the whole process effortlessly!

THE FIG

We started with the fig and imaged it in a AT style. We first cut the sections of our tissue sample — the fig. After collecting and placing them, we recorded all the sections from the top with a high res camera — our budget alternative to a full-blown electron microscope.

The data was then uploaded to webknossos and we started the alignment of the individual slices into a coherent dataset. Additionally, we used the new webknossos quick-select to do a full segmentation (see below).

THE TOMATO

For the tomato (imaged in a SBEM style), we first cut one very thin slice off the top. After imaging it from the top, we repeated the whole process over and over again. Slice by slice we were able to image the whole tomato until the whole sample was consumed. Finally, the data was also uploaded, pre-processed and segmented in webknossos, similarly to the fig example above.

SEGMENTATION PROCESS

For both fruits, we used the quick-select tool in combination with the interpolation tool in webknossos. This made the annotation process incredibly fast. Within 3 minutes we were able to fully reconstruct our samples.

And for the last step, we let webknossos compute the 3D meshes. I love how distinctly the shapes of both fruits were captured by our simplified setup.

Here it is, two fruits reconstructed using two different microscopy techniques! While the resolution of our approach does not quite match its bigger cousins, our workflow is identical and nicely showcases how to image and work with volumetric EM datasets in webKnossos.

Try out webknossos for free at webknossos.org and follow us on Twitter to be kept in the loop about the experiments and new features of webknossos!