USE CASE · VOLUME EM
From EM stack to quantified organelles.
Align, explore and segment FIB-SEM, SBEM and serial-section data in the browser, then measure the cells and organelles you find.
Data: jrc_jurkat-1, OpenOrganelle (HHMI Janelia; Xu et al. 2021, Heinrich et al. 2021, Nature).
What volume EM demands
Volume EM turns a cell or a piece of tissue into thousands of images. Getting from there to a measurement takes more than a viewer.
Very large data sizes
A single SBEM or FIB-SEM run can produce terabytes of images. Copying them to every collaborator’s workstation is slow, and often impossible.
Stacks that need aligning
Serial sections drift, warp and tear. Before anyone can segment, the stack has to become one consistent volume.
Thousands of structures
Mitochondria, nuclei, membranes and vesicles: a single volume holds far more objects than anyone can segment by hand.
Results you can measure
An image is not a result. Volumes, surfaces and counts have to come out of the data, in a way others can reproduce.
From stack to structure, in WEBKNOSSOS
Five steps, one platform. Use the ones your project needs: some labs annotate everything by hand, others annotate a few examples and let a model do the rest.
01
Align and explore
Upload or stream your FIB-SEM, SBEM or serial-section data, align it in WEBKNOSSOS, and move through the full volume at native resolution in the browser.
Data: Luke Noon, CIPF-ISABIAL, 3D-EM Unit (Valencia, Spain).
02
Correlate with light microscopy
Bring in the fluorescence that told you where to look. Register it onto your EM with matching landmarks or BigWarp, then overlay it to find your cells of interest in the ultrastructure.
Data: Drawitsch et al., eLife 2018.
03
Annotate by hand, with help
Paint cells and organelles with brush and trace tools, fill the slices in between with interpolation, or click an object to select it with AI quick-select. Many labs stop here; others use these annotations as ground truth.
Data: Motta et al., Science 2019.
04
Segment automatically
Detect mitochondria with a pre-trained model, or train a model on your own annotations for any structure. Run it on a bounding box, check the result, then on the whole dataset.
Data: Motta et al., Science 2019.
05
Quantify your structures
Turn every segment into a 3D mesh, then measure volume, surface area and bounding box per object, and export the numbers as CSV.
Data: Motta et al., Science 2019.
More tools for volume EM
Beyond the core workflow, WEBKNOSSOS has many smaller features that save time along the way.
Sharing links
Send collaborators or reviewers a link to the exact view, and link figures in your paper to the data.
Volume EM on WEBKNOSSOS
Published volume EM datasets you can open and explore in your browser.
“WEBKNOSSOS played a central role in our research. We stored all of our datasets there and used it for data sharing, collaboration, and annotation.”
“WEBKNOSSOS is very useful for safely storing large datasets and managing access across multiple users.”
Want us to align or segment your data?
Our team aligns your serial sections and runs automated segmentation of cells and organelles, delivered in WEBKNOSSOS.








