Microscopy is a key technique to visualize cells and tissues in health and disease. Electron microscopy visualizes the smallest structures of which cells and tissues are composed by bombarding it with electrons and maps those that come through, yielding greyscale images. Recent advances in automation has digitalized EM to such an extend that large areas can be imaged at biomolecular resolution, similar to panorama photos. The resulting large-scale maps are equivalent to black-and-white satellite images in which it is increasingly difficult to navigate if roads, waterways, buildings and other objects remain un-annotated. Analysis is therefore not only labor-intensive but also prone to different interpretations of the person who annotates.
Scientists have now supplemented large-scale EM with Röntgen (X-ray) detection to analyze human pancreas and skin tissue in an unprecedented way. The addition of X-ray imaging adds to the greyscale electron microscopy by also measuring the atoms present in the tissue, allowing color coding of the elemental composition to reflect similarities and differences between the structures present. Furthermore, a data-driven workflow was designed to automatically find and map elemental patterns in the signal such that subcellular features can be visualized and isolated with minimal intervention. Broad implementations of the presented methodology will aid the interpretation of biological ultrastructure, ultimately accelerating the understanding of health and disease.
This research is a collaboration between UMCG and TUD and made possible by NWO.
Publication: Automated analysis of ultrastructure through large-scale hyperspectral electron microscopy
