Researchers: Willem Diepeveen, Rob Tovey, Tatiana Bubba, M. Benning, C.-B. Schönlieb, O. Öktem, C.E. Yarman

Electron microscopy is a powerful tool in the physical, biological, and industrial sciences advancing areas from nanotechnology to drug discovery. There are three key features for this success, the visible is resolution, the current world record is held at 0.04nm, smaller than the radius of a Hydrogen atom (link). The second is that electrons interact much more richly with matter than photons, this enables spectral imaging techniques which allow us to identify different atomic configurations within a sample and colour these regions as seen in the image opposite (from [Nicoletti et al., 2013]). The final attribute which makes Electron Tomography invaluable is the word ‘tomography’, essentially the ability to recover full 3D data from a scan. The ability to see all details of a sample, even when they are beneath the surface, distinguishes electron microscopy from visible light microscopy. 

The goals of this project are two-fold, we wish to address the computational challenges and also further the functionality of electron microscopy by leveraging new aspects of the interactions between electrons and matter. On the computational side, we have already proposed an algorithm which is capable of recovering high resolution reconstructions despite a lot of missing data [Tovey et al., 2019]. Moving forward, we are also looking to improve the speed and stability of algorithms by proposing novel bases for reconstruction. If a sample admits a natural parametrisation then it should be better to reconstruct in that parametrisation rather than a standard discretised grid. To advance the analysis of electron diffraction patterns, we are currently investigating the physical model of electron imaging with the aim of reconstructing strain of a 3D sample. This is a great example of recovering multi-dimensional parameters (strain is a matrix) from grey-scale images.