University of Helsinki
Our first colloquium of the new academic year will be held on 7th September. The speaker is Samuli Siltanen, Professor of Industrial Mathematics here in Helsinki. He is an expert in inverse problems, particularly as applied to medical imaging
We think that his work will be of great interest to physicists too – as it has connections with physics in a variety of ways, ranging from the imaging technologies on which it relies through to the mathematical techniques employed in his work. His talk, titled Three-dimensional X-ray vision by sparse tomography, will introduce us to his latest research.
Here is the abstract:
Traditional X-ray tomography is used routinely in hospital CAT-scans and in industrial non-destructive testing. There the idea is to collect a large number of X-ray projection images from all around the object, interpret the data as line integrals over a non-negative X-ray attenuation coefficient function, and reconstruct the inner structure of the target. Typically, the reconstruction algorithm of choice is some variant of Filtered Back-Projection (FBP). However, in many practical applications there are radiation dose restrictions or geometric obstacles preventing the collection of a comprehensive dataset. In such cases the FBP algorithm does not perform optimally. In recent years there has been tremendous progress in the development of robust reconstruction algorithms for sparse-data tomography. One of the successful approaches is variational regularization with a sparsity constraint; this approach is closely related to compressed sensing. The new algorithms open up novel imaging possibilities in the fields of dental imaging, welding inspection, adaptive optics in telescopes and environmental monitoring. Changing the physics does not necessarily change the mathematics of reconstruction, so the methods apply as well to electron tomography, neutron tomography and more.
After the 30 minute talk, there will be a cocktail reception. Welcome!
Update 10.9.2018: Samuli’s slides are available to download here.