Felix Lucka

Deep Learning for Reflection-Mode Ultrasound Imaging

15 Sep 2020

About a year ago we started an exciting project on using deep learning to improve various aspects of reflection-mode ultrasound imaging. I did not know much about this imaging modality up to this point and I’m happy to know a little bit more now. Georgios Pilikos just presented our current results (virtually) at the 2020 IEEE Symposium (IUS) Ultrasonics. Check out the conference papers on an end-to-end deep delay-and-sum (DSA) imaging approach and an end-to-end deep US data compression. Thanks to everyone involved, in particular to Georgios.

Atomic Super-Resolution Tomography

21 Jul 2020

I had the pleasure to be involved in a project of Poulami Somanya Ganguly where we tried to reconstruct a nanocrystal at atomic resolution from electron microscopy images using a grid-free super resolution approach. Check out the paper!

Minisymposium at ICIAM 2019

06 Aug 2019

The International Congress on Industrial and Applied Mathematics (ICIAM) will take place in the beautiful town of Valencia, Spain in mid July. It’s gonna be a big, exciting meeting with thousands of participants. Check out the program. Joana Grah, Martin Benning and I are organizing a symposium on “Deep Learning and Inverse Problems”.

Minisymposium at AIP 2019

06 Jul 2019

The Applied Inverse Problems Conference (AIP) will take place in Grenoble, France, this year. Together with Sophia Coban, Tatiana Bubba, Samuli Siltanen, I organize a minisymposium called “Tomographic Imaging: Recent Advances, Exciting Applications and New Horizons”.

X-ray Data Collection for Machine Learning

22 May 2019

Together with my colleagues at CWI, we just released a large collection of tomography scans designed for developing machine learning approaches to tomographic image reconstruction. All details can be found in the accompanying paper on arXiv. We also prepared a collection of Python and MATLAB scripts for loading, pre-processing and reconstructing the data on on github. The complete data set is on zenodo and can be found via the following links: 1-8, 9-16, 17-24, 25-32, 33-37, 38-42. A big thanks to all my colleagues, in particular to Henri der Sarkissian!