KI4D4E

Objectives & Goals

The overall objective of the project is to leverage Artificial Intelligence techniques for more efficient use of synchrotron radiation sources with micro-computed tomography (4D-CT).  A collaboration with academic and industrial partners aims to apply AI to improve the productivity of tasks related to compression for large volumes of data, visualization, correction of artifacts, image reconstruction, and simulation.

Despite the success of these methods, they often require massive amounts of training data. To address this, our contribution to the project is to provide highly diverse and quality training datasets to improve the performance of partner models. One direction of research is the application of a numerical simulator, where users get to setup virtual experiments. A second direction is the application of novel machine learning techniques, which are able to produce highly realistic synthetic images. The goal is to provide a framework and a database for scientists interested in testing their algorithms on artificial data.

Areas of Research

- Numerical Simulation of 3D/4D CT data

- Automated generation of synthetic data via Machine Learning

- Reduction of reality gap between simulated and real data

Project Team

- Dr. Tomáš Faragó

- Dr. Alexey Ershov

- Guilherme Ribeiro da Silva

- Gabriel Lefloch

Collaboration

Prof. Dr. Sven Simon, ITI, chair and department Computational Imaging Systems, University of Stuttgart

Prof. Dr. Tomas Sauer, Chair of Mathematics with a focus on Digital Image Processing, University of Passau

Prof. Dr. habil. Andreas Maier, Chair of Computer Science, Pattern Recognition, Friedrich-Alexander University of Erlangen-Nuremberg

Prof. Dr. Simon Zabler, Fraunhofer Society, FHG institutes with expertise in the X-ray and CT imaging

Dr. Ingo Manke, Institute for Applied Materials Research, Helmholtz Center Berlin for materials and energy GmbH

Dr. Julian Moosmann, Institute for Materials Physics, Dr. Berit Zeller-Plumhoff, Institute for Metallic Biomaterials, Department of Imaging and Data Science, Helmholtz Center Hereon

Prof. Dr. Werner Lehnert, Forschungszentrum Jülich GmbH, IEK-14, teaching and research Research area modeling in electrochemical process engineering, RWTH Aachen,

Dr. Paul Tafforeau, ESRF, the European Synchrotron

Dr. Bernhard Hesse, Xploraytion GmbH, Berlin

Prof. Martin Müller, Richi Kumar, Helmholtz Center Hereon, German Engineering

Materials Science Center (GEMS) at the Heinz-Maier Leibnitz Center (MLZ)

Dr. Alessandro Tengattini, HFR at ILL Grenoble, Imaging Instrument NeXT

Dr. Robin Woracek, ESS Lund, Imaging Instrument ODIN

Dr. Marian Willner, MITOS GmbH, Garching near Munich

Dr. Andreas Wiegmann, math2market GmbH, Kaiserslautern

Funding

KI4D4E project is funded by the German Federal Ministry for Education and Research (BMBF) under the funding code 05D23VK1.