Dr. rer. nat. Florian Schaff

Courses and Dates

Offered Bachelor’s or Master’s Theses Topics

AI in Physics: Convolutional neural networks for dark-field X-ray CT reconstruction

Grating-based X-ray dark-field imaging uses scattering of X-rays to create an image of an object, rather than conventional X-ray attenuation. The combination of X-ray scattering with imaging allows us to map information about structures that are much smaller than the resolution of the imaging system over a large field of view. X-ray dark-field imaging can be combined with computed tomography (CT) to create three-dimensional images of the scattering distribution inside an object.  DF-CT was recently implemented for the first time into a clinical CT here at TUM

(https://www.bioengineering.tum.de/en/news/details/new-technology-for-clinical-ct-scans).

The goal of this project is to use convolutional neural networks (CNNs) to remove sampling artefacts in DF-CT images. Due to the unavailability of training data from the DF-CT machine, a technologically similar experimental setup and apply transfer learning will be used. The student will acquire, process and prepare training data, as well as train and apply CNNs.

Character of thesis work: mainly computational physics & image processing

For more information, please contact: Dr. Florian Schaff (florian.schaff@tum.de), or Prof. Franz Pfeiffer (franz.pfeiffer@tum.de).

suitable as

• Master’s Thesis Applied and Engineering Physics

Supervisor: Franz Pfeiffer