Daniel Frey

PhD candidate at Technical University of Munich (TUM), affiliated with the Chair of Biomedical Physics and the Munich Institute of Biomedical Engineering, in close collaboration with the Chair of AI in Healthcare and Medicine.

Research

My research focuses on advancing the clinical translation of human-scale dark-field computed tomography (DFCT), an X-ray modality based on small-angle scattering with strong potential for pulmonary imaging. I develop methods for DFCT streak artifact reduction and image quality enhancement using (self)-supervised techniques and learned representations, aiming to bridge physics-based domain knowledge with modern machine learning.

Background

I hold Bachelor's degrees in Biochemistry and Physics from TUM, where I was first introduced to dark-field imaging. I went on to complete the Biomedical Engineering and Medical Physics (BEMP) Master's program, which I finished with a thesis on the human-scale DFCT prototype. Alongside my studies, I gained practical engineering experience at the TÜV SÜD electrical safety department, and developed expertise in deep learning workflows for neuroimaging as part of the Morphometry Group at TUM University Hospital.

CONFERENCE ABSTRACTS & PRESENTATIONS

• MultipleMS Spinal Cord MRI [Oral Presentation]
  ECTRIMS 2024, Copenhagen, DK  ·  Conference  ·  Slides
  D. Frey, J. McGinnis, M. Mühlau
  Status update on MultipleMS, a multi-center longitudinal cohort study for the analysis of spinal cord lesions in MS.

• Robust Sparse-view Dark-field CT with 3D Gaussian Splatting [Poster Presentation]
  IEEE ISBI 2026, London, UK  ·  Conference  ·  Abstract  ·  Poster
  D. Frey*, T. Dorosti*, J. McGinnis, T. Hiu, F. I. Ozlugedik, J. B. Thalhammer, S. Peterhansl, D. Rueckert, F. Pfeiffer, F. Schaff
  Demonstrates that 3D Gaussian splatting enables enhanced tomographic reconstruction in dark-field imaging compared to FDK.

• Streak-reduced Human-scale Dark-field CT with 3D Gaussian Splatting [Poster Presentation by T. Dorosti]
  IEEE ISBI 2026, London, UK  ·  Conference  ·  Abstract  ·  Poster
  T. Dorosti*, D. Frey*, J. B. Thalhammer, J. F. Hilmer, P. Bleuel, S. Peterhansl, J. McGinnis, D. Rueckert, D. Pfeiffer, F. Pfeiffer, F. Schaff
  Repurposes 3D Gaussian splatting as streak artifact filter by post-processing reconstructed dark-field CT volumes.

• X-ray Dark-field CT Radiomics for Lung Phantom Assessment [Poster Presentation by S. Demianova]
  IEEE ISBI 2026, London, UK  ·  Conference  ·  Abstract  ·  Poster
  S. Demianova*, D. Frey*, P. Bleuel, J. F. Hilmer, L. Kayser, D. Pfeiffer, T. Koehler, F. Pfeiffer
  Proposes a lung phantom assessment methodology leveraging dark-field CT texture features via radiomic analysis.

* Equal contribution