I am a researcher working at the intersection of computer vision, systems neuroscience, and bioinformatics, focusing on cross-domain challenges of advanced visual computing methods to specialized biomedical imaging and computational biology workflows. My interests include large-scale multimodal data, 2-photon microscopy, thermal imaging, face analysis, and bridging methods from computer vision and deep learning into modern bioinformatics pipelines.
My current work develops computational methods for motion analysis in biomedical imaging, addressing challenges from microscopic neuronal recordings to human facial dynamics. I focus on creating robust algorithms that handle the unique constraints of biological imaging: low signal-to-noise ratios, non-rigid deformations, and the need to preserve subtle physiological signals while correcting artifacts.
Motion Correction for Functional Microscopy
Developing variational optical flow methods specifically tuned for 2-photon calcium imaging, enabling accurate tracking of neuronal activity in freely behaving organisms. This work addresses the critical need for subpixel-precision registration in low-SNR conditions while preserving functional signals. Complete ecosystem for motion correction across platforms: [Flow-Registration Suite].
Motion Magnification and Microexpression Analysis
Creating Lagrangian-based amplification techniques that selectively enhance subtle movements in video, with applications ranging from revealing facial microexpressions to visualizing small physiological motions. The approach uses sparse optical flow decomposition to achieve local magnification while maintaining spatial coherence.
Volumetric Registration and Motion Analysis for 3D Microscopy
Extending 2D registration methods to native 3D, addressing the computational and algorithmic challenges of dense volumetric flow estimation for modern light-sheet and multi-plane imaging systems.
Modern Python implementation of Flow-Registration for 2D motion correction in microscopy. Provides both a programmatic API and integration with popular imaging tools. [GitHub] [PyPI]
FlowReg3D (In Development)
Native 3D extension of Flow-Registration for volumetric microscopy data. Designed for dense, subpixel-precision motion estimation in light-sheet and multi-plane imaging.
The Visual Computing for Neuroscience Lab was part of the Systems Neuroscience and Neurotechnology Unit (SNNU), Saarland University, from 2018 to 2024. During that time, I have served as lab head and lead computer vision researcher closely collaborating with chairs and researchers at Saarland University, Saarland University Hospital, TU-Berlin, the Okinawa Institute of Science and Technology, the University of Chicago and others. The lab provided frameworks and solutions for cross-domain imaging, motion correction, and multi-camera setups. Check out my GitHub for implementations of some of the projects, in particular related to motion magnification and 2P motion correction. And a 2020 Saarland University talk on the scope of the research (only in German). Some resources: