About

Hi! I’m Christian, a PhD Candidate at the University of Tromsø in Norway, working in the Machine Learning Group and SFI Visual Intelligence. My research interest deal with the use of deep learning methods for complex spatial and temporal datasets, such as dynamic Positron Emission Tomography (PET) data.

Recently, I have been involved in methods for predicting the arterial input function (AIF) directly from the PET images with the use of an efficient in-house developed deep learning model, and extending this idea to incorporate physical (anatomical) information about the tracer kinetics during the learning phase of the model.

Background

My master thesis in applied physics and mathematics was concluded July 2024, on the subject of polygonal building extraction from aerial imagery. This program consisted of a foundation in mathematics, statistics, and physics, before a specialization in machine learning and statistics. During these 5 years, I worked as a teaching assistant for 3 semesters; 2 of which holding programming workshops in classical mechanics, and 1 semester of health data analytics.

I also have professional experience from working at KSAT—a leading satellite ground station provider located in Tromsø. First, on the ground network side (1 year), then in earth observation (2 years).

Interests

For me, the PhD is a golden opportunity to work with my hobbies: developing algorithms, theory and projects related to deep learning and AI. In particular, I’m a firm believer of the democratization of research and AI, through e.g. user-friendly deployment and idea sharing. One such example is through the use of GUI implementations of our research in Napari.

Also, the use of large-scale computing solutions such as the LUMI-supercomputer has been a prolific extension to the kinds of projects possible to run.

Self-hosting and managing my own servers (such as this one) is also a keen interest of mine. Not only through allowing full control of the services I use, but also as a means to learn and manage my online privacy and presence.