Skip to content

Christian Salomonsen

I am a PhD candidate at UiT The Arctic University of Norway, working in the Machine Learning Group and SFI Visual Intelligence. My research focuses on deep learning for dynamic medical imaging, with a particular interest in dynamic Positron Emission Tomography (PET), physics-informed learning, kinetic modeling, and practical AI tools for research.

A central part of my PhD work is developing efficient deep learning methods for predicting arterial input functions directly from PET images. I am also interested in how anatomical structure, tracer dynamics, and physiological constraints can be incorporated into learning-based models to make them more robust, interpretable, and useful in practice.

Portrait of Christian Salomonsen

Background

Before starting my PhD, I worked with applied physics and mathematics, computer vision, and remote sensing. My master's thesis focused on uncertainty-guided polygon generation for building detection from aerial imagery.

This background has shaped my interest in methods that combine machine learning with domain knowledge, especially in settings where data are limited, expensive, noisy, or difficult to annotate.

Research interests

I am broadly interested in machine learning methods for spatial and temporal data, especially when they are connected to physical processes or scientific measurement. Current interests include:

  • deep learning for dynamic PET imaging
  • arterial input function prediction
  • kinetic and physiological modeling
  • physics-informed and model-constrained learning
  • uncertainty, robustness, and reproducibility in medical imaging
  • research software, visualization tools, and practical AI systems for scientific workflows

Research software and open tools

I care about making research methods more open, inspectable, and usable. This includes reproducible workflows, open research software, and tools that make advanced methods easier to understand and apply.

Outside my main research work, I enjoy self-hosting, server administration, and building small systems that give me more control over my own digital infrastructure.