Gijs van Tulder

I am a researcher in machine learning for medical image analysis, combining deep learning and computer vision techniques with applications on MRI, CT, and X-ray. I am most interested in the methodological aspects of medically-inspired problems, such as domain adaptation and representation learning.

In the Pattern Recognition Laboratory of the TU Delft, I work on image analysis for osteoarthritis imaging, in collaboration with clinical researchers at the Erasmus MC in Rotterdam. In this project I focus on the analysis of longitudinal image data (i.e., time series) and the analysis of data from heterogeneous sources (domain adaptation), in order to improve the predictions of disease progression and improve our understanding of the methods involved.

Following my MSc in computer science at the TU Delft, I worked as a PhD student and researcher at the Biomedical Imaging Group Rotterdam, part of the radiology department of the Erasmus MC, on a wide range of more and less practical applications. My PhD thesis addresses the topic of domain adaptation and transfer learning in a medical imaging context. Subsequently, I was part of the Data Science group of the Radboud University in Nijmegen, where I worked on transfer learning for breast cancer screening and taught deep learning to MSc students.

See [my website) for more details and publications.