The focus of this PhD fellowship is on developing deep learning methods for spatio-temporal medical image analysis, such as dynamic positron emission tomography or echocardiography. The PhD student will address challenges in uncertainty prediction, explainability, self-supervised learning, multimodal and temporal information fusion, and the integration of medical domain knowledge to enhance network training and task performance.

The core of the position involves creating new deep learning techniques for spatio-temporal data analysis with limited labels. Key research aspects include estimating and modeling uncertainty, particularly developing new methods for spatio-temporal imaging data. The aim is to establish trustworthy and reliable deep learning solutions for clinical use. Developing explainable methods for spatio-temporal data is a top priority, including method evaluation using medical domain knowledge from experts or additional imaging modalities that provide complementary anatomical information.

You are expected to collaborate with other members of the Visual Intelligence center and the Machine Learning research group, across various partners and innovation areas. You will contribute to the centre’s seminars and be part of a network of young researchers in deep learning in the Visual Intelligence Graduate School.