TU Delft is a top tier university and is exceedingly active in the field of Artificial intelligence. The “Knowledge-Driven AI Lab” (KDAI Lab) is devoted to the development and application of AI in applied sciences. We aim to strengthen today’s data-driven AI through integrating fundamental knowledge of imaging physics.
AI techniques are transforming the entire chain of medical imaging, from physical acquisition and image analysis, to its diverse biomedical applications. The current success of AI can be largely attributed to the development of deep learning techniques that learn from large-scale data. The inherent dependency on the available training data, however, raises issues in theory and in practice, including among others, the limited generalizability when distribution shift occurs, and poor explainability when the set of parameters is large and unintuitive. This poses serious concerns when AI is to be applied to a clinical scenario where the stakes are high.
KDAI has 2 PhD vacancies for AI research in medical imaging:
(1) Research in AI methods integrating imaging physics. Your role is to develop AI methods that go beyond the data-driven principle and actively integrate the imaging physics of data (e.g. MRI, CT). The goal is to make the AI algorithms fundamentally robust and generalizable to data sources, including different manufacturers and medical centers.
(2) Research on Explainable medical imaging AI. You role is to investigate explainable AI (XAI), in the scenario of medical imaging. This includes post-hoc understanding and visualization of complex deep neural networks for medical image analysis, and integration of explainability into the medical imaging AI algorithm design.