DIRECT

In the DIRECT (Deep Image REConsTruction) project we develop deep learning-based image reconstruction methods for Cone Beam CT (CBCT), tomosynthesis and MRI reconstruction. Many applications in imaging and radiotherapy would benefit from real-time reconstruction methods, that is, from methods that can directly map the measured machine data to the reconstructed image. Even more so, the downstream task if often something else: for instance in radiotherapy the primary goal of creating a medical image is not so much the image itself, but the location of the tumor and the organs.

People

Jonas Teuwen

Jonas Teuwen

Assistant professor

Diagnostic Image Analysis Group

Nikita Moriakov

Nikita Moriakov

Postdoctoral researcher

Diagnostic Image Analysis Group

 Roeland Dilz

Roeland Dilz