Imaging Physics in the Context of Radiotherapy/ Medical Physics (Landry) 1

Department / Institute
Department of Radiation Oncology, University Hospital of the LMU Munich
Subject area
Imaging Physics in the context of radiotherapy/ medical physics
Project title
Deep learning based 4D cone beam computed tomography scatter correction
Name of supervisor
Prof. Dr. Guillaume Landry
Number of open positions
1
Language requirements
Proficiency in English
Academic requirements
Candidates with experience in medical physics and deep learning would be ideal. Prerequisites are good programming knowledge (python, pytorch/ kera/ tensorflow) and knowledge of the physics of particle therapy and imaging. Familiarity with medical imaging topics would be desirable. This can be expected from a physics graduate student, who had ideally worked on a topic related to medical radiation physics in their master thesis. Equivalent engineering backgrounds may also be compatible.
Project time plan
36 months
Contact
csc.international@lmu.de

Project description

In ion therapy of cancer, beams of protons or heavier ions are used to treat tumors while sparing healthy tissue by exploiting their finite range. Prediction of this range is performed using x-ray computed tomography (CT) images. However, in the treatment room, we find cone beam CT (CBCT) scanners that exhibit image artifacts due to the detection of scatter by the large detector array. Correcting for scatter enables in-room close calculation and potentially plan adaptation based on CBCT images. Furthermore, when treating lung tumors, having motion information from a 4D-CBCT scan is beneficial. In our group, extensive experience with CBCT scatter correction exists including deep learning methods. However, adapting these to 4D-CBCT requires additional refinements to account for the diminished imaging quality from undersampling. The goal of this project will be to translate a classical non-AI 4D-CBCT correction method, which is too slow for clinical use to a deep-learning based workflow. The project offers the possibility to acquire advanced deep learning-based image processing skills.

What are you looking for?