Position: Research Fellow in Machine Learning for Radiation Therapy
Department: Medical Physics in Radiation Oncology
Code number: 2019-0291
The German Cancer Research Center is the largest biomedical research institution in Germany. With approximately 3,000 employees, we operate an extensive scientific program in the field of cancer research.
The new BMBF funded ARTEMIS project is focused on the establishment of a unique MR-guided ion-beam therapy prototype for clinical application. The continuous monitoring with high soft-tissue-contrast allows to complement the precision of ion beam irradiation with an accurate organ and target volume tracking. As up till now ion treatment planning relies on the planning CT to stopping power conversion, substitution of the CT by the MRI modality poses the challenge to predict the stopping power distribution of each patient from an MRI.
We invite applications for an enthusiastic research fellow to join the collaboration between DKFZ and the University Hospital Heidelberg to develop a correlation model for pseudoCT generation in a MR-only planning scenario utilizing cutting edge machine learning methodologies.
In ARTEMIS, you will work within a research group at DKFZ collaborating with researchers and clinicians in the field of radiation oncology. You will exploit the unique data source generated in Heidelberg, a hub for MR-guided treatments, to establish a prediction model for creation of digital patient twins to optimize radiation treatment planning.
- Analyze anonymized MR-CT datasets of patients and phantom measurements
- Design and establish machine learning models and pipelines for MR-only pseudoCT generation process (or directly stopping power maps)
- Scrutinize the model robustness and accuracy in comparison to image registration based pseudoCT generation methodology
- Present your findings in international conferences and in scientific publications
- Design and implement a deployment channel for model utilization to the clinical partner site
- PhD (or master’s degree plus 3 years relevant experience) in computer science, computational physics, medical informatics or equivalent
- Demonstrated experience in machine learning on medical images
- Previous experience with medical imaging data like CT and MRI and radiation treatment concepts is desired
- Working experience with deep learning models (e.g. UNET) utilizing GPU-accelerated Tensorflow is desired
- Solid programming skills in Python (C++ and/or CUDA is a plus)
- Excellent interdisciplinary communication, presentation, and scientific writing skills in English
You are a highly motivated data scientist with the desire to contribute to the improvement of ion-beam therapy planning. You are passionate about getting clinically relevant knowledge out of real-world datasets and are looking forward to collaborative interdisciplinary work. You have an independent and pro-active work style but also interact well in a team environment. The position will give you the opportunity to work on a unique clinical prototype for MR-guided ion-beam therapy. You will have opportunities to supervise and mentor junior researchers during the appointment. The option exists to earn a doctoral degree if desired.
- Interesting, versatile workplace
- International, attractive working environment
- Campus with modern state-of-the-art infrastructure
- Salary according to TV-L including social benefits
- Possibility to work part-time
- Flexible working hours
- Comprehensive further training program
- Access to the DKFZ International Postdoc Program
Earliest possible start date: as soon as possible
Duration: The position is limited to 2 years with the possibility of prolongation.
The position can in principle be part-time.
Application deadline: 04.10.2019
Dr. Kristina Giske
Phone +49 (0)6221/42-2579
Please note that we do not accept applications submitted via email.