Radboudumc collaborates with the UMC Groningen and the NKI-AVL to investigate cardiac toxicity associated with radiotherapy for lung cancer patients. In the past years, there is increased awareness that cardiac toxicity is or will be relevant for the treatment of lung cancer, however, no adequate models are available that predict cardiac toxicity. In the project, you will develop such models using data from 6 institutes adding to a total of 4000 patients. You will analyze the data using advanced techniques such as Deep Learning, and use different Machine Learning techniques to automatically detect abnormalities and learn from the data. The aim of the project is to construct and validate robust prediction models that calculate the risk of cardiac toxicity, radiation pneumonitis and survival
ProjectOne of the reasons that robust dose-effect relationships of the heart have not yet been established is due to the use of relatively small cohorts, and the cumbersome and time consuming nature of manual delineation of the heart. Therefore, the first step within the project is to optimize automatic heart delineation, to optimize and validate the method published by our collaborators at the NKI and to compare the performance with state-of-the-art deep learning models to further improve the methodology.
As a second step, these results will be applied to generate and validate algorithms that enable the automatic detection of radiotherapy induced toxicity (pericarditis and pneumonitis), based on imaging before and after treatment.
Finally, machine learning techniques will be used to create prediction models. This study will be performed in close collaboration with 6 institutes. The project will help to allocate the appropriate treatment and optimize radiotherapy for the individual patient and results in implementation into daily clinical practice.
Tasks and responsibilitiesYou will be part of a research project funded by the Radboudumc, and as a member of the Radboud Institute of Health Siences (
RIHS) be stationed at the Radiotherapy department. You will collaborate closely with the image analysis group of the departement of
Radiology (DIAG). Your responsibilities will consist of :
- The acquisition of the large multi-institutional cohort, this will involve site visits in the US, Germany and the Netherlands;
- Using existing software and developing new software to automatically delineate relevant structures in CT and automatically asses induced toxicity of the CT's available in the cohort;
- Using machine learning techniques to create prediction models;
- Of all the above projects you write scientific papers that will be submitted to peer reviewed papers;
- Within 4 years you will finish writing a thesis for your promotion according to rules of the Radboud university.