Radiomics and Machine Learning to predict response to treatment in pancreatic cancer


Pancreatic cancer is among the most lethal cancers worldwide. Most patients present a poor prognosis, due to the cancer being detected typically at a late stage, a fact that makes the cancer often inoperable. To try and reduce the size of the cancer, and thus increase the chance of a successful surgery, chemotherapic treatment is sometimes administered before surgery takes place. However, the response to chemotherapy varies significantly among patients. Some individuals respond only partially, while others do not seem to respond at all, resulting in months of treatment with potentially severe side effects and without any actual benefit. The ability to predict the response to chemotherapy before or at the beginning of the treatment could therefore help select the most appropriate treatment options for each patient, potentially resulting in important improvements in pancreatic cancer care.

Aim and Tasks

The aim of this project is to develop a model able to predict the response to chemotherapy in patients affected with pancreatic cancer. For this, you will use a dataset of CT perfusion images, a state-of-the-art imaging modality, of pancreatic cancer patients acquired at our institution. You will develop algorithms to extract and quantify relevant imaging biomarkers from these images, related to cancer anatomy and function, and you will then use these biomarkers to develop a machine learning model for outcome prediction.

You will be supervised by a postdoctoral researcher expert in machine/deep learning for medical image analysis, and by the radiologist responsible and leader for pancreatic cancer research at our department.

The project will involve the following tasks:

  • Organization of the dataset
  • Development of image analysis algorithms for imaging biomarker quantification from CT perfusion images
  • Statistical analysis of these biomarkers
  • Development of a machine learning model to predict the response to therapy based on the extracted and analyzed biomarkers
  • Analysis and discussion of the results obtained

To be successful in this project, you should:

  • ... be a creative and enthusiastic MSc student in biomedical engineering, medical physics, technical medicine, bioinformatics or similar, with interests in biomedical research and medical imaging.
  • ... have knowledge in quantitative programming (Matlab, Python).

If you already have some knowledge on machine learning, statistics, or in biomedical research, then this is definitely a plus!

Additional Information

  • No financial compensation will be given for this project
  • The optimal duration of the project is approximately 6 months, although it can be adjusted according to your needs.
  • The project can result in a Master’s thesis.
  • Location: Department of Medical Imaging, Radboud University Medical Center (remote working is possible for part of the time)

Information and Application

For further information about this project, or to apply, please contact Marco Caballo ( To apply, please send your CV and a short letter (half a page) explaining why you are interested in this project, and which skills you expect to learn/improve from it.


The Department of Medical Imaging is one of the most active research departments of the Radboudumc, counting more than 100 researchers, and ranked in 2021 the fourth in the whole world for quality and quantity of research activity. The Radboudumc is a leading academic center for medical science, education and health care with over 8,500 staff and 3,000 students. Radboudumc strives to be a leading developer of sustainable, innovative and affordable healthcare to improve the health and wellbeing of people and society in the Netherlands and beyond. This is the core of our mission: To have a significant impact on healthcare.