PhD positions: Predicting radiotherapy toxicity with machine learning and images



Working Enviroment

The PhD students will be part of a dynamic, multidisciplinary research team from UMCG including other (PhD) students, technical medicine graduates, computer scientist, and clinical researchers. The supervision team consists of a Technical Physician, Medical Physicist, Computer Scientist, and Radiation Oncologist. PhD students are registered within the Graduate School of Medical Sciences in Groningen, and graduate with a PhD degree from the University of Groningen, Faculty of Medicine.
The Department of Radiation Oncology in the UMCG is the first in the Netherlands with an integrated state of the art photon-proton clinic. Currently, radiotherapy treatments are provided for over 4,500 patients annually. The department has the largest prospective follow-up program of systematically collected clinical, toxicity and tumour data of head and neck cancer patients in the world. The department is part of the UMC Groningen Comprehensive Cancer Center, with 21 multidisciplinary tumour boards and provides the highest level of oncologic services. The department is actively involved in patient care, research, education, and training. The UMC Groningen Proton Therapy Center, part of our Department, started to treat patients with IBA pencil beam scanning technology in January 2018.
The University Medical Centre Groningen, located in the centre of Groningen, is one of the Netherlands’ eight university medical centres and the largest employer in the Northern Netherlands. It has an ambitious, dynamic, international environment with state-of-the-art facilities. More than 13,000 employees provide patient care, are involved in medical education, and perform cutting-edge scientific research, focused on healthy ageing. Data science and large-scale studies are important focal points in Groningen.

Job description

An increasing number of head and neck cancer patients survive years after radiotherapy, due to improved treatment and increase in favourable tumour subtype. Nevertheless, these survivors often suffer from persistent severe side effects, gravely impacting their quality of life. The PRI2MA project aims to reduce radiation-induced toxicities by improving toxicity prediction before treatment to guide treatment decision-making. For example, this will allow better dose distributions with less toxicity, better selection of patients for proton versus photon therapy, and improve management of toxicity care.

Recent work has been performed to develop comprehensive individual toxicity risk profiles (CITOR) based on validated toxicity models. The proposed PRI2MA project, aims to improve the predictive power of these models with deep learning by adding information on 3-dimensional (3D) radiation dose distributions and multimodality imaging. Prospectively collected toxicity scores, imaging and radiation dose are currently available for more than 1500 patients at our department (Radiotherapy, UMCG). Additionally, this project is performed in collaboration with MD Anderson Cancer Center (Houston, Texas), which is the largest cancer centre in the US and who provides validation data, at which PhDs will have the opportunity for a research intern (not mandatory).

The project is funded by KWF for 2 PhD student for 4 years, one PhD will focus on the use of medical imaging for prediction of toxicities, while the other will focus on deep learning prediction models. Close collaboration between these PhD students (and PI) is essential. The final product will result in a decision-support tool that is integrated into routine clinical practise. The PRI2MA decision support tool will be designed for a broad application space, which includes patient-understandable expected toxicity reporting, head and neck toxicity care management, treatment comparison and alteration. The development of this tool will be a collective effort of the PI and the PhDs.

PhD project 1
This project aims to incorporate high-dimensional medical imaging information into existing prediction models, which are currently based on radiation dose parameters and clinical factors only. Image biomarkers will be extracted from organs-at-risk that have been segmented on CT, FDG-PET and anatomic/quantitative MR images. Machine learning approaches will be deployed to deal with the high-dimensional image data and improve the prediction of toxicities. The final task aims synergize image information from PhD project 1 and the 3D dose distribution deep learning models from PhD project 2.

PhD project 2
This project aims to improve toxicity predictions by developing a convolutional neural network (CNN) architecture that – instead of using discrete dose-volume parameters – can deal with entire 3D dose distributions and multi-organ segmentations as a full image. Additionally, we will advance towards deep learning models that can predict multiple toxicities in a single model simultaneously. The final task aims synergize image information from PhD project 1 and the 3D dose distribution deep learning models from PhD project 2.

Applicants are invited to write for 1 of the PhD positions within the project.


- Candidates with a master’s degree in Technical Medicine, Biomedical Engineering, Physics, Computer science or related field.
- Experience with deep neural networks or medical imaging processing.
- Technical proficiency in Python and/or Matlab.
- Aware of recent developments in deep learning.
- Excellent communication skills in English.
- Flexible and collaborative attitude.
- Available to start the PhD project this year (aimed start date September/October 2021).

Conditions of employment

- A fulltime PhD position (36 hours). First appointment for one year. If candidate is proven suitable, the appointment will be extended for another 3 years.
- Salary according to the Dutch scale for PhD project research = € 2495,- gross per month in the first year up to a maximum of € 3196,- gross per month in the last year.
- A holiday allowance of 8% of the gross annual salary and a year-end bonus of 8.3%
- Minimum of 23 holidays per year in case of fulltime employment.

The UMCG has a preventive Hepatitis B policy. The UMCG can provide you with the vaccination, should it be required for your position.

In case of specific professions a ‘Certificate of Good Conduct’ is required.

More information
Sanne van Dijk, MSc, PhD, email: ;

Additional information

Applying for a job

Please use the the digital application form at the bottom of this page - only these will be processed. You can apply until July 13st, 2021. Within half an hour after sending the digital application form you will receive an email- confirmation with further information.

Digital application form