Medical Informatics

This group is a subgroup of Life Sciences. Coordinator of this subgroup: Peter A.N. Bosman.
This group is a subgroup of Life Sciences.
Coordinator of this subgroup: Peter A.N. Bosman.
On the application side, the subgroup of Medical Informatics focuses on applications of computer science and mathematics in medicine and health. Projects typically target demand-driven research questions that, when solved, enable improved (decision) support for medical experts (i.e. clinicians). As such, the Medical Informatics group focuses on fundamental research questions that underlie translational research, connecting models and algorithms from mathematics and computer science to real-world medical practice.

On the fundamental research side, the subgroup of Medical Informatics focuses on the design and application of practically relevant algorithms, mostly in the computational intelligence domain (i.e. metaheuristic optimization (especially (hybrid) evolutionary algorithms) and machine learning). A particular focus is on the design of multi-objective optimization algorithms since real-world optimization problems are often multi-objective, meaning that several objectives need to be optimized simultaneously. This introduces an inherent trade-off between multiple goals (e.g. maximize radiation delivered to tumor cells and minimize radiation delivered to healthy cells).
A list of projects and (associated) researchers in the Medical Informatics subgroup is provided below.
Improving Childhood Cancer Care when Parents Cannot be There - Reducing Medical Traumatic Stress in Childhood Cancer Patients by Bonding with a Robot Companion.

This is a project funded by the Dutch Technology Foundation STW (Stichting Technische Wetenschappen), the Dutch Cancer SocietyKWF (Koningin Wilhelmina Fonds), and the Top consortia for Knowledge and Innovation Life Sciences & Health TKI-LSH (Topconsortium voor Kennis en Innovatie Life Sciences & Health) and co-funded by companies ASolutions BV, Focal Meditech, Brocacef, and Cancer Health Coach. The project supports a postdoc (at 1fte for 2 years) at CWI, 1 PhD student at the Technical University of Delft, and 1 PhD student at AMC (Academic Medical Center). In this project, we develop new technology in AI and robotics and employ these directly in a real-world scenario, to benefit children suffering from childhood cancer. Children treatedfor cancer in a hospital often experience high levels of stress and anxiety. Not only are they faced with a life-threatening disease, they also find themselves in a strange environment, surrounded by doctors, nurses and other medical personnel. Moreover, during various types of treatment, close family may not be in close proximity, e.g., due to exposure risks in case of radiation treatment, or simply because parents cannot guarantee 24/7 presence. This project unites and advances the state-of-the-art in automated human-robot interaction design and machine learning to provide children with a humanoid robot companion that is capable of being there when humans cannot and that can address a child’s distress, pain and fear to minimize medical traumatic stress. This is of key importance, e.g., to ensure successful, uneventful treatment, to lower risks associated with involving relatives and clinicians (e.g., radiation exposure), and to reduce significant adverse psychological reactions (e.g., posttraumatic stress) to the entire experience.
ICT-based Innovations in the Battle against Cancer - Next-Generation Patient-Tailored Brachytherapy Cancer Treatment Planning.

This is a project funded by the Dutch organisation for scientific research NWO (Nederlandse organisatie voor Wetenschappelijk Onderzoek) within the IPPSI-TA programme (Innovatieve publiek-private samenwerking in ICT (IPPSI) - Technology Area (TA)) and Elekta Brachytherapy, funding a postdoc (at 1fte for 3 years) and a PhD student at CWI, and two PhD students at AMC (Academic Medical Center).  In this project, we tackle demand-driven ICT-related fundamental research challenges that will enable innovations in the important application area of health. We particularly consider the design of brachytherapy (BT) cancer treatment plans. BT is an important type of radiotherapy whereby radiation sources are placed near tumors (using devices such as applicators or interstitial needle implants). ICT plays a crucial role in BT. Software is used to compute, using 3D medical imaging data (e.g., CT- and MRI scans), and ultimately execute, a BT plan to deliver radiation that kills tumor cells. It is, however, impossible to only irradiate tumor cells, putting nearby healthy organs at risk. This makes BT planning hard, requiring medical experts to carefully interact with planning software to find satisfactory plans (see figure for typical example of BT irradiation plan). In this project, we identify several key ICT-related fundamental challenges in optimization and machine learning that will innovate these research areas as well as BT planning and will thereby ultimately improve the patient's quality of life. In addition to designing novel MO optimization algorithms, we design novel machine-learning algorithms to, for the first time, learn what makes a good plan using previously approved BT plans. This allows presenting much better plans to medical experts directly, before fine-tuning is required. We furthermore target exploiting the ever-increasing, ever-cheaper, power of parallel (co-)processing hardware to ensure results are obtained sufficiently fast, enabling planning with state-of-the-art radiation models. Finally, we develop and apply the algorithms that are needed to reach the next level in personalized BT cancer treatment planning by robustly designing devices based on patient-specific data (i.e., medical images) and using 3D printing technology to actually create these devices to realize the best-possible BT cancer treatment plans.
3D dose reconstruction for children with long-term follow-up - Toward improved decision making in radiation treatment for children with cancer.

This is a project funded by KiKa (stichting Kinderen Kankervrij), funding one PhD student at CWI, one PhD student and one postdoc (at 0.15fte for 4 years) at AMC (Academic Medical Center). Childhood cancer survivors (CCS) are at significant risk of treatment-related late adverse effects (AEs).  Radiation treatment (RT) is a cornerstone of childhood cancer treatment. It is imperative to understand the underlying relation between radiation dose and AEs to enable the design of the least toxic, yet effective RT plan and to select the best-suited treatment type (e.g., RT with protons versus photons). Currently, late health outcomes cannot always be fully accounted for, because radiation-risk estimates in long-term follow-up studies of CCS are mainly based on rough estimates of average doses to organs, without considering RT-exposed organ volumes. Detailed knowledge of dose and volume effects from 3D dose distributions in relation to multiple AEs is urgently needed. This poses a major challenge because patients with long-term follow-up were treated at a time (<1990) when RT was still planned on 2D simulator films instead of the currently used 3D computed tomography (CT) scans. Therefore, 3D dose distributions are typically not available for these patients. Phantom-based dose-reconstruction methods exist; however, a more individualized technique is preferred. The aim of this project is to develop and validate a novel approach to reconstruct 3D dose distributions of historically-treated patients for whom RT planning CT scans are not available. To this end, state-of-the-art learning and optimization algorithms need to be developed and applied to match each historically-treated patient with recently-treated patients for whom 3D CT scans are available, followed by geometrical and dosimetric reconstructions.
Deformable Image Registration.

This is partly a CWI/AMC internal project and partly a KWF-funded project titled "Optimized targeting for surgery and radiotherapy of breast cancer with a DCIS component" that funds two PhD students at NKI (Netherlands Cancer Institute), one of which is co-supervised by senior researchers of CWI and AMC (Academic Medical Center). Deformable image registration (DIR) concerns deforming one image to make it look like another image. DIR holds huge potential for healthcare. Consider, e.g., breast cancer, and having a CT image of a breast that contains a tumor and having another CT image, taken later, where the tumor was surgically removed. Knowing how the anatomy of the breast changed is very important so that follow-up radiation therapy can narrowly target the area that formerly contained the tumor to destroy any remaining cancerous cells without damaging too many healthy cells. Widespread clinical uptake of DIR is however still lacking, mainly because existing approaches perform DIR by assigning weights to multiple important objectives and summing them into a single objective to be optimized. Tuning these weights for individual cases to obtain desired results is hard. With a multi-objective approach such weight-tweaking becomes obsolete, thereby removing a major hurdle to unleashing the important (clinical) potential of DIR. Researchers of the Medical Informatics group at CWI pioneered the use of multi-objective optimization for DIR with several pilot studies and publications and continuous to improve and extend research and development on this topic, including the design and use of (model-based) evolutionary algorithms to solve some of the hardest DIR problems (i.e., with large anatomical differences (e.g., due to surgery, see example above)). Existing methods have limited success here.
MSc Projects.
Talented students are always welcome to come and do their MSc-thesis work. Students can either propose a topic (will be reviewed before acceptance) or work on topics that contribute to the Medical Informatics subgroup (that are proposed by researchers in the group). Both fundamental algorithmic projects (on evolutionary computation) are possible as well as more application-oriented projects. Feel free to contact Peter A.N. Bosman for more information.
Researchers at CWI Life Sciences research group
Researchers by (project-based) association
  • Cees Witteveen (TU Delft, Algorithmics group, Full Professor)
  • Coen Rasch (AMC, Department of Radiation Oncology, Full Professor)
  • Mark Neerincx (TU Delft, Interactive Intelligence group, Full Professor)
  • Martha Grootenhuis (AMC, Department of Pediatric Psychosocial Care, Full Professor)
  • Arjan Bel (AMC, Department of Radiation Oncology, Head Clinical Physics)
  • Tanja Alderliesten (AMC, Department of Radiation Oncology, Senior Researcher)
  • Hans Merks (AMC, Department of Pediatric Oncology, Oncologist/Senior Researcher)
  • Koen Hindriks (TU Delft, Interactive Intelligence group, Associate Professor)
  • Dirk Thierens (Utrecht University, Department of Computer Science, Assistant Professor)
  • Cécile M. Ronckers (AMC, Emma Children's Hospital (DCOG-LATER), Researcher)
  • Irma van Dijk (AMC, Department of Radiation Oncology, Postdoc)
  • Krzysztof L. Sadowski (Utrecht University, Department of Computer Science, PhD Student)
  • Kleopatra Pirpinia (NKI, Department of Radiation Oncology, PhD Student)
  • Ziyuan Wang (AMC, Department of Radiation Oncology, PhD Student)
  • Stef Maree (AMC, Department of Radiation Oncology, PhD Student)
  • Marjolein van der Meer (AMC, Department of Radiation Oncology, PhD Student)
  • To be appointed on STW/KWF project (PhD Student at TU Delft)
  • To be appointed on STW/KWF project (PhD Student at AMC)

Companies by (project-based) association