Artificial Intelligence-based Innovations

The vision is to enrich and innovate health care in the digital domain by developing and tailoring novel artificial intelligence (AI) techniques to meet clinical challenges, while keeping a keen eye on what is ultimately needed to translate these developments into clinically readily usable tools. In other words, to bridge the gap between fundamental computer science and mathematics and clinical applications. A special focus is to strive for innovations in (radiation) oncology with AI.

Advancing cancer care

The department of Radiation Oncology of the LUMC is involved in multidisciplinary research aimed at improving the treatment and the quality of life of cancer patients, of which image-guided radiation treatment is an important part.

The AI-based Innovations research group aims to innovate the treatment of cancer by 1) developing solutions for tasks that are part of the image-guided radiation treatment workflow which are performed (mostly) manually and involve a time consuming, iterative, and non-insightful process (including the identification of the regions of interest in medical images and the optimization of radiation treatment plans), and 2) enabling better informed and improved clinical decision support to, for example, patient-specifically tailor the follow-up procedure and choice of preferred cancer treatment.

Advancing cancer care

The department of Radiation Oncology of the LUMC is involved in multidisciplinary research aimed at improving the treatment and the quality of life of cancer patients, of which image-guided radiation treatment is an important part.

The AI-based Innovations research group aims to innovate the treatment of cancer by 1) developing solutions for tasks that are part of the image-guided radiation treatment workflow which are performed (mostly) manually and involve a time consuming, iterative, and non-insightful process (including the identification of the regions of interest in medical images and the optimization of radiation treatment plans), and 2) enabling better informed and improved clinical decision support to, for example, patient-specifically tailor the follow-up procedure and choice of preferred cancer treatment.

 

Innovation via an ecosystem of close collaboration

We go beyond the application of AI by closely collaborating with both AI experts who develop new techniques and clinical experts. This results in truly new ways to solve demand-driven challenges coming from the clinic. It also opens the door to understanding what the possibilities and limitations are of the latest techniques and what data we may still need to acquire to realize clinical innovation. In turn, this may raise new questions about how AI techniques should be developed or adapted. In doing so, the cycle of interaction that is central to the ecosystem of innovation, is established.

Three focus areas

Our research can be roughly subdivided into three research focus areas: image processing, optimization of treatment, and explainable AI.

The focus area on image processing includes (semi-)automatic delineation of regions of interest in medical images, and medical image registration (i.e., aligning two medical images). Both are of importance in the process of radiation treatment planning.

The focus area on optimization of treatment is currently dedicated to innovating internal radiation treatment (i.e., brachytherapy).

Explainable AI can be used for tasks like predictive modelling based on multi-modal data (such as medical images, texts, categories). This way relationships of the underlying data may be revealed. Moreover, the explainability component of these techniques is expected to enlarge the trust in these techniques and thereby the acceptance and clinical uptake.

One of the ultimate goals of the development and application of the explainable AI research projects is to gain improved insights and new knowledge on the relation between cancer treatment as a whole and (late) adverse effects. Not only can image-guided radiation treatment be improved based on this knowledge, also better informed decision making would be enabled regarding the preferred treatment modality (e.g., photon-based radiation treatment versus proton-based radiation treatment) for a specific patient.

 

Themes for Innovation