Artificial Intelligence (AI)

Ritse Mann, radiologist, and Jonas Teuwen, AI for Oncology-pioneer, joined forces in order to develop a smart MRI to increase the speed of diagnostics in breast cancer care. Their research resulted in an algorithm that can 'see' breast abnormalities more quickly during an MRI scan. If no abnormalities are found, the MRI scan can be stopped early. This means that the examination will only take 3 minutes for most women. If abnormalities are detected, the scan will take as long as it does now: about 20 minutes. This method will save a lot of time, which could make MRI more affordable in breast cancer screening, the early detection of breast cancer.

Jan-Jacob Sonke's research group focuses on adaptive radiotherapy. Their goal is to be more aware of the changes that may occur in patients during their treatment. They want to be able to deliver the right radiation dose to the right location, with as little damage as possible to the patient's healthy tissue. This poses a challenge in every treatment session because both the patient and the tumor are always in motion and change shape over time. Central to this treatment is the combination of daily imaging, strategies to detect and adapt to changes to the patient's anatomy, and methods to personalize the treatment.

Our radiology department uses AI applications every day to improve diagnostics. AI software can support radiologists in the detection and observation of abnormalities found on imaging. This includes the detection of metastases on lung CT scans and prioritizing examinations that are showing unexpected acute abnormalities.

_{*The software assists in the detection and observation of possible metastases found on lung CT scans. The software detects if and where these lung tumors are located, generates a 3D volumetric analysis, and can evaluate changes through time.} 

Radiologist Laurens Topff is investigating the added value of AI applications in everyday clinical practice. Together with his team of technical experts, he translates innovations in AI research into the clinic, in order to improve accuracy and efficiency. Their projects focus on both the automated analysis of medical imaging, as well as operational processes like predicting MRI scan times in order to improve patient planning.

Unfortunately, not every patient with cancer responds to treatment in the same way. It is often hard to predict whether a therapy will be effective. The research teams led by Jonas Teuwen, Hugo Horlings, and Lodewyk Wessels are working on AI algorithms that can combine all available data belonging to a particular patient in order to predict whether a certain treatment type will prove effective, and how many side effects the patient will experience. Only a small number of patients respond to immunotherapy, for instance, but many do experience side effects when receiving the treatment. Better predictions that can show whether a patient might respond favorably can contribute to better and more effective treatment. Combination treatment involving multiple drugs is a common cancer treatment that tends to improve the effects of the individual drugs. Together with the University of Amsterdam, the AI for Oncology Lab is developing revolutionary algorithms.

The Netherlands Cancer Institute has joined forces with various other knowledge institutes in Amsterdam in order to invest in Artificial Intelligence over the coming 10 years. This collective is called AI Technology for People. The collective has invested in the development of responsible AI technologies, setting up research studies, attracting scientists, and educating students into AI. The Netherlands Cancer Institute collaborates with the following partners: Amsterdam Economic Board, Amsterdam UMC, Centrum Wiskunde & Informatica (the national research institute for mathematics and computer science), the municipality of Amsterdam, the Amsterdam University of Applied Sciences, Sanquin, the University of Amsterdam, and VU Amsterdam.
The Netherlands Cancer Institute is working on the AI for Oncology Lab together with the University of Amsterdam, and on the Partnership for Online Personalized AI-driven Adaptive RT (POP-AART) lab together with the University of Amsterdam and Elektra. Both labs are part of the Innovation Center for Artificial Intelligence, part of AI technology for people.

More information