Developing new artificial intelligence (AI)-based strategies for improving precision radiotherapy. That is what the Netherlands Cancer Institute, the University of Amsterdam (UvA) and Elekta are focusing on within the Partnership for Online Personalised AI-driven Adaptive RT (POP-AART) lab. The lab receives funding from the Rijksdienst voor Ondernemend Nederland (RVO).
When irradiating tumours, giving the right dose of radiation at the right place with minimum damage to healthy tissue is the goal. In practice, this is difficult, partly because the patient and tumour are constantly moving and can change shape over time.
Optimising adaptive radiotherapy
The lab is part of the Innovation Centre for Artificial Intelligence. The parties involved aim to optimise adaptive radiotherapy using AI methods. This form of radiotherapy takes into account changes that may occur in patients during the course of treatment.
The parties aim to pioneer the use of new deep learning methods in the lab to address challenges in modern radiation oncology. These include personalisation, speed and precision. "Our ambition is not only to develop highly innovative AI algorithms, but also to make them applicable to clinical practice so that patients can be treated even better," says Jan-Jakob Sonke, group leader of adaptive radiotherapy and associate professor at the UvA. Sonke is one of the lab's two directors.
The other director is Efstratios Gavves, associate professor of computer vision and deep learning at the UvA. Gavves: "The time is ripe to combine the power of artificial intelligence and big data with the cutting-edge expertise of the UvA and the Antoni van Leeuwenhoek, for an application with enormous impact: the fight against cancer."
Different workflows
In the lab, several workflows run in parallel. Each work stream focuses on a different element of the radiotherapy process. These include the quality of daily treatment images. But also strategies for detecting changes in a patient's anatomy and adjusting treatment accordingly. Another example are methods for continuously personalising the treatment of patients.
Maurits Wolleswinkel, President Linac Solutions & Chief Product Officer at Elekta, said, "We believe AI has a crucial role to play in helping clinicians deliver increasingly personalised and complex cancer treatment to patients. Elekta's participation in POP-AART underscores our commitment to precision radiation medicine."
Customised radiation plan
AI has long been contributing to the treatment of cancer. For example, in March 2020, a patient at Amsterdam UMC was treated according to a radiotherapy plan developed by AI software from the Centre for Mathematics and Computer Science (CWI). This software helps to determine the correct radiotherapy plan for tumours. The solution strikes a balance between the radiation dose and minimising damage to surrounding organs.
The software suggests multiple radiation plans in minutes. This allows the doctor to create a treatment plan for the patient faster.
Calculating probability of metastasis
Gynaecological oncologist Hanny Pijnenborg of Radboudumc is leading a European research project, funded by the KWF, into the use of AI in the treatment of uterine cancer. One in ten women with this condition will experience metastasis. When uterine cancer metastasises, it initially spreads to the lymph nodes. In some countries, doctors therefore remove the lymph nodes. This enables them to determine whether these contain metastatic cancer cells.
This is not happening in the Netherlands in view of the lasting problems it may cause. Pathologists therefore examine a small piece of tumour tissue. AI helps determine the risk of metastasis to the lymph node in individual patients.
Author: Wouter Hoeffnagel
