Revealing hidden information
The potential of artificial intelligence (AI) to fuel progress is undeniable. AI’s ability to find hidden patterns and gather insights from vast volumes of data is accelerating progress in everything from drug discovery to surgical robots. We are currently deploying and developing AI for both research and clinical applications. However, we are still learning as progress through AI demands new knowledge and skills.
Prof. Ronald Boellaard, head of the Imaging Methodology Group, is employing a type of AI that is called ‘deep learning’ to accurately analyze images to better predict treatment response and prognosis for patients with diffuse large B-cell lymphoma.
“Research is essential to get a clear picture of both opportunities and risks,” says Boellaard.
A range of possible treatment plans
Machine learning uses self-learning algorithms that enable highly accurate prediction of outcomes. The department of Radiation Oncology has developed AI software that quickly proposes multiple radiation treatment plans for patients with prostate cancer.
- Arjan Bel, head of Radiotherapy Physics, department of Radiation Oncology, explains: "While Amsterdam UMC has performed research using computer support for radiation treatment plans for years, the challenge was to make quality plans quickly. Our AI software is based on self-learning evolutionary algorithms that display intelligent search behavior. It calculates a range of possible treatment plans while balancing the proper radiation dose and possible side effects”. In a blind test during the software’s testing stage, nearly all radiation oncologists preferred the new AI-based treatment plans over those formulated by current standard practices.
Adapting to the patient
To put those treatment prediction plans to use, Amsterdam UMC has deployed six Varian Ethos™ therapy solutions: machines for radiation therapy. Ethos therapy integrates AI-driven software to deliver highly personalized and precise radiation treatment, from patient setup through treatment delivery, in only 15 minutes. The Ethos therapy adapts the initial treatment plan in response to the variability of the tumor's shape and changes in nearby organs. This greatly focuses the radiotherapy to the tumor area, while sparing neighboring healthy tissue. "Adaptive therapy is one of the most important developments in the field of radiotherapy, and the new Ethos solution will greatly increase our ability to offer this advanced treatment to patients," says Prof. Ben Slotman, head of the department of Radiation Oncology.
Prediction
Advanced medical image analytics is increasingly used to predict clinical outcome in cancer patients. With the use of radiomics, hundreds of disease features can be extracted and analyzed from a single medical image.
- Pim de Graaf, together with colleagues of the Dutch Retinoblastoma Center, is assessing the value of AI-guided radiomics for the detection of subtypes of a rare form of pediatric eye cancer and to improve the detection of risk factors for developing distant metastasis.
- Martijn van Oijen, associate professor in Medical Oncology, is developing and evaluating practical clinical decision tools created by AI for patients with cancer of unknown primary origin, providing greatly needed support to improve treatment effectiveness.
- With the project ‘Deep learning for tumor response evaluation’, Nina Wesdorp is showing that AI can really have a direct impact on patient care. Her collaborative project with analytics software leader SAS is improving outcomes for patients with colorectal cancer – the third most common cancer worldwide – by predicting patients who will respond well to chemotherapy and are likely candidates for life-saving surgery.
Artificial intelligence-assisted research
Our cancer researchers are utilizing AI and machine learning to gain insights from large volumes of data, and to build models of the complex interactions within tumor cells and their environment. The resulting advancement in our understanding of cancer biology directly translates into the development of more precise and improved therapies.
Computational analyses and the ability to interpret high-throughput analysis-based results are quickly becoming essential skills for all researchers
according to Roel Verhaak, professor of Computational Biology of Brain Tumors.
Accelerating progress
The participation of Amsterdam UMC in the ambitious 1-billion-euro collaborative initiative ‘AI Technology for People’ underscores the commitment to harnessing the power of AI, and the recognition that we are on the edge of a revolution in which supercomputing will drive major progress in cancer research and care.
For more information about AI contact prof. Ronald Boellaard.