The use of Artificial Intelligence (AI) is fueling a new revolution in biomedical research and healthcare. The enormous datasets generated by modern bio-medical techniques contain vast amounts of information – but extracting the knowledge locked within is beyond the processing ability of the human mind or normal computing systems.
Various forms of AI, however, are rapidly being developed that have an unprecedented ability to analyze these enormous datasets. By discovering hidden complex patterns and relationships, AI is primed to transform scientific research and usher in a new golden age of scientific discovery. In addition, the adoption of AI in healthcare is enabling dramatic improvements in everything from faster diagnoses to robot-assisted surgeries.
Unexplored territory
“We could see the groundbreaking potential of AI applications for health care and biomedical research, but how do you actually implement AI into the labs and clinic in a responsible and practical manner?” says Eefje van Kessel, care program manager at Cancer Center Amsterdam. “This was still pretty much unexplored territory.”
Enter Professor Geert Kazemier, currently the chair of the Cancer Center Amsterdam Executive Board. “We took the lead several years ago to initiate a working platform for Artificial Intelligence at Cancer Center Amsterdam. We launched a pilot program in collaboration with SAS Analytics Software & Solutions.”
SAS Analytics is a market leader in the field of data analytics with strong expertise in the development of novel algorithms for data analysis. Geert also chose SAS because the software provides a transparent summary of how each analysis is performed. This is a critical factor in the ongoing deployment of analytics, making it easier to audit the process and build more trust in AI.
In the initiative with Cancer Center Amsterdam, researchers brought their ideas for AI applications to SAS technicians. Together, they selected a number of pilot programs involving radiomics – the analysis of medical images – and proteomics – large-scale protein analysis.
“It was a win-win situation,” says Eefje. “Our scientists were able to interrogate datasets using SAS models and they were getting unprecedented results. For SAS, they could develop other AI applications using the knowledge gained through our collaboration and market these tools to other organizations.”
AI Pilot Projects at Cancer Center Amsterdam – one example
In the CAESAR project, Geert Kazemier leads a team of researchers including Nina Wesdorp and Michiel Zeeuw who are working to improve care for patients with liver-metastatic colon cancer. AI analyzes medical imaging of liver tumors pre- and post-systemic chemotherapy to see how the tumor is responding to treatment. The SAS AI models provide highly accurate total tumor volume and a 3D representation of each tumor, allowing doctors to make better decisions regarding which patients should have lifesaving surgery or a different treatment strategy.
“We also want to make personalized predictions. Together with SAS expertise and software, we train these predictive models with data from rich and large datasets,” says research associate Michiel Zeeuw. “The ultimate goal of the CAESAR research team and SAS is clinical implementation of the developed models to optimize the care of tomorrow for patients with cancer.”
The CEASAR collaboration has already resulted in 6 scientific manuscripts (see below).
Data-driven exploration
Applying AI models also opens a completely different way of performing research. Eefje: “Normally, scientists would analyze data with a specific question in mind. But with AI, the technology can analyze a dataset and find correlations that were previously not noticeable or anticipated. So, in a way, it is the data itself that drives the output of the AI models.”
Creating a better future for patients
Inspired by the research successes and improvements in patient care that were facilitated by the AI pilot projects at Cancer Center Amsterdam, the Amsterdam UMC board of directors decided to get onboard. They took over the contract with SAS earlier this year, and AI technology is now being deployed for the benefit of patients, care providers, and researchers throughout Amsterdam UMC.
“It is really rewarding to see the difference we are making with AI,” says Geert. “We’ve worked hard to get this ball rolling, through successful public-private partnerships and through funding from the Cancer Center Amsterdam Foundation and Health Holland. Now you can see the momentum building, with more projects that incorporate AI models and availability of more financial resources. Together we are truly shaping a future where improved treatments and care for patients are facilitated by Artificial Intelligence.”
For more information about AI-supported projects at Cancer Center Amsterdam contact Eefje van Kessel.
Do you have questions about the application of AI in patient care at Amsterdam UMC? Read this Q&A here (in Dutch).
People involved at Cancer Center Amsterdam
Eefje van Kessel
CAESAR project
Nina Wesdorp
Michiel Zeeuw
Geert Kazemier
Wesdorp, N.J., et al.(2020) Advanced analytics and artificial intelligence in gastrointestinal cancer: a systematic review of radiomics predicting response to treatment. European journal of nuclear medicine and molecular imaging. https://doi:10.1007/s00259-020-05142-w.
Wesdorp, N.J., et al.(2021) The Prognostic Value of Total Tumor Volume Response Compared With RECIST1.1 in Patients With Initially Unresectable Colorectal Liver Metastases Undergoing Systemic Treatment.Annals of Surgery Open; 2:e103. https://doi:10.1097/as9.0000000000000103.
Wesdorp, N.J., et al.(2021) Advanced image analytics predicting clinical outcomes in patients with colorectal liver metastases: A systematic review of the literature. Surgical Oncology; 38:101578. https://doi.org/10.1016/j.suronc.2021.101578.
Van 't Erve, I., et al. (2021) KRAS A146 Mutations Are Associated With Distinct Clinical Behavior in Patients With Colorectal Liver Metastases. JCO Precision Oncology: 1758-67. https://doi:10.1200/PO.21.00223.
Wesdorp NJ, et al. (2022) Interobserver Variability in CT-based Morphologic Tumor Response Assessment of Colorectal Liver Metastases.Radiology: Imaging Cancer; e210105. https://doi: 10.1148/rycan.210105.
Wesdorp, N.J., Zeeuw,J.M., et al.(2022) Deep learning models for automatic tumor segmentation and automatic total tumor volume assessment of patients with colorectal liver metastases. (Submitted)
Boris Janssen: AI in pathology
Bart Westerman: Toxicity atlas
An atlas for treating brain tumors: toxicity mapping by Amsterdam UMC and Medstone
Using artificial intelligence to improve brain tumor treatment
This article was created for Cancer Center Amsterdam.
© 2022 NHBC – All rights reserved.