AI model proves to be a heavyweight in tumor assessment

Physicians evaluate treatment response by measuring tumor growth. The current, diameter-based RECIST criteria (see box) are of limited use for pleural mesothelioma, since this cancer type grows in a thin, irregular layer along the lung wall. Where, then, do you measure the diameter to determine whether the therapy is working? This leads to uncertainty and frustration among patients and physicians.

AI experts, radiologists, and pulmonologists from the Netherlands Cancer Institute (NKI) have now solved this problem. Together, they developed the AI model ARTIMES, which can measure the entire volume of a tumor and compare it to previous scans. “This is unfeasible for humans,” says pulmonologist Sjaak Burgers. “It is difficult and time-consuming to determine for every pixel whether it is tumor tissue or not. Checking whether the AI model has done it correctly, however, is simple.”

Outperforming humans

“We are the first in the world to demonstrate that AI outperforms humans in this area, and that physicians can actually base their decisions on it,” says Kevin Groot Lipman, technical physician and lead author of the study published today in Lancet Oncology. The researchers built and tested the AI model with the help of radiologists and more than 11,000 CT scans from over 2,000 patients at 121 hospitals worldwide.

Pulmonologists at the NKI have embraced the model. The fact that they can already use it so soon after its development is no small feat. After all, an AI model that simply measures tumor volume alone does not tell them when to start or stop treatments. That is why the researchers took the crucial next step and developed guidelines to support physicians in their decision-making.

Reviewed by physician

“We can now assess tumor response to treatments much more accurately,” says Burgers. “We can discover the lack of response much sooner than before. This allows a patient to stop treatment earlier and, if possible, switch to a different treatment. This not only provides certainty but also spares our patients unnecessary side effects and reduces healthcare costs.” The AI model's measurements are always reviewed by a physician.

Via the in-house exemption of the present EU regulations, only physicians at the Netherlands Cancer Institute are permitted to use the model, as that is where it was developed. “We obviously want patients worldwide to benefit from this,” says Groot Lipman. “We are in the process of getting the model approved for use in other hospitals. We are also eagerly awaiting a proposal from the EU to simplify the approval process for this type of medical device.”

Shock

“I expect this model to come as a shock to physicians and researchers outside the mesothelioma field,” he predicts. Starting today, researchers worldwide can begin using the mesothelioma AI model themselves, which is publicly available online. The NKI is already testing AI models for lung cancer and brain metastases. “This is going to open up a whole new field of research. We expect that AI will also be able to help with many other types of tumors.”

The model is expected to improve the search for new treatments by making clinical trials more reliable and efficient. In a test using data from eight clinical trials, the research team demonstrated that the AI measurements are much more accurate than the criteria currently in use (see box). “This allows us to better assess the efficacy of new treatments in clinical trials.”

Research at the NKI is financially supported by KWF Dutch Cancer Society and the AVL Foundation.